Fuzzy Logic Applications in Computer Science and Mathematics

Fuzzy Logic Applications
in Computer Science
and Mathematics
Scrivener Publishing
100 Cummings Center, Suite 541J
Beverly, MA 01915-6106
Advances in Learning Analytics for Intelligent Cloud-IoT Systems
Series Editor: Dr. Souvik Pal and Dr. Dac-Nhuong Le
This book series involves different computational methods incorporated within the system
with the help of analytics reasoning and Sense-making in big data, which is centered in the
cloud and IoT-enabled environments. The series seeks volumes that are empirical studies,
theoretical and numerical analysis, and novel research findings. The series encourages
cross-fertilization of highlighting research and knowledge of data analytics, machine learning,
data science, and IoT sustainable developments. .
Publishers at Scrivener
Martin Scrivener ([email protected])
Phillip Carmical ([email protected])
Fuzzy Logic Applications
in Computer Science
and Mathematics
Edited by
Rahul Kar
Dac-Nhuong Le
Gunjan Mukherjee
Biswadip Basu Mallik
and
Ashok Kumar Shaw
This edition first published 2023 by John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA
and Scrivener Publishing LLC, 100 Cummings Center, Suite 541J, Beverly, MA 01915, USA
© 2023 Scrivener Publishing LLC
For more information about Scrivener publications please visit www.scrivenerpublishing.com.
All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or other-
wise, except as permitted by law. Advice on how to obtain permission to reuse material from this title
is available at http://www.wiley.com/go/permissions.
Wiley Global Headquarters

111 River Street, Hoboken, NJ 07030, USA
For details of our global editorial offices, customer services, and more information about Wiley prod-
ucts visit us at www.wiley.com.
Limit of Liability/Disclaimer of Warranty

While the publisher and authors have used their best efforts in preparing this work, they make no rep
resentations or warranties with respect to the accuracy or completeness of the contents of this work and
specifically disclaim all warranties, including without limitation any implied warranties of merchant-
ability or fitness for a particular purpose. No warranty may be created or extended by sales representa
tives, written sales materials, or promotional statements for this work. The fact that an organization,
website, or product is referred to in this work as a citation and/or potential source of further informa
tion does not mean that the publisher and authors endorse the information or services the organiza
tion, website, or product may provide or recommendations it may make. This work is sold with the
understanding that the publisher is not engaged in rendering professional services. The advice and
strategies contained herein may not be suitable for your situation. You should consult with a specialist
where appropriate. Neither the publisher nor authors shall be liable for any loss of profit or any other
commercial damages, including but not limited to special, incidental, consequential, or other damages.
Further, readers should be aware that websites listed in this work may have changed or disappeared
between when this work was written and when it is read.
Library of Congress Cataloging-in-Publication Data
ISBN 978-1-394-17453-9
Cover image: Pixabay.Com

Cover design by Russell Richardson
Set in size of 11pt and Minion Pro by Manila Typesetting Company, Makati, Philippines
Printed in the USA
10 9 8 7 6 5 4 3 2 1
Contents
Preface xiii
1 Decision Making Using Fuzzy Logic Using Multicriteria 1
Panem Charanarur, Srinivasa Rao Gundu and J.Vijaylaxmi
1.1 Introduction 2
1.2 Fuzzy Logic 5
1.3 Decision Making 6
1.4 Literature Review 7
1.5 Conclusion 10
Acknowledgment 11
References 11
2 Application of Fuzzy Logic in the Context of Risk
Management 13
Sudipta Adhikary and Kaushik Banerjee
2.1 Introduction 13
2.2 Objectives of Risk Management 14
2.3 Improved Risk Estimation 15
2.3.1 Point-Wise Calculations on a Curve 15
2.3.2 Estimation of a Curve 15
2.3.3 Accuracy in Quantification is Raised 16
2.3.4 The Problems with the Basic Quantification
Approach 16
2.4 Threat at Quantification Matrix 17
2.4.1 Qualitative Matrix 17
2.4.2 Errors in Scaling 17
2.4.3 Band Width at Various Scales 17
2.5 Fundamental Definitions 18
2.5.1 Positioning Statement 18
2.5.2 Risk Under the Level of Tolerance 19
v
vi Contents
2.5.3 Risk Elimination 19

2.6 Fuzzy Logic 19
2.7 Risk Related to Fuzzy Matrix 20
2.8 Conclusion 26
Bibliography 26
3 Use of Fuzzy Logic for Controlling Greenhouse Environment:
A Study Through the Lens of Web Monitoring 29
Kaushik Banerjee and Sudipta Adhikary
3.1 Introduction 29
3.2 Design (Hardware) 30
3.2.1 Sensor for Measuring Soil Moisture 30
3.2.2 Sensor for Measuring Humidity and Temperature 31
3.3 Programming Arduino Mega Board 31
3.3.1 Fuzzification 32
3.3.2 Fuzzy Inference 32
3.3.3 Communication via Remote Connections
and a Web Server 33
3.4 Implementation of a Prototype 34
3.5 Results 35
3.6 Conclusion 37
Bibliography 37
4 Fuzzy Logics and Marketing Decisions 41
Mohammed Majeed
4.1 Introduction 41
4.2 Literature 42
4.2.1 Fuzzy Logic (FL) 42
4.2.2 FL Application in Marketing 43
4.2.2.1 Communication and Advertising 43
4.2.2.2 Customer Service and Satisfaction 43
4.2.2.3 Customer Segmentation 43
4.2.2.4 CRM 44
4.2.2.5 Pricing 44
4.2.2.6 Evaluation of a Product 44
4.2.2.7 Uncertainty in the Development of New
Products 45
4.2.2.8 Decision Making 45
4.2.2.9 Consumer Nation Identity (CNI) 46
4.2.2.10 Quality of Service 46
4.3 Conclusion 46
Contents vii
4.4 Further Studies 47

References 47
5 A Method for Ranking Fuzzy Numbers Based on Their Value,
Ambiguity, Fuzziness, and Vagueness 51
Sunayana Saikia and Rituparna Chutia
5.1 Introduction 51
5.2 Preliminaries 54
5.2.1 Definitions and Concepts 54
5.3 The Designed Method 56
5.4 Validate the Reasonableness of the Suggested Ranking
Algorithm 68
5.5 Comparative Analysis and Numerical Examples 75
5.6 Application 87
5.7 Conclusions 94
References 94
6 Evacuation of Attributes to Translucent TNSET
in Mathematics Using Rough Topology 99
Kala Raja Mohan, R. Narmada Devi, Nagadevi Bala Nagaram,
Sathish Kumar Kumaravel and Regan Murugesan
6.1 Introduction 99
6.2 Basic Concepts of Rough Topology 100
6.2.1 Conditional Attribute 100
6.2.2 Decision Attribute 101
6.2.3 Rough Topology 101
6.2.4 Lower Approximation 101
6.2.5 Upper Approximation 101
6.2.6 Boundary Region 101
6.2.7 Basis 101
6.2.8 Information System 101
6.2.9 Core 101
6.3 Algorithm 102
6.4 Information System 102
6.5 Working Procedure 104
6.6 Conclusion 104
References 104
7 Design of Type-2 Fuzzy Controller for Hybrid
Multi-Area Power System 107
Susmit Chakraborty, Arindam Mondal and Soumen Biswas
7.1 Introduction 108
viii Contents
7.2 Plant Model 108

7.3 Controller Design 109
7.3.1 Proportional Integral Derivative (PID) Controller 110
7.3.2 Fractional Order Proportional Integral Derivative
(FOPID) Controller 111
7.3.3 Type-2-Fuzzy Logic 111
7.4 Levenberg–Marquardt Algorithm 115
7.5 Optimization of Controller Parameters Using CASO
Algorithm 116
7.6 Result and Analysis 116
7.6.1 Without Disturbances 116
7.6.2 With Disturbances 119
7.7 Conclusion 121
Appendix 121
References 122
8 Alzheimer’s Detection and Classification Using Fine-Tuned
Convolutional Neural Network 125
Anooja Ali, Sarvamangala D. R., Meenakshi Sundaram A.
and Rashmi C.
8.3 Methodology 133
8.3.1 Dataset 134
8.3.2 Pre-Processing 134
8.4 Implementation and Results 134
8.5 Conclusion 138
References 138
9 Design of Fuzzy Logic-Based Smart Cars Using Scilab 143
Josiga S., Maheswari R. and Subbulakshmi T.
9.2 Literature Survey 145
9.2.1 Fuzzy Logic for Automobile Industry 146
9.2.2 Fuzzy Logic for Smart Cars 146
9.2.3 Fuzzy Logic for Driver Behavior Detection 147
9.2.4 Fuzzy Logic Applications for Common Industry 148
9.3 Proposed Fuzzy Inference System for Smart Cars 149
Contents ix
9.3.2 Membership Functions 150

9.3.3 Rule Base 153
9.3.4 Example Rules 153
9.3.5 Defuzzification 153
9.4 Implementation Details and Results 155
9.5 Conclusion and Future Work 156
References 157
10 Financial Planning and Decision Making for Students
Using Fuzzy Logic 159
G. Surya Deepan and T. Subbulakshmi
10.3 System Architecture 163
10.3.1 Input 164
10.3.3 Membership Function 165
10.3.3.1 Necessity 166
10.3.3.2 Cost Percentage 166
10.3.3.3 Quality 167
10.3.4 Fuzzy Rule Base 167
10.3.5 Fuzzy Output 168
10.3.6 Defuzzification 169
10.4 Conclusion and Future Scope 169
References 170
11 A Novel Fuzzy Logic (FL) Algorithm for the Automatic
Detection of Oral Cancer 173
M. Praveena Kiruba bai and G. Arumugam
11.1.1 Significance of Pre-Processing 174
11.2 Image Enhancement 174
11.3 Gabor Transform 175
11.4 Image Transformation 175
11.5 Adaptive Networks: Architecture 176
11.5.1 Classification of Images 176
11.6 Results and Discussions 177
11.7 Conclusion 177
Bibliography 177
x Contents
12 A Study on Decision Making of Difficulties Faced by Indian

Workers Abroad by Using Rough Topology 179
Nagadevi Bala Nagaram, R. Narmada Devi , Kala Raja Mohan,
Regan Murugesan and Sathish Kumar Kumaravel
12.1.1 Problems Faced by the Indian Workers 180
12.2 Fundamental Idea of Rough Topology 182
12.2.1 Conditional Attribute 182
12.2.2 Decision Attribute 182
12.2.3 Rough Topology 182
12.2.4 Lower Approximation 182
12.2.5 Upper Approximation 182
12.2.6 Boundary Region 182
12.2.7 Basis 183
12.2.8 Information System 183
12.2.9 Core 183
12.3 Algorithm 183
12.4 Information System 183
12.5 Working Procedure 185
12.6 Conclusion 185
References 186
13 Case Study on Fuzzy Logic: Fuzzy Logic-Based PID
Controller to Tune the DC Motor Speed 187
Devendra Kumar Somwanshi
13.1.1 DC Motor 189
13.1.2 DC Motor Speed Control Methods 189
13.1.2.1 PID Controller 189
13.1.2.2 Fuzzy-Based PID Controller 189
13.1.2.3 Micro Controller-Based PID
Controller 190
13.1.2.4 Genetic Algorithm-Based PID
Controller 190
13.2.1 Common Findings 190
13.2.2 Comparative Analysis of Research Works
Reviewed 191
13.2.3 Strengths in the Literature Reviewed 191
13.2.4 Weaknesses in the Literature Reviewed 195
13.2.5 Findings in the Literature Reviewed 195
Contents xi
13.3 Design of Fuzzy-Based PID Controller 196

13.3.1 Fuzzy Controller 196
13.3.2 Flowchart for Fuzzy Controller 196
13.3.3 Fuzzy Logic Controller Membership Function
and FAM Table 197
13.3.4 Rules for the Fuzzy Controller 200
13.3.5 Simulation Diagram of FLC 202
13.3.6 Fuzzy-Based PID Controller 203
13.3.6.1 Fuzzy Block Design 203
13.3.6.2 Flowchart for Fuzzy-PID Controller 204
13.3.6.3 Simulation Diagram of Fuzzy-PID
Controller 204
13.4 Experimental Work and Results Analysis 205
13.5 Conclusion and Future Scope 207
References 209
14 Application of Intuitionistic Fuzzy Network Using Efficient
Domination 213
A. Meenakshi, J. Senbagamalar and A. Kannan
14.2 Efficient Domination in Intuitionistic Fuzzy Graph (IFG) 215
14.3 Main Frame Work 217
14.3.1 Construction of IFN from Sub IFN 217
14.4 Secret Key 219
14.4.1 Encryption Algorithm 219
14.4.2 Decryption Algorithm 223
14.5 Illustration 224
14.5.1 Construction of IFN from Sub IFN 224
14.5.2 Secret Key 228
14.5.3 Encryption Algorithm 228
14.5.4 Decryption Algorithm 230
14.6 Conclusion 231
References 231
15 Analysis of Parameters Related to Malaria
with Comparative Study on Fuzzy Cognitive Maps
and Neutrosophic Cognitive Maps 233
Regan Murugesan, Sathish Kumar Kumaravel, Kala Raja Mohan,
Narmada Devi Rathinam and Suresh Rasappan
15.2 Parameters of Malaria 235
xii Contents
15.3 Fuzzy Cognitive Map 235

15.3.1 Matrix Representation of FCM 235
15.4 Neutrosophic Cognitive Map 240
15.4.1 Matrix Representation of NCM 241
15.5 Comparison and Discussion 246
15.6 Conclusion 247
References 247
16 Applications of Fuzzy Cognitive Maps and Neutrosophic
Cognitive Maps on Analysis of Dengue Fever 249
Sathish Kumar Kumaravel, Regan Murugesan,
Nagadevi Bala Nagaram, Suresh Rasappan and G. Yamini
16.2 Parameters of Dengue 251
16.3 Fuzzy Cognitive Maps 251
16.3.1 Matrix Representation of FCM 252
16.4 Neutrosophic Cognitive Map 257
16.4.1 Matrix Representation of NCM 258
16.5 Comparison and Discussion 263
16.6 Conclusion 264
References 265
17 A Comprehensive Review and Analysis of the Plethora
of Branches of Medical Science and Bioinformatics
Based on Fuzzy Logic 267
Partha Sarker and Siddhartha Roy
17.2 Previous Work 271
17.3 Fuzzy Logic in Medical Fields and Bioinformatics 271
17.3.1 Applied Fuzzy Logic in Medical Areas 271
17.3.2 Applied Fuzzy Logic in Bioinformatics 272
17.4 Review of Published Work and In-Depth Analysis 273
17.5 Conclusion 273
References 277
Index 279
Preface
The prime objective of developing this book was to provide meticulous

details about the basic and advanced concepts of fuzzy logic and its all-
around applications to different fields of mathematics and engineering.
The book caters to a certain level of professional knowledge, academicians,
students, and researchers. The basic steps of fuzzy inference systems start-
ing from the core foundation of the fuzzy concepts are presented in this
book. The fuzzy theory is a mathematical concept and, at the same time,
it is applied to many versatile engineering fields and research domains
related to computer science. The fuzzy system offers some knowledge
about uncertainty and also is related to the theory of probability. A fuzzy
logic-based model acts as the classifier for many different types of data
belonging to several classes. Covered in this book are topics such as the
fundamental concepts of mathematics, fuzzy logic concepts, probabil-
ity and possibility theories, and evolutionary computing to some extent.
The combined fields of neural network and fuzzy domain (known as the
neuro-fuzzy system) are explained and elaborated through many highly
regarded research papers. Each chapter has been produced in a very lucid
manner, with grading from simple to complex in an effort to accommodate
different audiences.
The application-oriented approach is the unique feature of this book.
Apart from the theoretical discussion, the problems and the allied case
studies concerned with the topics discussed in this book will be of great
interest to a broad audience. The problems and the case studies furnished
in this book are worthwhile to researchers and academicians, as well. This
book comprises state-of-the-art information on a wide range of various
subjects, all directly or indirectly connected to the overarching topic.
Fuzzy logic and its application have evolved significantly and, through
many research paths, have arrived at the current stage. With concern paid
to the students of different types of engineering, this book also addresses
some additional aspects. Primarily the book focuses on:
xiii
xiv Preface
a) The myriad modern research information in the field of

computational intelligence, presented with references to
many published papers
b) The pertinent information and research in the field of fuzzy
systems, its different variants, and evolutionary computing
c) The future research directions in the field of fuzzy logic-
based computational intelligence, which provides an effec-
tive means of research in the field of classification of items,
from different species and so forth
d) Providing a compact treatise on the fuzzy-based com-
putational intelligence and how it applies to evolutionary
computing
The material of this book was developed and arranged so that read-
ers can easily grasp the fundamental concepts of the subject and gradu-
ally move to more advanced levels through functional assessments of the
matter in both broad and analytical ways. The target readership includes
researchers, professionals, and students willing to pursue their career fur-
ther in the field of computation in the fuzzy domain.
We express our sincere thanks with ample acknowledgment to all our
colleagues, friends, and students for their invaluable suggestions and feed-
back in the development of this book, including the provision of more
important and relevant information. We must offer our heartfelt gratitude
to our family members, for without their support and endurance, this book
would have been an impossible task. Lastly, we are very much grateful to
the editors at Scrivener and Wiley.
We wish every reader an insightful, perceptional, and informative jour-
ney into this book, the world of fuzzy logic systems, and its application
paradigm.
The Editors
Rahul Kar, Dac-Nhuong Le, Gunjan Mukherjee,
Biswadip Basu Mallik and Ashok Kumar Shaw
July 2023
1
Decision Making Using Fuzzy
Logic Using Multicriteria
Panem Charanarur1, Srinivasa Rao Gundu2* and J.Vijaylaxmi3
Department of Cyber Security and Digital Forensics, National Forensic

1
Sciences University, Tripura Campus,Tripura, India

2
Department of Computer Science, Government Degree College-Sitaphalmandi,
Hyderabad, Telangana, India
3
PVKK Degree & PG College, Anantapur, Andhra Pradesh, India
Abstract
Fuzzy set theory and multicriteria decision making were initially introduced
in the early 1970s. As a consequence, a wide variety of innovative solutions
have been tried and confirmed through the use of fuzzy multicriteria decision
making. The following section provides a quick overview of fuzzy multicriteria
decision-making categories, as well as some of their earliest and most recent uses.
Uncertainty and ambiguity are synonyms for the adjective “fuzzy.” Since there are
numerous instances in real life when we are not able to discern whether a given
condition is true or untrue, the flexibility that fuzzy logic gives is quite helpful
when making decisions. There will always be some degree of inaccuracy and
unpredictability to every circumstance. The development of fuzzy set theory and
the rise of decision making are closely linked. It was necessary to highlight the
groundbreaking applications of fuzzy multicriteria decision making (MCDM) in
order to encourage more research in this field. Many real-world Malaysian cases
are used to demonstrate the broad range of applications that may be pulled from
the various approaches described in this study. Some cutting-edge intelligent
methods to MCDM are intended to help spread the news about fuzzy MCDM.
Keywords: Fuzzy, fuzzy set, multicriteria decision making (MCDM), decision
theory, logical thinking
*Corresponding author: [email protected]
Rahul Kar, Dac-Nhuong Le, Gunjan Mukherjee and Biswadip Basu Mallik and Ashok Kumar Shaw (eds.)
Fuzzy Logic Applications in Computer Science and Mathematics, (1–12) © 2023 Scrivener Publishing LLC
1
2 Fuzzy Logic Applications in Computer Science and Mathematics
1.1 Introduction
Determining what to do is an NP-complete issue, and it has applications in
a wide range of domains. There were a lot of researchers that looked into
the subject in the mid-20th century. Through his work on the lens model,
Brunswick pioneered the use of mathematical or statistical models to cap-
ture individual variances in decision-making procedures in 1947, paving
the way for future research.
It is possible to analyze many different decision-making circumstances
using policy capture, according to the researchers. Sequences of judgment
stimuli that have been produced using controlled signals are shown to par-
ticipants in order to capture and model the participants’ subsequent judg-
ments, according to this approach. Changing environmental signals in a
systematic way addresses interactions with the environment, whereas the
deployment of expert decision-making systems addresses interactions with
extraneous validity. Brehmer and Brehmer followed up on their research
into this method by conducting an analysis into the amount to which indi-
viduals utilize various decision policies and the extent to which they are
aware of the processes they apply when making judgments.
In today’s fast-paced climate, enterprises, industries, and the govern-
ment all require competent, rational decision-making, and they all need it
urgently. It is needed that it must be chosen between one or more options
while making a decision. In the framework of cognitive brain processes, it
is possible to come up with a wide range of options for making a choice.
A final choice is reached at the end of every decision-making process.
Decisions may be made in terms of action or opinion depending on the
outcome. Long-term or short-term planning, working at the highest or
lowest levels of management, having the ability to make sound decisions
is a need. Using the advanced decision-making tools offered by Decision
Theory [1], the Decision Maker can make better choices when confronted
with tough choices.
The following is an example of a wide definition of Decision Theory:
One must pick the action that is most likely to achieve one’s objectives, as
established by the decision maker, from among the numerous possibilities
when provided with a list of options. Before making a decision, a person
who wants to choose the best possible course of action must evaluate all of
the potential consequences and outcomes associated with each option. The
selection problem is subjected to logical and quantitative investigation.
To better comprehend or prescribe actions to strengthen the coherence
between various alternatives offered by the scenario and the aims and value
Decision Making Using Fuzzy Logic Using Multicriteria 3
systems of the agents participating in the decision process, there are a vari-
ety of ways for modeling decision scenarios. Building a relational or func-
tional model is the basis for mathematical decision analysis [2].
Many different methodologies have been used to study the decision-
making process of humans. Experts recommend that while evaluating a
person’s options, they take into account their psychological needs, pref-
erences, and desired values. From a cognitive standpoint, it is essential to
perceive decision-making as a continuous process that is linked to one’s
immediate environment. It is necessary to examine the logic and rational-
ity of decision-making and the consistency of the choice that results from
it from a normative perspective to understand the process. Another way
to look about it is as a process that leads to the discovery of an appropri-
ate solution to a problem that has been addressed. There are two kinds of
assumptions that may influence a decision: explicit and implicit [3]. An
emotional or intellectual choice may also lead to a decision.
Logical thinking is required for technical decision-making in which
experts draw on their expertise to make well-informed judgments about
the present situation. A study employing natural techniques found that
professionals rely more on intuitive Decision Making than organized
approaches when faced with tighter deadlines, higher stakes, or more
uncertainty. A recognition-primed decision approach is used in order to
fit a collection of indicators into the expert’s competence and swiftly arrive
at an acceptable plan of action [4]. There have been several recent efforts
that have formalized the incorporation of uncertainty into decision mak-
ing. Several studies have underlined the importance of human judgment
and the inherent flaws in decision making as essential components in
the assessment of human performance. Psychologists have been working
on mathematical and computer models for many years to better under-
stand how humans make choices. They may be used to a broad variety
of tasks and circumstances in a wide range of organizations and settings.
AI research has concentrated on how intelligent beings interact with their
environment and make decisions, rather than how they think.
Scientists study how individuals interact with their environment and
try to mimic their decision-making processes in order to learn more about
these people. Most of the time, public policy is at issue in these kinds of
operations. All of the decision making problems often include the concept
of ambiguity or vagueness. Many psychologists turn to probability theory
for help when presented with this problem. When using probabilistic mod-
els instead of statistical models, there are two key limitations: It should be
highlighted that certain natural sources of uncertainty may not exist in a
form that can be explained by a universally agreed probability model [5].
Many cognitive processes exhibit an unpredictability that defies the predic-

tions of probability theory and randomness, and this uncertainty may be
difficult to explain. Since Lotfi Zadeh got his start in fuzzy sets theory, the
latter field has grown to include soft computing, which incorporates tech-
niques, such as fuzzy systems with neural networks and genetic algorithms.
As per Lotfi Zadeh the fuzzy sets and fuzzy logic, which are mathemat-
ical systems that directly translate into natural language, may be used in
combination with other mathematical systems to represent the intricate
interactions between variables that occur in everyday language. Model-
free estimators or universal approximations, reasoning imprecision, and
fuzzy rule representations are all part of the fuzzy system method to simu-
lating human judgment and decision making [6].
Fuzzy logic is becoming an increasingly vital instrument to have on hand
whether we work in construction engineering or management research. As
a consequence of the lack of comprehensive data sets for modeling, fuzzy
logic may be used to reflect the subjective uncertainty in the construction
sector. There are several ways to develop a hybrid system, such as using
fuzzy logic, evolutionary algorithms, and artificial neural networks, but
the most important is to mix these methods with each other. Additionally,
unique applications in planning and scheduling, estimation, bidding, pro-
ductivity; project control; structuring projects; process improvement; and
risk analysis will be discussed in this session. With a focus on fuzzy logic
and fuzzy hybrid techniques, we could conduct a detailed examination and
provide recommendations on how to adapt them for construction applica-
tions in particular [7].
Image processing, analysis, indexing, and retrieval are becoming more
important due to the increasing availability on the Internet of enormous
picture datasets. Using low-level content-based qualities, content-based
image retrieval (CBIR) is able to give results that are quite consistent.
“Semantic gap” has long been a problem in CBIR due to the challenges
in precisely describing images at the lowest feasible level and having that
description be understood by people. Scholars are working hard to fill up
the semantic gaps that have emerged in image processing and retrieval
studies. It is essential to create fast techniques for extracting useful and
succinct characteristics from photos and the creation of flexible similarity
measures for object matching, as well as the automated tagging of visual
information with semantic ideas in order to meet the needs of users. There
are ways of dealing with all of these issues via the use of fuzzy techniques
to image processing, which may provide useful tools and simple processes
for extracting pictures based on content and ideas [8].
1.2 Fuzzy Logic

People use the word “fuzzy” when something is confusing or ambiguous.
In the real world, we often encounter situations in which we are unable to
determine whether a given condition is true or false. Fuzzy logic allows us
to think more freely in these types of circumstances. In this way, the flaws
and uncertainties of any given circumstance may be taken into account
[9]. There are two possible truth values in the Boolean system: 1.0 and 0.0.
1.0 represents the absolute truth, while 0.0 represents the ultimate false-
hood. Floppy systems lack the logic of an absolute true value and an equal
but opposite value, as in a binary system. Fuzzy logic, on the other hand,
includes an intermediate value that might be either half true or completely
untrue.
Architecture of a fuzzy system is divided into four main components;
It is the initial phase in the decision-making process, and it is made up
of rules and IF-THEN conditions proposed by specialists to manage the
decision-making system using linguistic information. Numerous recent
breakthroughs in fuzzy theory have led to practical methods for design-
ing and adjusting fuzzy control systems. There are less ambiguous rules
as the consequence of most of these developments. A fuzzy set algorithm
is used to transform inputs like integers into fuzzy sets. Temperature and
pressure data, RPM readings, and so on are examples of crisp inputs, which
are the exact sensor values that are sent into the control system for pro-
cessing. When a fuzzy input matches one of a set of rules, this component
determines which rules should be activated depending on the input field.
Control actions are then generated by combining the rules that have been
triggered. There are four steps involved in defuzzifying the fuzzy sets that
are generated by the inference engine and may be utilized in other applica-
tions. With the help of an expert system and various defuzzification tech-
niques, it is possible to reduce the quantity of error that is generated. There
are several advantages to using fuzzy logic systems, including the fact that
they are able to work with input data that have been corrupted in any way.
A fuzzy logic system’s construction is clear and easy to understand. There
is nothing complicated about fuzzy logic, which is founded on the mathe-
matical concepts of set theory [10].
In many cases, it may provide a quick and effective solution to complex
problems since it mirrors human thinking and decision-making. Only a lit-
tle quantity of memory is required since the approaches may be conveyed
with a short bit of data. As a consequence of the use of fuzzy logic by many
experts, there was an uncertain situation. Fuzzy logic does not lend itself to
a well-defined method when it comes to solving problems. As a result, it is

difficult or impossible to establish the qualities of our technique in the vast
majority of cases. Accuracy is sometimes sacrificed while employing fuzzy
logic since it may work with both exact and imprecise data. Spacecraft and
satellite altitudes are controlled using fuzzy logic in a number of indus-
tries, including the aerospace sector. Motorists use it for speed control and
traffic management in the car system. Corporate decision-support systems
and employee evaluations make heavy use of it, especially in large orga-
nizations. Uses include adjusting solution and dry material pH, as well as
chemical distillation operations. A number of high-volume applications of
artificial intelligence depend on fuzzy logic. Natural language processing
is one example.
It is commonly used in modern control systems, such as expert systems,
and it works well. As a technique to speed up human decision making,
fuzzy logic is often used in combination with Neural Networks. Data are
gathered and transformed into more relevant data via the creation of fuzzy
sets, which are made up of incomplete facts.
1.3 Decision Making

A decision maker may utilize decision theory or decision analysis to deter-
mine the optimal approach for a given circumstance when faced with a
large number of alternatives and an uncertain future. Consider the follow-
ing scenario: a clothing company wants to make large quantities of a new
style because they feel it would be well received by the public and, hence,
popular. The manufacturer, on the other hand, would prefer to create
fewer units if it anticipates that the product’s appeal and demand would be
insufficient. Manufacturers of seasonal garment items must make a man-
ufacturing quantity decision before the demand for their products can be
assessed. This is a significant issue. Once the items are on the shelves and
available for purchase by consumers, it will be hard to determine whether
or not they are well liked by those who purchase them. The choice of the
optimal production volume decision from a collection of options when
there is uncertainty about future demand is a challenge for Option Theory
Analysis. The theory of decision theory starts with the following assump-
tions, which are based on the notion that all decision-making situations
share a set of identical properties.
Decision makers are individuals or groups of individuals who are in
charge of determining the best course of action to pursue from among the
numerous alternatives available. The decision maker is presented with a
variety of alternative courses of action, or “courses of action.” It is the pur-

pose of decision analysis to choose the most advantageous choice from a
group of alternatives in order to achieve a certain objective. In many cases,
circumstances beyond of the decision maker’s control, such as the status of
the environment, impact whether a choice will be successful or unsuccess-
ful. These outcomes are referred to as “states of nature” in certain circles.
There is a payout associated with each conceivable course of action and
natural situation, and this payout quantifies the total benefit to the decision
maker resulting from a particular combination of these factors in each case.
When dealing with a specific problem, a payout table lists all of the states
of nature that are mutually exclusive and collectively exhaustive, as well as
a set of recommended actions or strategies for dealing with that issue. The
payoff is calculated for each unique combination of natural circumstance
and course of action that occurs. Consider the following scenario: there are
m conceivable occurrences or natural states and n alternative courses of
action in the issue under consideration. A reward compatible with strategy
Aj will be given to the decision-maker as a consequence of this outcome.
In the future, it will be able to represent Si using the notation pij (where
I = 1,..., m and j = 1, ..., n).
A term used to describe the gap between what may have occurred in
a natural condition and what actually transpired is “opportunity loss.”
During the computation of opportunity losses, each and every probable
natural occurrence is taken into consideration separately. Consider the
existence of a natural phenomenon known as Si. “The payments for each
of the n strategies are represented by the numbers pi1, …, pin.” Allowing
for the possibility that Mi is the largest of these values. Using A1 as a
decision-making tool will have negative consequences for M1–p11.
1.4 Literature Review

Use of one’s mental abilities is required for some kinds of passwords. In
1965, Zadeh created the concept of fuzzy logic, which has since become a
universally accepted method of reasoning in computer science. It has taken
some time for different schools of thought to adopt this new paradigm since
it was first provided in a preset form for technological reasons. Many west-
ern scientists have voiced skepticism about the use of fuzzy logic from its
birth, concerned that it may compromise the integrity of previously estab-
lished scientific principles. When it was all happening on stage, the show
was fantastic. There were several examples of the advantages of employing
a variety of systems, spanning from mathematics to engineering. After all,
the community voted on fuzzy logic as the best option. During the devel-
opment of the country’s first subway system, fuzzy logic controllers were
employed, which opened in Tokyo in 1987. Most intelligent products on
the market today use fuzzy logic-based technology. Fuzzy sets and fuzzy
logic have been proven to be useful tools in the fields of management and
decision sciences, as well as engineering [11].
Various classification strategies have been described by, differ-
ent authors, and others. An example of a categorizing approach is the
Analytical Hierarchy Process (AHP), which may be used to find the
optimum solution for a given issue. Using AHP reduces the time needed
to identify the best alternative and speeds up the decision-making pro-
cess, according to study results. Every feature has the potential to either
improve or reduce utility monotonically overtime, according to TOPSIS
methodology. Consequently, it is clear that there is an optimal solution
for both the good and bad aspects of this circumstance. Final ranking
is established by the proximity of these options to one another, which
is arranged alphabetically by their closeness to one another. And also
authors employed an aggregation technique to combine preferences in
one of the most innovative models yet produced for merging decision
makers’ preferences [12]. Fuzzy goal programming approaches were uti-
lized to overcome this problem. They discovered that they had a posi-
tive impact. Methods like described by may also utilize the phrase “grey
interval numbers,” in which the input variables are referred to as “grey
numbers.” Many researchers have studied MADM. MADM has also been
the subject of many studies.
Multi-attribute decision-making process

First and first when it comes to making a real-world choice, a collec-
tion of possibilities must be evaluated for their quality. It is impossi-
ble for a single decision maker to handle such circumstances because
of their complexity. It is common practice in the real world to include
experts who explain their judgments on the performance of alternatives
when insufficient information is available. To what extent a member of
a decision-making group’s viewpoint should be taken into account is an
important issue. It is important to keep in mind that different people
have different specializations, and hence their perspectives are of differ-
ing importance in different areas because of this. When it comes to calcu-
lating the relative importance of various decision-makers, this is a topic
on which scholars are working.
In both management and decision science, multiattribute decision mak-

ing is a key issue that is only growing in importance. A noncomplex situation
is one in which the opinions of decision makers (DM) are completely known
before any choices have been taken in this area. In this essay, we will not be
addressing this kind of issue. A scenario in which the DMs’ various points
of view are represented in verbal terms rather than numerically intrigues us
more than any other. In a situation like this, the value of the DMs cannot be
emphasized. A group of DMs is usually given the task of choosing from a
restricted set of options when it comes to making a decision.
The DMs are responsible for the following:
i) To get their comments on the weights of the qualities, and ii) offer
their thoughts on the options for specific attributes, I asked them to do
so. There are several uses for MADM, and this problem is critical to the
future of the technology. There are many factors to consider while making
a choice, including both qualitative and quantitative features, which are
frequently based on incorrect data and human judgment. Because of this,
fuzzy MADM seems to be an effective tool for coping with these decision-
making issues. Judgments made by leaders in most circumstances are
vague and consequently cannot serve as exact numerical values in most
cases. The usage of triangular fuzzy numbers has been used to quantify
uncertainty in supplier selection situations. Step-by-step instructions for
creating an analogy like the one displayed above utilizing the multi
attribute decision making procedure are described below.
Start
Step 1: Identifying the options
Step 2: Determining C
Step 3: Assembling a group to make key decisions
Step 4: Making decisions on one’s own
When l is 1, 2, …, n, the X and W powers (l) are determined.

Individual choice matrices are normalized in this step.
Step 5: Constructing the aggregated decision matrix

Step 6: Weighted-normalized individual decision matrices are computed
Step 7: Determining Bj, where j = 1, 2, …, n
Step 8: Locating Lj, where j = 1, 2, …, and so on, until n
Step 9: Identifying the people who make the decisions
Step 10: Calculating the relative merits of several options

Step 11: Choosing the best option
The primary concept was to allocate weights to decision makers based on

how consistent their opinions were with the most compromising approach
in each characteristic. A quadratic programming model was built in order
to determine the most and least compromising solutions in each charac-
teristic as a result of these findings. After that, the proximity coefficient of
each decision maker was taken into account while calculating his or her
weight in various aspects. Individual preferences were aggregated, and a
final choice was reached using the weights assigned to each preference.
Compared to earlier techniques, the suggested algorithm offers a num-
ber of benefits. For example, this model may be employed in a fuzzy envi-
ronment, which is advantageous. Second, distinct weights are calculated for
each decision maker in each decision-making characteristic, which is more
in line with the reality that different persons have varying competencies in
different attributes, as shown by the results of the experiment. In the third
place, as in the majority of prior research, the average of decision-maker
views is seen as the positive ideal answer, with the total square of distances
being kept to a minimum. It is also necessary to solve a quadratic pro-
gramming model in order to discover the negative optimal solution. Then,
these points serve as reference points for determining the relative impor-
tance of different decision-makers. An issue of selecting a maintenance
plan is investigated using the suggested technique, and a decision-making
committee comprised entirely of specialists was established to solve the
problem. We anticipate that adopting various expert weights will result in
findings that are more acceptable to a broader range of participants.
1.5 Conclusion
Microsoft, Amazon, and Google are leading the way in low-cost cloud
computing technologies. Because of security concerns, it is a key roadblock
to widespread usage of cloud computing. A cloud’s infrastructure security
is very important. There have been several investigations on the safety of
cloud infrastructure, but there are still some unanswered questions and
new difficulties to contend with. The current cloud architecture raises a
number of security issues, which are detailed in this article. Infrastructure-
related challenges affecting cloud computing’s business model were exam-
ined in the study. The literature-based solutions to security concerns at
different levels were also examined in this research. In order to help in their
settlement, previously unresolved issues are made public. Flexible, elastic,

and multi-tenant cloud architectures have revealed a slew of new problems
at every stage of the infrastructure lifecycle. Because of the vast variety of
issues it may bring, multi-tenancy has the biggest influence on all levels of
infrastructure.
Acknowledgment
We, the authors of this book chapter, would like to express our thanks to
the late Mr. Panem Nadipi Chennaih for the support and development of
this book chapter, it is dedicated to him.
References
1. Mishra, R., A fuzzy approach for multi criteria decision making in web
recommendation system for e-commerce. 2013 Eleventh International
Conference on ICT and Knowledge Engineering, pp. 1–4, 2013.
2. Yusuf, H. and Panoutsos, G., Multi-criteria decision making using fuzzy logic
and ATOVIC with application to manufacturing. 2020 IEEE International
Conference on Fuzzy Systems (FUZZ-IEEE), pp. 1–7, 2020.
3. Morente-Molinera, J.A., Wang, Y., Gong, Z.-W., Morfeq, A., Al-Hmouz, R.,
Herrera-Viedma, E., Reducing criteria values in multi-criteria group deci-
sion making methods using hierarchical clustering methods and fuzzy ontol-
ogies. IEEE Trans. Fuzzy Syst., 30, 6, pp. 1585–1598, 2021.
4. Zimmermann, H.J. and Sebastian, H.J., Application of fuzzy logic to engi-
neering design and configuration problems-a survey. Proceedings of IEEE 5th
International Fuzzy Systems, vol. 2, p. 1120, 1996.
5. Yeh, C.-H. and Deng, H., An algorithm for fuzzy multi-criteria decision
making. 1997 IEEE International Conference on Intelligent Processing Systems
(Cat. No.97TH8335), vol. 2, pp. 1564–1568, 1997.
6. Darestani, S.A., Tadi, A.M., Mirzaei, S., Evaluation of projects risks for dairy
industry using best-worst multi-criteria decision-making. 2020 IEEE 7th
International Conference on Industrial Engineering and Applications (ICIEA),
pp. 1110–1115, 2020.
7. Khan, S.A. and Rehman, S., On the use of unified and-or fuzzy aggrega-
tion operator for multi-criteria decision making in wind farm design pro-
cess using wind turbines in 500 kW–750 kW range. 2012 IEEE International
Conference on Fuzzy Systems, pp. 1–6, 2012.
8. Diana, A. and Solichin, A., Decision support system with fuzzy multi-
attribute decision making (FMADM) and simple additive weighting (SAW)
in laptop vendor selection. 2020 Fifth International Conference on Informatics

and Computing (ICIC), pp. 1–7, 2020.
9. Chatterjee, K., Kar, M.B., Kar, S., Strategic decisions using intuitionistic fuzzy
vikor method for information system (IS) outsourcing. 2013 International
Symposium on Computational and Business Intelligence, pp. 123–126, 2013.
10. Han, Z.Q. and Qin, H.Y., An incomplete probabilistic linguistic multi-
criteria group decision-making method based on statistical variance. 2021
33rd Chinese Control and Decision Conference (CCDC), pp. 5985–5989, 2021.
11. Sürmeli, G., Kaya, İ., Erdoğan, M., A fuzzy multi-criteria decision mak-
ing approach for choosing a logistics center location in Turkey. 2015 6th
International Conference on Modeling, Simulation, and Applied Optimization
(ICMSAO), pp. 1–6, 2015.
12. Tyagi, S.K., Sharma, S.K., Shukla, V.K., Interpretive structural modelling
using fuzzy linguistic information. 2019 4th International Conference on
Information Systems and Computer Networks (ISCON), pp. 347–351, 2019.
2
Application of Fuzzy Logic in the
Context of Risk Management
Sudipta Adhikary* and Kaushik Banerjee
Department of LAW, Brainware University, Barasat, West Bengal, India
Abstract
There have been numerous successful applications of industrial risk analysis to
power, nuclear, petroleum, and chemical sites since the end of World War II. Every
one of us will encounter situations in the course of a typical day where we will
instinctively try to avoid taking measures to lessen or manage potential danger.
The objective of a risk calculation is to identify and quantify the potential dangers
associated with a certain situation or event. Assessing the potential for harm and
identifying the specific elements that pose the greatest threat is the primary goal
of risk management. The goal of this risk assessment procedure is to quantify risks
and select suitable controls. In this research, we apply fuzzy logic to the problem
for creating a fuzzy model that may be used to better handle the uncertainties that
crop up at every stage of the risk calculation procedure.
Keywords: Risk management, fuzzy logic, fuzzy model, risk calculation, matrix
2.1 Introduction
Very often, we confronted with a variety of situations that prompt us to
think in terms of eliminating or at least mitigating risk.
13
Start
System definition
Hazard identification
Analysis of
options for Risk assessment
further risk
minimization
No Risk
accepted?
Yes
Safety
Figure 2.1 Evaluation of risk as it occurs.
Risk evaluation and assessment can be done by maintaining proper

sequence demonstrated in flow diagram (Figure 2.1).
Controlling risks entails, by definition:
• Coverage under insurance.

• Respecting legal norms and constitutional mandates.
• In the event of an emergency, it is important to make sure
that we do not repeat past blunders that is one way to rely on
our own personal experience.
• Leaning on one’s own knowledge and instincts to establish
safety measures.
2.2 Objectives of Risk Management

The aforementioned methods fail to consider regarding the needs of prop-
erly identifying all risks, allocating sufficient resources, and choosing effec-
tive controls. This led to the development of what we now call it as “risk
assessment,” which is basically the most accepted strategy to reduce risk at
a reasonable cost. The goal of this risk assessment procedure is to quantify
risks and select suitable controls. Risks are quantified when we determine
their relative size and significance. Thus, we can determine which threats
require immediate action and which ones can wait, and then divide our
limited resources among them accordingly. To determine the best means
of mitigating each risk, it is necessary to weigh the level of risk that can be
tolerated against the cost of implementing the controls that would reduce
Fuzzy Logic for Risk Assessment 15
the risk most efficiently. The risk assessment matrix is a useful instrument
for making qualitative evaluations of potential dangers. The concept of risk
as the potential for undesirable outcomes in the event of a given accident
scenario, along with the frequency with which such outcomes occur, is the
foundation of the risk matrix.
2.3 Improved Risk Estimation

Risk level = Consequence × Likelihood. We need to evaluate both the
potential negative outcomes of a scenario and the chances of those negative
outcomes occurring in order to get a sense of the amount of risk associated
with that scenario. There are two methods for calculating potential danger.
2.3.1 Point-Wise Calculations on a Curve

We will start with “single point estimation,” which describes the first
method. As a proxy for the entire area under the curve, this method only
estimates a single node on the risk profile graph. If a fire were to break out
in the structure, it would cause $1 million in damages, and the probability
of such a loss (the annual fire risk) is 5%. Hence, the annualized risk is
$5,000 since risk = capital × duration = $1,000,000 × 0.005% every year.
2.3.2 Estimation of a Curve

The alternative technique is known as “multiple point estimations.” This
involves making many point estimates on the curve in order to make an
approximation, and then making an area estimate for the resulting approx-
imation. If we take the same example and try to estimate the potential
threat in connection to a peril of a house, we may have a need to calculate
probability that a peril would cause the destruction of $10,000 and every
other possible consequence to get an accurate picture of the total poten-
tial threat regarding the concern peril in connection with the mentioned
property. The area under the curve provides a better representation of the
level of threat involved. This presents a bit of a challenge, as not only is
estimating the area under the curve more difficult than estimating the area
around a point (which can be done with the basic C × L computation), but
it also takes significantly more time. With the exception of extremely large
risks with prohibitively expensive potential controls, single-point approxi-
mations are usually adequate for most risks. A decent place to begin when
trying to sort out the big risks from the little ones is using a single point
Likelihood/year
0.1
0.01
0.001
$1 $1,000 $10,000 $1,000,000
Consequences
Figure 2.2 Curve used to estimate risk.
risk quantification. If necessary, we can apply more complex strategies to

more significant threats (Figure 2.2).
2.3.3 Accuracy in Quantification is Raised

It is best to use a number and a unit to assess both the financial costs (in
dollars) and the likelihood (in percentage/year). Once the estimates have
been made, they are plugged into the formula C × L = R. A dollar per year
becomes the risk measurement. For more information on how to calculate
risk with a single data point, see the preceding example. Given that we can
only use a single point for presenting the zone which is confined within the
potential risk curve, it makes sense to pick the point through the highest
risk value, defined here as the consequence multiplied by the likelihood
value.
2.3.4 The Problems with the Basic Quantification Approach

• Assessing the probability of a lesser implication: The
annual risk from a peril (even a small fire) in a construc-
tion (multi-storeyed building), for instance, could be as high
as 20%, though the risk of the mentioned peril completely
abolishes the construction could be as low as 1%. Since the
probability of a fire occurring is only 1%, estimating the fire
risk by using the probability of a fire occurring (20%) would
result in a twentyfold overestimation. Use only the estimated
probability of experiencing the negative outcomes you have
calculated.
• Gauging intrinsic risk: Inherent risk is the amount of dan-
ger present when safeguards aren’t in place. Risk assessments
that attempt to quantify Inherent Risk were widely used, but

this practice has fallen out of favor as more people become
aware of its limitations.
• A few of many flaws in the inherent risk approach: A risk
assessment is meant to aid in the allocation of resources
according to the requirements of the actual business envi-
ronment. When we eliminate constraints in a controlled
manner, we are not measuring the real world. This approach
also has been argued, as the potential controls are fallible,
the integral risk creates an accurate snapshot of actual hap-
pening if they did fail. Of course, we were not in the actual
world here. Be sure to factor in every potential outcome
when calculating the repercussions. If we do not include all
possible negative outcomes, our estimate of the risk is bound
to be low.
2.4 Threat at Quantification Matrix

When it comes to quantifying risks, the risk quantification matrix is widely
used. Depending on the matrix design, it can be either relatively accu-
rate or very inaccurate. Examples of possible matrix mistakes include the
following:
2.4.1 Qualitative Matrix

The repercussions and probabilities are measured in these matrices using
only words, with no defined units. Consequences and probabilities are dif-
ficult to quantify in words.
2.4.2 Errors in Scaling

If the matrix’s consequences and likelihood scales are not uniform, this
type of error is incorporated. If the risks are located near the extremes of
the significance and probability scales, the same mistakes are made.
2.4.3 Band Width at Various Scales

The wide ranges of the likelihood and consequence categories are to blame
for these errors. Affordable means between $250,000 and $1M, while mod-
erate means between $1M and $2M. This range of uncertainty means that
Almost
Certain 5 6 7 8 9
> 1 : 1y
Likely
1 : 1y
4 5 6 7 8
Likelihood
Possible
1 : 5y
3 4 5 6 7
Unlikely
1 : 10y
2 3 4 5 6
Rare
< 1 : 30y
1 2 3 4 5
Insignificant Minor Moderate Major Catastrophic

<$100k $100k–$250M $250k–$1M $1M–$5M > $5M
Consequence
- Extreme
- High
- Moderate
- Low
Figure 2.3 Scaling and bandwidth approximation errors.
risks that are potentially orders of magnitude apart in scale are treated the
same. Thus, a $50,000 annual risk ($1,000,000 × 1/20th of a year) and a $50
annual risk ($250,000 × 1/5th of a year) may both be classified as a level 3
risk (Figure 2.3).
2.5 Fundamental Definitions

Manufacturing and processing always carry some degree of risk, which is
why there are safety standards in place.
2.5.1 Positioning Statement

Managing risk efficiently and cheaply precludes spending a lot of time
measuring each possible threat. On the contrary, we need precise risk
quantification so that we can allocate resources sensibly and decide which
controls will most efficiently bring down risks to an acceptable level while
staying within budget. The complexity of a risk assessment should be in
line with the magnitude of the risk, the cost of controls, or the potential
value added from accompanying the valuation. This concept is known as
the “value proposition.”
Intolerably
high risk
Tolerable
risk
Broadly
accepted
risk
Figure 2.4 Risk under the tolerance limit.
2.5.2 Risk Under the Level of Tolerance

Tolerable levels of risk can be exceeded, but there is a breaking point. To
the point when the benefits outweigh the risks. The range of acceptable risk
is between these two extremes (Figure 2.4).
2.5.3 Risk Elimination

Taking precautions to lessen a risk’s occurrence is what we call “risk reduc-
tion.” However, the question of how safe is safe remains. To what extent,
then, must the danger be mitigated? In order to address these inquiries,
one can choose one of two routes. The quantitative approach involves tally-
ing up the potential dangers posed by each potential hazard and contrast-
ing the total with the acceptable level of risk. Using a qualitative ranking
that considers both likelihood and consequences is at the heart of the qual-
itative approach.
2.6 Fuzzy Logic
For the purpose of representing knowledge, fuzzy logic employs a set of

mathematical philosophies constructed on grades of membership. Issues
of partiality and impenetrability in membership and in the truth are dis-
cussed. It seeks to mimic human cognition by emulating how we under-
stand the world, make decisions, and use common sense in a linguistic
context.
0 0 0 1 1 1 0 0.2 0.4 0.6 0.8 1

(a) Boolean Logic (b) Multi-valued Logic
Figure 2.5 Multi-valued vs Boolean logic.
Young Old
0.6 Old
0.4 Young
40 50 70 Age
Figure 2.6 Classification of fuzzy age.
X µ(X) Inference µ(Y) Y

Fuzzifier Defuzzifier
Engine
Fuzzy
Knowledge
Base
Figure 2.7 Fundamental of fuzzy structure.
A fuzzy inference system has three main parts (Figures 2.5–2.7):
• A rule base with a collection of fuzzy guidelines.

• A record with the definition of the membership function
utilized in the fuzzy principles.
• A cognitive machinery that applies the extrapolation way to
the principles and the input evidences to reach at a sensible
conclusion.
2.7 Risk Related to Fuzzy Matrix

Following from R. Nait’s Fuzzy Risk Graph Model for determining Safety
Integrity Level, the goal of this study is to laid on fuzzy logic to a tradi-
tional risk matrix, taking advantage of the method’s ability to smooth out
rough edges and pinpoint solutions to problems without clear cut offs or
predetermined constraints. Fuzzy sets, defined by a membership function

taking on values between 0 and 1, need the selection of a category for each
variable prior to application. In this study, the usual risk matrix depicted in
Figure 2.8 is employed.
Classes of sternness and levels of likelihood are presented in MIL-STD-
882D’s risk matrix layout, which looks like this:
According to Table 2.1, we have assigned reasonable intervals to each
risk level for the purposes of this study.
This study found that the Gaussian membership function was the most
common and intuitive option for such structures. There are nine variables
in a fuzzy risk assessment matrix, and Figure 2.9 shows the membership
function in connection with risk management.
The probability is given over a logarithmic scale, as presented through
Figure 2.9. because the likelihood values are logarithmically spaced. Fuzzy
logic was implemented in Matlab to create a fuzzy risk matrix (Table 2.2).
Due to its widespread acceptance in the field of capturing expert knowl-
edge and its ability to allow us to define the proficiency in a more instinc-
tive manner, the Mamdani technique was chosen over the sugeno method.
However, the Sugeno method is very appealing in control problems, espe-
cially for dynamic nonlinear systems. Parameters regarding risk graph had
been tested with every feasible combinations of crisp input patterns using
all five Defuzzification techniques (Table 2.3). A range of three values was
used to verify each crisp input, as shown in Figure 2.10.
Consequency
(Severity) Catastrophic Critical Marginal Negligible
Probability
I II III IV
(Likelihood)
Frequent A HIGH HIGH SERIOUS MEDIUM

Probable B HIGH HIGH SERIOUS MEDIUM
Occasional C HIGH SERIOUS MEDIUM LOW
Remote D SERIOUS MEDIUM MEDIUM LOW
Improbable E MEDIUM MEDIUM MEDIUM LOW
Figure 2.8 Risk graph model.

Table 2.1 Risk level

(through the lens of a
case study).
From To
L 0 25
M 25 50
S 50 75
H 75 100
IMPROBABLE REMOTE OCCASIONAL PROBABLE FREQUENT

1
0.5
0
–5 –4.5 –4 –3.5 –3 –2.5 –2 –1.5 –1 –0.5 0
input variable “LIKELIHOOD”
NEGLIGIBLE MARGINAL CRITICAL CATASTROPHE

1
0.5
0
4 4.2 4.4 4.6 4.8 5 5.2 5.4 5.6 5.8 6
input variable “SEVERITY”
LOW MEDIUM SERIOUS HIGH

1
0.5
0
0 10 20 30 40 50 60 70 80 90 100
input variable “RISK”
Figure 2.9 Membership function (risk management).

Table 2.2 Suggestions for rating the seriousness of mishaps.

Description Category Environment, safety, and health result criteria
Catastrophic I Could result in death, permanent total disability, loss
exceeding $1M, or irreversible severe environmental
damage that violates law or regulation.
Critical II Could result in permanent partial disability, injuries or
occupational illness that may result in hospitalization
of at least three personnel, loss exceeding $200K but
less than $1M, or reversible environmental damage
causing a violation of law or regulation.
Marginal III Could result in injury or occupational illness
resulting in one or more lost work days(s), loss
exceeding $10K but less than $200K, or mitigatible
environmental damage without violation of law
or regulation where restoration activities can be
accomplished.
Negligible IV Could result in injury or illness not resulting in a lost
work day, loss exceeding $2K but less than $10K, or
minimal environmental damage not violating law or
regulation.
Table 2.3 Suggested mishap probability levels.

Fleet or
Description* Level Specific individual item inventory**
Frequent A Likely to occur often in the Continuously
life of an item, with a experienced.
probability of occurrence
greater than 10-1 in that life.
Probable B Will occur several times in Will occur
the life of an item, with a frequently.
less than 10-1 but greater
than 10-2 in that life.
(Continued)
Table 2.3 Suggested mishap probability levels. (Continued)

Fleet or
Description* Level Specific individual item inventory**
Occasional C Likely to occur some time in Will occur several
the life of an item, with a times.
less than 10-2 but greater
Remote D Unlikely but possible to occur Unlikely, but can
in the life of an item, with reasonably be
a probability of occurrence expected to
less than 10-3 but greater occur.
Improbable E So unlikely, it can be assumed Unlikely to occur,
occurrence may not but possible.
be experienced, with a
probability of occurrences
less than 10-6 in that life.
Figure 2.11 summarizes the results of the test combinations.

One of the test cases with the input pattern for the parameters indicated
in Table 2.4 is depicted in this picture.
The total MF’s may readily express the quantification of the potential
threat in a fuzzy form, eliminating the need to defuzzify through the applica-
tion of FIS. Figure 2.12 shows that, once we remove the defuzzification chal-
lenge, the pattern of the resulting MF indicates a moderate amount of risk.
Parameter Range
Range for
test value-2
Range for Range for

test value-1 test value-3
Figure 2.10 Graph regarding fuzzy risk model.

52 46
128 134
SOM Defuzzification LOM Defuzzification
39 32
141 148
Centroid Defuzzification Bisector Defuzzification
17
163
MOM Defuzzification
Figure 2.11 Test for safety of fuzzy model by using centroid, bisector and MOM
defuzzification.
Table 2.4 Test input design for security through fuzzy

graph.
Linguistic value Calibrated range
Likelihood Probable (B) 0 : -1
Severity Negligible (IV) 2 : 10
RISK (CONV) M
Linguistic value Range test value
Likelihood Cc -0.75
Severity Fa 4.2
RISK (CRISP) 33
LIKELIHOOD = –0.75 SEVERITY = 4.2 RISK = 33

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
–5 0 4 6
Figure 2.12 Aggregation of fired membership functions.
2.8 Conclusion
This study provided evidence that fuzzy logic theory may be used for
decision-making tool While conventional risk matrices are easy to use,
they can produce wildly varying outcomes. The use of qualitative risk fac-
tors is an inherently subjective aspect of the risk matrix approach. On the
flip side, there is a contradiction between the numerical interpretation of
risk characteristics using crisp intervals and the natural gradual transition
between intervals. This paper demonstrates that the fuzzy logic is very
much fruitful when it is used to traditional risk matrix for finding out risk.
Bibliography
Duffus, J. and Worth, H., Risk assesment & risk management, in: The Science of
Chemical Safety Essential Toxicology #6, IUPAC, Scotland, 2001.
Matthew, T., Advanced Risk Assessment, Riskcentral, 2013.
BASF, Risk matrix as a tool for risk assesment. Workshop Bieleschweig 4,

14./15/9.2004.
Nait-Said, R., Zidani, F., Ouzraoui, N., Fuzzy risk graph model for determining
safety integrity level. Hindawi Publishing Corporation, 2008, Article ID
263895, 2008.
Understanding safety integrity level. Magnetrol Int., Bulletin 3, 41-299.3, 49–54,
2011.
Martin McNeill, F. and Thro, E., Fuzzy Logic: A Practical Approach, Academic
Press Professional, Inc., San Diego, CA, USA, 1994.
3
Use of Fuzzy Logic for Controlling
Greenhouse Environment: A Study
Through the Lens of Web Monitoring
Kaushik Banerjee* and Sudipta Adhikary
Department of LAW, Brainware University, Barasat, West Bengal, India
Abstract
We discuss about the construction and application of an economical system
employing through the fuzzy logic for observing and controlling of greenhouse
from remote location. This system was designed to monitor and operate the green-
house remotely. In order to monitor and manage the ambient temperature, water
content in soil, relative moisture content, and lighting, an Arduino Mega board is
introduced. This will be used as an integral part of the control system. An ‘Arduino
Ethernet Shield’ was used so that the device could maintain contact with the web-
site. As a result, it was feasible to set up an area-based(local) network in order
to observe and regulate the mentioned variables of greenhouse through an auto-
mated mode or a manual fashion. Without using a mathematical model of the
plant, it was also possible to demonstrate that fuzzy logic is an excellent method
for controlling nonlinear systems. Through our research, we were able to make the
most efficient use of available resources while developing the device(greenhouse).
Keywords: Fuzzy control, green house, web monitoring, fuzzy logic, Arduino
3.1 Introduction
The required administration for creation a greenhouse provides two sig-
nificant obstacles. It starts with worthful use of the land and followed by
most economical uses of available power and water. As a result, the usage
29
of high-tech greenhouses that are focused on the creation of optimal tem-

perature conditions in order to boost plant growth while simultaneously
lowering production costs and energy consumption is required.
3.2 Design (Hardware)

The arrangement that was built to manage the variables in the greenhouse
is shown as a block diagram in Figure 3.1, which may be seen below.
3.2.1 Sensor for Measuring Soil Moisture

A sensor is a hinge on the resistive qualities of the concerned soil mass and
a voltage divider that uses dual conductors. These were created to observe
the entire region that is the subject of this research, so that these variables
may be measured (Figure 3.2). The setup that was allowed for obtaining an
equivalent reception that was dependent on the soil water content. If the
soil is damp, the value of the resistor R1 will drop, which will cause the out-
put voltage to rise. On the other hand, for the dry soil, the value of R1 will
rise, which will cause the output voltage to fall. The fact that the sensor was
calibrated via the use of experimental experiments is notable. The setup is
shown in Figure 3.2.
Disturbances
Relative
Temperature Humidity
Greenhouse
Monitoring and Temperature and Humidity Sensor

Control
Soil Moisture Sensor
Arduino
Arduino
PC Ethernet
Mega Final Control
Shield Elements
Electrical Pump
Humidifier
Arduino Relay Extractors
Modules Heater
Lighting
Figure 3.1 Framework of the block diagram.

Fuzzy Logic for Greenhouse Environment 31
+5VDC
Conductors
Analog Pin
R2
R1 (Soil)
Figure 3.2 Device for measuring water content.
VDD VDD
5k Pin1
Arduino Pin2
DATA
Digital DHT 11
I/O Pin
Pin4
(a)
GND
(b)
Figure 3.3 (a) DHT11. (b) Circuit diagram.
3.2.2 Sensor for Measuring Humidity and Temperature

In order to achieve both high dependability and outstanding long-term
stability, a DHT11 sensor had been used, which provides output digitally
based on relative moisture content and temperature. It was decided to
design a single-wire protocol for the purpose of transmitting data in packs
of 8 bits each. The DHT11 sensor as well as its assemblage to the Arduino,
3.3 Programming Arduino Mega Board

For temperature and relative moisture content for the fuzzification and
defuzzification processes is shown through the help of flow chart in Figure
3.4. A comparable method was used for both the watering and lighting
phases of the concerned project.
START
Reading Configuration Parameters

in EEPROM
Data Acquisition of the Temperature

and Humidity Sensor
Fuzzification
(Determining Membership
Values)
Defuzzification
(Control Actions)
END
Figure 3.4 Chart for the fuzzification and defuzzification processes.
3.3.1 Fuzzification
There were found to be 10 fuzzy sets for defining levels that were associated
with the input variables. The triangular shapes were chosen to represent
membership functions because of their ability to affect the degree of mem-
bership, which can range from 0 to 1. Temperature and relative moisture
content for the concerned fuzzy set are shown in Figure 3.5, which is a rep-
resentation of the triangle function. A procedure is carried out according
to the criteria that the user sets on the website in order to guarantee that
the membership range is between 0 and 1.
3.3.2 Fuzzy Inference
By using the fuzzy matrix, different amalgamations of variables were pro-

duced, which allowed the control actions to be determined for any climatic
state inside the greenhouse.
MLT MBT OT MABT MALT

1
Membership Grades
0.5
0
(a)
MLRH MBRH ORH MABRH MALRH

1
Membership Grades
0.5
0
(b)
Figure 3.5 (a) Temperature. (b) Relative humidity.
3.3.3 Communication via Remote Connections and a Web

Server
A remote communication system between the user may be implemented
with the help of the Arduino Ethernet Shield. On these web servers
Figure 3.6 Web interface.

Web Interface
Configuration Panel: Set the initial system

values and press the ok button.
Server Status: Choose the

operating mode.
Automatic Manual
Fuzzy control enabled. Wait for the Manual configuration enabled.

system get to stabilize with the The user is able to manage the
defined values. final control elements (ON/OFF).
In both modes, the user is able to monitor the

configuration parameters, the final control elements
status and the environmental sensors values.
End
Figure 3.7 Web interface (flowchart).
(Figure 3.6), facts are displayed using HTML in order to generate a web
page based on readings. The HTML code that was appended was derived
from the sample code, and it was necessary in order to accomplish the
requirements of the prototype (Figure 3.7).
3.4 Implementation of a Prototype

A greenhouse made of acrylic was constructed so that the system could be
tested in it (Figure 3.8).
(a) (b)
Figure 3.8 Prototype setup for greenhouse.
3.5 Results
Obtained findings explored that the relative moisture content rises to a
value of 54%, which falls within the parameters that have been established.
The amount of time that was required for the variables to stabilize was 4
minutes.
Figure 3.9 illustrates how the system may maintain a constant tempera-
ture while simultaneously lowering the relative moisture content to 54%,
60
55
Humidity (%)
50
Relative
45
40
35
30
0 Minutes 4
(a)
50
Temperature (°C)
40
30
20
10
0
0 Minutes 4
(b)
Figure 3.9 Ranges are defined at set points (a), humidity, (b) temperature.
placing it within the parameters of the range that can be controlled by the
set points (Figure 3.10). Ten minutes were allotted for the stabilization
process.
Through the adaptation of this, the efficacy of irrigation system could be
tested. The length of each irrigation was 5 seconds, and there was a gap of
5 minutes between each irrigation. In all, nine irrigations were performed.
This variation ranged from 30% to 100%. On the other hand, after the
ninth irrigation, an average moisture level that remained constant at 93%
was measured, which confirmed that the system was operating as intended
(Figure 3.11.)
80
Humidity (%)
70
Relative
60
50
40
30
0 Minutes 10
(a)
40
Temperature (°C)
30
20
10
0
0 Minutes 10
(b)
Figure 3.10 Distinct range of the set points: (a) humidity, (b) temperature.
100
90
80
70
Soil Moisture
60
50
40
30
20
10
0
0 Minutes 5
Figure 3.11 Set points in the defined range (a) humidity, (b) temperature.
3.6 Conclusion
The findings have been illustrated the enhancements that are possible to
attain in terms of greenhouse temperature management via the use of a
fuzzy controller in conjunction with particle swarm optimization. It has
also been suggested that SD cards could be used in order to expand the
amount of storage space available on the prototype.
Bibliography
Azaza, M., Tanougast, C., Fabrizio, E., Mami, A., Smart greenhouse fuzzy log-
ic-based control system enhanced with wireless data monitoring. ISA Trans.,
61, 297–307, 2016. [CrossRef] [PubMed].
Castañeda, A. and Castaño, V.M., Smart frost control in greenhouses by neu-
ral networks models. Comput. Electron. Agric., 137, 102–114, 2017.
[CrossRef].
Cruz, E. and Hahn, F.F., Remote monitoring of greenhouse, in: Proceedings of the
American Society of Agricultural and Biological Engineers Annual International
Meeting, New Orleans, LA, USA, pp. 3554–3561, July 26–29, 2015.
Coates, J., Chipperfield, A., Clough, G., Wearable multimodal skin sensing for the
diabetic foot. Electronics, 5, 45, 2016. [CrossRef].
Groener, B., Knopp, N., Korgan, K., Perry, R., Romero, J., Smith, K., Stainback,
A., Strzelczyk, A., Henriques, J., Preliminary Design of a low-cost green-
house with open source control systems. Proc. Eng., 107, 470–479, 2015.
[CrossRef].
Iliev, O.L., Sazdov, P., Zakeri, A., A fuzzy logic-based controller for integrated
control of protected cultivation. Manage. Environ. Qual., 25, 75–85, 2014.
[CrossRef].
Jiang, J.A., Wang, C.H., Liao, M.S., Zheng, X.Y., Liu, J.H., Chuang, C.L., Hung,
C.L., Chen, C.P., A wireless sensor network-based monitoring system with
dynamic convergecast tree algorithm for precision cultivation management
in orchid greenhouses. Precis. Agric., 17, 766–785, 2016. [CrossRef].
Li, S.J., Li, M.Y., Wang, X.D., Design of greenhouse environment controller based
on fuzzy adaptive algorithm, in: Proceedings of the 27th Chinese Control and
Decision Conference, Qingdao, China, pp. 2644–2647, May 23–25, 2015.
Li, J. and Chong, S., An energy conservative wireless sensor networks approach for
precision agriculture. Electronics, 2, 387–399, 2013. [CrossRef].
Márquez, M., Ramos, J., Cerecero, L., Lafont, F., Balmat, J., Esparza, J., Temperature
control in a MISO greenhouse by inverting its fuzzy model. Comput. Electron.
Agric., 124, 168–174, 2016. [CrossRef].
Maher, A., Kamel, E., Enrico, F., Atif, I., Abdelkader, M., An intelligent system for
the climate control and energy savings in agricultural greenhouses. Energ.
Effic., 9, 1241–1255, 2016. [CrossRef].
Montoya, A., Guzmán, J., Rodríguez, F., Sánchez, J., A hybrid-controlled approach
for maintaining nocturnal greenhouse temperature: Simulation study.
Comput. Electron. Agric., 123, 116–124, 2016. [CrossRef].
Milik, A., Bajer, L., Krejcar, O., Design and realization of low-cost control for
greenhouse environment with remote control. IFAC-PapersOnLine, 48, 368–
373, 2015. [CrossRef].
Mesas, F., Verdú, D., Meroño, J., Sánchez, M., García, A., Open-source hardware
to monitor environmental parameters in precision agriculture. Biosyst. Eng.,
137, 73–83, 2015. [CrossRef].
Nikolidakis, S., Kandris, D., Vergados, D., Douligeris, C., Energy efficient auto-
mated control of irrigation in agriculture by using wireless sensor networks.
Comput. Electron. Agric., 113, 154–163, 2015. [CrossRef].
Nicolosi, G., Volpe, R., Messineo, A., An innovative adaptive control system to
regulate microclimatic conditions in a greenhouse. Energies, 10, 722, 2017.
[CrossRef].
Outanoute, M., Lachhab, A., Ed-Dahhak, A., Selmani, A., Guerbaoui, M.,
Bouchikhi, B., A neural network dynamic model for temperature and rel-
ative moisture content control under greenhouse, in: Proceedings of the 3rd
International Workshop on RFID and Adaptive Wireless Sensor Networks,
Agadir, Morocco, pp. 6–11, May 13–15, 2015.
Revathi, S. and Sivakumaran, N., Fuzzy based temperature control of greenhouse.
IFAC-PapersOnLine, 49, 549–554, 2016. [CrossRef].
Rivera, J., Raygoza, J., Ortega, S., Figueroa, A., Begovich, O., FPGA-based startup
for AC electric drives: Application to a greenhouse ventilation system.
Comput. Ind., 74, 173–185, 2015. [CrossRef].
Song, Y., Wang, J., Zhang, X., Greenhouse environment parameters optimization
and wireless monitoring based on maximize profit margin. Sens. Lett., 14,
1129–1137, 2016. [CrossRef].
Srbinovska, M., Gavrovski, C., Dimcev, V., Krkoleva, A., Borozan, V., Environmental
parameters monitoring in precision agriculture using wireless sensor net-
works. J. Clean. Prod., 88, 297–307, 2015.
Verma, H., Jain, M., Goel, K., Vikram, A., Verma, G., Smart home system based
on internet of things, in: Proceedings of the 3rd International Conference on
Computing for Sustainable Global Development, New Delhi, India, pp. 2073–
2075, March 16–18, 2016.
Van Beveren, P., Bontsema, J., Van Straten, G., Van Henten, E., Optimal control of
greenhouse climate using minimal energy and grower defined bounds. Appl.
Energy, 159, 509–519, 2015. [CrossRef].
Wang, S.W. and Zhang, C.L., Study on farmland irrigation remote monitoring
system based on Zig Bee, in: Proceedings of the International Conference
on Computer and Computational Sciences, Noida, UP, India, pp. 193–197,
January 27–29, 2015.
Wang, J.-M., Yang, M.-T., Chen, P.-L., Design and implementation of an intelligent
windowsill system using smart handheld device and fuzzy microcontroller.
Sensors, 17, 830, 2017. [CrossRef] [PubMed].
4
Fuzzy Logics and Marketing Decisions
Mohammed Majeed *
Marketing Department, Tamale Technical University, Tamale-Ghana, Ghana
Abstract
The goal of this chapter is to highlight areas of marketing that requires fuzzy logic
and how fuzzy logic affect marketing decisions. Fuzzy logic uses anthropologi-
cal knowledge and decision to assist acceptable perceptive in order to influence
a conclusion. The areas FL can be applied to marketing includes: communication
and advertising, customer service and satisfaction, customer segmentation, CRM,
pricing, evaluation of a product, uncertainty in the development of new products,
decision making, consumer nation identity (CNI), and quality of service. If the
problem of building a details and explanatory method for taking marketing deci-
sions based on fuzzy logic approaches is solved in this chapter, it will be impossible
to make an operative marketing decision based on a well-knowledgeable choice.
Keywords: Marketing, customer, fuzzy logic, decision making, product,

uncertainty
4.1 Introduction
The World Wide Web and other forms of information technologies are
becoming increasingly prevalent. As a result, the popularity of mobile
(web) sites continues to rise year after year. Since its inception in the mid-
60s of the 20th century, fuzzy logic has grown in popularity as one of the
most widely utilized artificial intelligence [18, 22] in theory and practice.
If you’re interested in learning more about fuzzy logic, you can check out
this page on investopedia for more information on the subject [5, 9]. Fuzzy
logic can be summarized as an appeal to one’s instincts. It’s all about trust-
ing your gut instincts and making the proper choice. Using fuzzy logic
*Email: tunteya14june@[email protected]; [email protected]
41
in the creation of marketing models gives marketers a strong tool [14].

If-then rules have emerged as a new technique for marketers to do mar-
keting audit, the fuzzy logic marketing model. This new modeling tech-
nique provides marketers with tools that are long-term and sustainable for
generating corporate performance. When it comes to marketing, today’s
consumers are more savvy and dynamic than ever before, as well as more
competitive and volatile. For example, a process expert’s report can be for-
malized and simulated using this tool. Secondly, it provides a simple solu-
tion to the modeling techniques that are complicated. The weighted sum
of the factors it considers determines the degree of effect. Fuzzy logic con-
stantly regards and incorporates situations of a different sort. As a result, it
is possible to build multicriteria methods that take expert knowledge into
account. This article’s goal is to demonstrate how fuzzy logic can be used in
business and marketing.
The significance of the research in the real world. Fuzzy logic meth-
ods are currently being used in marketing and management to generate
solutions. As a result of today’s high-functioning software products on
the IT market, participants in business processes, particularly those in
small enterprises, are unable to achieve these requirements for user liter-
acy. There must be a clear, adaptable marketing information, as well as a
methodical area that matches to the executive judgments in order to give
the necessary functionality for management decision making. This ensures
that there will be no false negatives, such as concluding a given segment
can be hired from, when in fact this would lead to a service failure at a
particular location in the network.
4.2 Literature
4.2.1 Fuzzy Logic (FL)
According to Zadeh [32], “fuzzy sets” (fuzzy logic’s essential concept) were
first proposed in 1965. To describe the human mind, it was used. According
to fuzzy logic, logical assertions are valued according to their degree of truth-
fulness [8]. For fuzzy logic, it is important to consider the utility of impreci-
sion as well as the relevance of accuracy [16]. There are numerous advantages
to using a wide range of artificial intelligence approaches in combination
with each other, including as genetic algorithms, microbial fodder, fuzzy
logic, and neural networks [20]. Fuzzy logic has been utilized successfully
in energy and resource control [9, 10, 17]. It considers a number of factors,
and the combined weight of those factors determines the degree of influence.
Fuzzy Logics and Marketing Decisions 43
When it comes to time series forecasting, fuzzy logic is an excellent choice [1,
16] for building a marketing channel evaluation model.
4.2.2 FL Application in Marketing

Fuzzy mathematics is beginning to alter viewpoints in a variety of busi-
ness and marketing fields [14]. It is possible to gain insights into the type
and strength of correlations between variables using traditional statistical
approaches such as regression analysis [8]. The competitor’s growth strat-
egy, product life cycle, product kind, and operating assets are the inputs to
the fuzzy inference system that serves as a decision support system.
4.2.2.1 Communication and Advertising

According to Abdoli and Sheykhemaeli [1] Decision making for both cus-
tomers and marketing managers can be improved by using a fuzzy-based
inference engine to invest in targeted promotions with a laborious collec-
tion of contacts. A fuzzy software tool can be used by marketing experts to
better identify new customers by accounting for data diversity and inaccu-
racy [7].
4.2.2.2 Customer Service and Satisfaction

Additionally, the fuzzy logic methodology is used sense impracticable cus-
tomer desires. The consumer can, alongside vendor, actually identify and
modify them [15]. Along with the improving quality in customer expe-
rience, it will raise the number of transactions and client happiness, the
production capacity will also increase [24]. Improvements efficacy would
be realized as anticipated and might boost purchaser fulfilment. At the
conclusion, it is expected to reduce customers’ complaints, particularly the
distributing efficiently and successfully [28].
4.2.2.3 Customer Segmentation

All potential and present consumers are segmented into internal homog-
enous and heterogeneity subsets (customer segments) based on market
reactions, as well as the work in one or more customer categories [8].
Customer splitting up based on user preferences is made possible through
the application of fuzzy logic. There is evidence to support the use of
customer segmentation within an e-mobility service to guide customers
toward more environmentally friendly modes of transportation (such
as electric cars, public transportation, and trains, for example) [15, 29].
Qualitative and quantitative characteristics are frequently seen in data.
Additionally, non-numeric qualities need to be taken into account. Class
descriptions can be improved by the addition of linguistic phrases, such as
“comfortable,” “medium comfortable,” and “luxurious.” These numbers are
not really clear. It is feasible to cope with these ambiguous consumer data
by using fuzzy logic, fuzzy categorization, and compensating fuzzy logic. It
is feasible to treat each customer individually with fuzzy logic [21].
4.2.2.4 CRM
Data mining can benefit from the fuzzy model. Using direct marketing,
this example examines whether to visit the consumer in person, write him
a letter, or not to touch him in any way all [15]. Six input variables, each
with three or four qualities, three rule boxes, and a single output variable,
each with three attributes, are used to solve the problem [13].
4.2.2.5 Pricing
Studies on how attitudes and culture influence customer purchasing deci-
sions are becoming more common as marketing research expands [30] as
are studies that use fuzzy logic and look for behavioral reactions of cus-
tomers. Improved accuracy in the indorsed value points and a better client
involvement can be achieved by using fuzzy logic [12]. Because it reduces
expenses, fuzzy logic has a wide range of applications. This tolerance is
exploited to produce tractability, resilience, and a low solution cost in fuzzy
logic [13]. When using the fuzzy price system to make pricing decisions for
each of a company’s products in a market, all of the competitor’s pricing
changes and their product mix are used as input to the algorithm [14].
4.2.2.6 Evaluation of a Product

Clustering is an effective decision-making tool for many management
tasks. As the name suggests, clustering is the process of organizing a col-
lection of items so that those belonging to the same group (or cluster) are
much more related to one another than those in other categories are (clus-
ters). Clusters are commonly thought of as being composed of individuals
who live in close proximity to one another. It is possible to combine fuzzy
clustering with neural networks and evolutionary algorithms to gain the
benefits of both [13]. The technique may be applied to any shop’s prod-
ucts, and the more products and characteristics there are, the better off the
shop will be by adopting it. Therefore, with fuzzy logic, buyers can be self-
assured that a manufactured goods is chosen according to all their favor-
ites and necessities [13].
4.2.2.7 Uncertainty in the Development of New Products

Uncertainty is inherent in the essence of NPD [11]. There is a lack of con-
fidence in the new technology’s market potential among NPD managers.
As a result, new product development (NPD) executives are unsure of how
to use the latest know-hows to create new goods that match client expec-
tations. Customers’ inability to express their wants, as well as managers’
struggles to translate technology developments into product features and
benefits, both contribute to this uncertainty. Lastly, senior management is
confronted with ambiguity over the amount of cash to invest in pursuit of
quickly shifting markets as well as the timing of investment.
4.2.2.8 Decision Making

The praxiological approach to economics will be used to compile, analyze,
and systematize a firm’s management judgments when making market-
ing decisions [26]. It provides as a guide and procedures for predicting,
decision making, and evaluations in an environment of ambiguity, ambiv-
alence, impressions, and subjective. In order to provide customers with
accurate information about their preferences, fuzzy logic should be used.
To better understand their customers’ preferences, managers and research-
ers might use fuzzy logic to construct a collective decision-making model
[33]. A study by Al Ganideh et al. [3] asserted that fuzzy logic provides
more accurate insights than regression analysis to international business
researchers studying customer behavior. In today’s global economy, data
warehouses are becoming a major growth area. Data analysis and deci-
sion support operations can be improved by using it as a key component
of business intelligence (BI) [19]. Intelligence and knowledge can be suc-
cessfully stored in an enterprise-wide data warehouse through the fuzzy
approach. To learn more about client preferences, managers might use it to
construct fuzzy logic decision making models.
Management decision-making processes are based on data mining sys-
tems (fuzzy information retrieval), incorporated smart system (ISS) [26].
Facing real-world scenarios, complicated and dynamic situations, and
fuzzy logic is used to handle a number of challenges, notably those linked
to control of complex industrial processes and decision systems in general,
the resolution and compression of data [6].
4.2.2.9 Consumer Nation Identity (CNI)

The influence of socio-psychological variables such as dogmatism, conser-
vatism, and world-mindedness on national identification levels is explored
using fuzzy logic approach. It is important for international marketers to
understand the concept of national identity, because customers all around
the world are exposed to more foreign products in their native nations [2].
Marketing managers in both the United States and abroad can benefit from
a better understanding of consumer identity. Understanding the CNI helps
marketers boost global marketing operations including promotional cam-
paigns and successful and efficient worldwide strategies [23, 27]. As stated
by Al Ganideh and Aljanaideh [4], fuzzy logic can help researchers and
managers by offering accurate values for particular consumers based on
their dogmatism, conservatism, and global mindedness scores (collectively
describes CNI).
• The term “world-mindedness” refers to a person’s attitude

toward humanity as a whole [4].
• Conservatism refers to a desire to hold on to long-standing
traditions and social structures, and to only gradually intro-
duce new ones [4, 2].
• A dogmatic way of seeing the world in which everything is
either black or white is called dogmatism [4].
4.2.2.10 Quality of Service

An alternate approach to analyzing service quality is becoming accepted
by researchers in the field of management sciences [30] using fuzzy set
theory [25, 31]. As a technique to combine subjective and objective knowl-
edge, triangular fuzzy numbers (TFNs) are utilized to construct an overall
quality index for each of the segments evaluated in the research [32]. As
a result, FL shows that triangular fuzzy numbers and resemblance to ideal
solutions can be used to measure customer satisfaction.
4.3 Conclusion
Fuzzy logic is a bridge between mathematics and human thought and
action. Fuzzy logic and its applications in management and marketing
research are discussed in this chapter. Intelligence and information can
be successfully stored in an enterprise-wide data warehouse using the
fuzzy approach. In addition, managers can use fuzzy logic to construct

group decision models to better understand client preferences and pref-
erences. Fuzzy logic is promoted in this chapter as a method that academ-
ics might use in their business and marketing studies. Research should be
done to see if fuzzy logic mathematics can forecast productivity based on
feedback using various marketing approaches. For instance, all kinds of
items can profit from adopting this approach, and the more products you
have to choose from, the more features you have to choose from, the bet-
ter. Customer preferences can be expressed in natural language and then
translated into predicate calculus utilizing the properties of compensating
fuzzy logic. One of the choices for multicriteria decision making is to use
fuzzy logic in a marketing firm, such as a retail shop. It is critical, then, to
aid in the management of risk and uncertainty by providing fuzzy logic
approaches for decision making that improve the efficiency of businesses.
Only if the difficulty of establishing an information and analytical system
for making marketing decisions based on fuzzy logic methods is overcome
in this chapter can a successful marketing decision based on informed
choice be produced. The answer in this chapter is both relevant and prac-
tical. Decision assistance in services marketing can be accomplished with
the satisfactory classification of target settings and the provision of suit-
able resources, such as trustworthy information and statistical, intelligent
agents of marketers.
4.4 Further Studies

In the future, worldwide internet users will rise at a faster rate. In recent
years, e-shopping has seen a steady rise in popularity due to the expansion
of the Internet and the accompanying increase in users. An investigation
into fuzzy logics analysis of consumer loyalty in the e-shopping environ-
ment is the best course of action here.
References
1. Abdoli, M. and Sheykhesmaeli, S., Identification of factors influencing the
evaluation of marketing channels. Second National Conference on Marketing
Research, Tehran, 2015, (in Persian).
2. Al Ganideh, S.F., Consumer ethnocentrism in the Jordanian market: The
impact of socio-psychological variables on consumer ethnocentric tenden-
cies. J. Bus. Retail Manage. Res., 5, 1, 11–20, 2010.
3. Al Ganideh, S.F., El Refae, G., Aljanaideh, M., Can fuzzy logic predict con-
sumer ethnocentric tendencies? An empirical analysis in Jordan. J. Phys. Sci.
Appl., 1, 5, 100–106, 2011.
4. Al Ganideh, S.F. and Aljanaideh, M., Using fuzzy logic to analyze marketing
data: The impact of socio-psychological variables on the national identity of
Jordanians. Transnatl. Corp. Rev., 5, 2, 104–114, 2013, http://dx.doi.org/10.1
080/19186444.2013.11668682.
5. Andrew, L., Fuzzy logic – A call for more intuitive decision making, 2020,
https://www.andrewperry.com/.
6. Bandaru, S., Gaur, A., Deb, K., Khare, V., Chougule, R., Bandyopadhyay, P.,
Development, analysis and applications of a quantitative methodology for
assessing customer satisfaction using evolutionary optimization. Appl. Soft
Comput. J., 30, 265–278, 2015.
7. Bashiri, A., Mirhosseini-Alizamini, S.M., Mehdi Nasrabadi, M., Application
of fuzzy logic for advertising marketing campaigns. Control Optim. Appl.
Math. (COAM), 5, 2, (25–37), 2021.
8. Bezděk, V., Using fuzzy logic in business. Proc. – Soc. Behav. Sci., SIM 2013/
12th Int. Symposium Management., 124, 371–380, 2014, 1877-0428.
9. Bravo Hidalgo, D., Climatización solar de edificaciones. Cent. Azúcar, 42,
72–82, 2015.
10. Bravo Hidalgo, D. and León González, J.L., Divulgación de la investigación
científica en el Siglo XXI. Rev. Universidad y Sociedad, 10, 88–97, 2018.
11. Davila, T., An empirical study on the drivers of management control systems’
design in new product development. Account. Organ. Soc., 25, 383–409,
2000.
12. Dairo, A. and Szucs, K., Fuzzy expert pricing systems and optimization
techniques in marketing science, in: Fuzzy Expert Pricing Systems and
Optimization Techniques, Fuzzy Systems and Data Mining, VI, pp. 255–260,
2020.
13. Dostal, P. and Lin, C., Business applications of fuzzy logic, in: The Oxford
Handbook of Computational Economics and Finance, Chen, S.-H., Kaboudan,
M., Du, Y.-R. (Eds.), 2018.
14. Enache, I.C., Fuzzy logic marketing models for sustainable development.
Bull. Transilv. Univ. Bras., Ser. V: Econ. Sci., 8, 57, 267–274, 2015.
15. Guo, L.-X., Lin, C.-C., Huang, P.-F., Jhou, S.-Y., Chen, S.-C., Tsai, F.-S., Fuzzy
logic analysis for key factors for customer loyalty in e-shopping environ-
ment. Front. Psychol., 12, 742699, 2021.
16. Halabi, D. and Shaout, A., Website tenders evaluation using fuzzy logic.
Int. J. Inf. Syst. Serv. Sect., 11, 32–50, 2019, https://doi.org/10.4018/
IJISSS.2019040103.
17. Hidalgo, D.B. and Guerra, Y.P., Eficiencia energética en la climatización de
edificaciones. Rev. Publ., 3, 218–238, 2016.
18. Hernández, A.B. and Hidalgo, D.B., Fuzzy logic in business, management
and accounting. Open J. Bus. Manage., 8, 2524–2544, 2020, https://doi.
org/10.4236/ojbm.2020.86157.
19. Krishna, P.R. and De, S.K., A fuzzy approach to build an intelligent data
warehouse. J. Intell. Fuzzy Syst., 11, 23–32, 2001.
20. Kunsch, P.L. and Vander Straeten, M., The cost of a nuclear-fuel repository:
A criterion valuation by means of fuzzy logic, in: Evaluation and Decision
Models with Multiple Criteria: Case Studies, p. 311, Springer, Berlin, 2015,
https://doi.org/10.1007/978-3-662-46816-6_11.
21. Meier, A. and Werro, N., A fuzzy classification model for online customers.
Informatica, 31, 175–182, 2007.
22. Ouahli, J. and Cherkaoui, A., Team performance assessment within fuzzy
logic, 2019.
23. Phau, I. and Chan, K.W., Targeting East Asian markets: A comparative study
on national identity. J. Targeting Meas. Anal. Mark., 12, 2, 157–172, 2003,
http://dx.doi.org/10.1057/palgrave.jt.5740105.
24. Rahmatin, N., Santoso, I., Indriani, C., Rahayu, S., Widyaningtyas, S.,
Integration of the fuzzy failure mode and effect analysis (Fuzzy FMEA) and
the Analytical Network Process (ANP) in marketing risk analysis and miti-
gation. Int. J. Technol., 4, 809–818, 2018.
25. Stakhanov, D.V., Tselykh, A.N., Tselykh, L.A., Sergeev, N.E., On the issue of
adaptation of expert systems for decision support of applied management
tasks, SFU Bulletin. Technical Sciences, Technological Institute of the Southern
Federal University in Taganrog, Taganrog, 9, 159, 2014.
26. Stakhanov, D., Grishchenko, O., Fedortsova, S., Making management deci-
sions in the business environment based on fuzzy logic methods. Proceedings
of the VIII International Scientific and Practical Conference ‘Current problems
of social and labour relations’ (ISPC-CPSLR 2020), Advances in Social Science,
Education and Humanities Research, vol. 527, pp. 647–652, 2021.
27. Thelen, S.T. and Honeycutt, Jr. E. D., Assessing identity in Russia between
generations using the national identity scale. J. Int. Mark., 12, 2, 58–81, 2004,
http://dx.doi.org/10.1509/jimk.12.2.58.32900.
28. Ulfah, M., Trenggonowati, D.L., Ekawati, R., Ramadhania, S., The proposed
improvements to minimize potential failures using lean six sigma and multi
attribute failure mode analysis approaches. IOP Publishing. Broad Exposure
to Science and Technology 2019 (BEST2019). IOP Conf. Ser.: Mater. Sci. Eng.,
673, 012082, 2019.
29. Vom Berg, B.W., Valdés, A.R., Memari, A., Barakat, N., Gómez, J.M.,
Customer segmentation based on compensatory fuzzy logic within a sus-
tainability CRM for intermodal mobility, in: Studies in Computational
Intelligence, pp. 415–429, 2014.
30. Xia, X., Wang, Z., Gao, Y., Estimation of nonstatistical uncertainty using
fuzzy-set theory. Meas. Sci. Technol., 11, 4, 430–435, 2000.
31. Yeh, C.H. and Kuo, Y.-L., Evaluating passenger services of Asia-Pacific inter-
national airports. Transp. Res. E, 39, 1, 35–48, 2003.
32. Zadeh, L.A., Fuzzy sets. Inform. Control, 8, 338–353, 1965.
33. Liu, C. H., A group decision-making method with fuzzy set theory and
genetic algorithms in quality function deployment. Qual. Quant. 44, 1175–
1189, 2010.
5
A Method for Ranking Fuzzy Numbers
Based on Their Value, Ambiguity,
Fuzziness, and Vagueness
Sunayana Saikia* and Rituparna Chutia
Department of Mathematics, Cotton University, Guwahati, Assam, India
Abstract
Real-world issues are seen to involve uncertainty quite frequently. To prevent this
uncertainty, some parameters, like fuzzy numbers, can be taken into account. As
a result, ranking fuzzy numbers is crucial to bringing clarity to ambiguous situa-
tions in real life. We developed a novel fuzzy number ranking algorithm based on
value, ambiguity, fuzziness, and vagueness in this study. It is discovered that the
suggested ranking method has a high degree of discrimination and can rank fuzzy
numbers that cannot be distinguished by other recent and established methods.
Additionally, there are some numerical examples that show strong discrimination
power in comparison to current approaches. The suggested method also seems to
outperform in every scenario. Furthermore, the proposed strategy is applied in
this work to a university/college’s assistant professor recruitment procedure for a
certain field.
Keywords: Trapezoidal fuzzy numbers (TrFNs), ranking trapezoidal fuzzy
numbers, value, ambiguity, fuzziness, vagueness
5.1 Introduction
Generally, generalised intervals are used to represent fuzzy numbers. As a
result, fuzzy numbers cannot be ordered in the same way that real num-
bers can. As a result, at various times, multiple approaches for ranking
fuzzy numbers have been offered. Furthermore, the ordering of ambiguous
51
numbers has been thoroughly investigated. There are several research arti-
cles in the literature that describe various ways for ranking fuzzy numbers.
To summarise, there is no commonly accepted mechanism for ranking
fuzzy numbers. Existing approaches have been shown to provide results
that are sometimes counter-intuitive. Furthermore, some of these tactics
are nondiscriminatory and contradictory.
Here, a few of the ranking systems are carefully investigated. In 1976,
“[32], [33] provided the first unconventional approaches to the require-
ment of ordering fuzzy quantities using a maximising set”. “A variety of
fuzzy number ranking algorithms are analysed and compared by [8]”.
With the help of maximising set, [27] created a well-known ranking
strategy. [45],[46] “also proposed some suitable conditions to establish
the rationality validation of a fuzzy number ranking approach”. [11]
adopted the maximisation and minimization set strategy, but this rank-
ing method has certain drawbacks. As a result, “[7] suggested a new way
to address the constraints of methods based on maximising and mini-
mising sets”. To solve these restrictions, “[17] proposed a new ranking
approach based on the area between the centroid point and the origin”.
“[7] discovered certain flaws in [17] technique and suggested a new
ranking method for generalised fuzzy numbers to address them”. When
ranking fuzzy numbers, [47] “discovered that [17] ranking method was
flawed in a variety of scenarios”. [47] also developed an improved rank-
ing algorithm. [48] presented a new ranking approach for generalised
fuzzy numbers to address the restrictions of the centroid point formula.
Furthermore, [34], [36], [16] “presented a ranking approach for fuzzy
quantities based on the left integral values through an index of opti-
mism, and [50] enhanced this method”. [13] “proposed a new technique
based on the areas of the positive side, the areas of the negative side, and
the heights of generalised fuzzy numbers”. Unfortunately, when there
is compensation of areas, [12] and [13] fail. [14] “introduced a rank-
ing algorithm for generalised fuzzy numbers with varied left and right
heights”. [44] “introduced a new idea for ranking fuzzy numbers by con-
sidering the area at different decision levels”. Under some circumstances,
however, [14], [39], and [44] fail to identify trapezoidal and triangular
fuzzy numbers. [41] “developed a new ranking algorithm for exponen-
tial trapezoidal fuzzy numbers based on variance and also employing
the mellin transform of the fuzzy numbers”. [18] “suggested a modified
epsilon-deviation degree approach based on the ill-defined magnitude
‘value’ and the ambiguity of the fuzzy set”. [20] “also developed a new
notion of parametric form of fuzzy numbers with defuzzifiers at differ-
ent heights, as well as the concept of decision levels”. [10] “developed
A Method for Ranking Fuzzy Numbers 53
a ranking method for non-normal p-norm trapezoidal fuzzy numbers

based on an integral value approach”. [35] “adapted [10]’s technique for
ranking non-normal p-norm trapezoidal fuzzy numbers in their paper”.
In addition, [22] rank p-norm generalised fuzzy numbers.
[23] “proposed a new technique for ranking trapezoidal fuzzy numbers
at the decision-making level based on their value and ambiguity”. Using the
notion of fuzzy distance, [3] suggested a novel ranking approach. In addi-
tion, [4] “suggested a new ranking algorithm based on fuzzy numbers’
centroid points and maximum crisp value”. [29] “introduced a novel rank-
ing approach that took the core, margin, and α-cuts into account for each
of the fuzzy values”. [31] “also introduced a novel ranking algorithm based
on parameterized defuzzification of a fuzzy integer”. [1] “suggested a
new method for ranking trapezoidal fuzzy numbers, but it has significant
flaws”. [28] “suggested a new ranking approach to address the shortcom-
ings of [1]’s method”. Furthermore, [28]’s ranking algorithm has several
shortcomings. As a result, [30] “introduced the ranking approach
to solve the shortcomings of [28]’s ranking method”. For comparing
triangular intuitionstic fuzzy numbers, [40] “suggested a novel ranking
approach based on value index and ambiguity index”. On the concept of
α-cut, β-cuts, and area on the left side of intuitionstic fuzzy numbers, “[25]
suggested a novel parametric ranking approach for intuitionstic fuzzy
numbers”. [43] “introduced a novel parametric ranking approach based
on α-cuts and hesitation degrees, which are two important ideas in
intuitionstic fuzzy sets”. “[24] devised a new approach for ranking interval
type-2 fuzzy integers based on ambiguity and value index”. “[19] proposed
a novel approach of ranking Z numbers at decision-making stages based
on value index and ambiguity”. “[21] proposed a new method to the value,
ambiguity, and θ function using a single-valued neutrosophic number”.
[37], [5], and [38] are some recent works.
Numerous modern methods for ranking fuzzy numbers typically fall
short of putting them in the proper order. Additionally, it has been demon-
strated that some current techniques for ranking fuzzy numbers are unreli-
able in the following situations: (1) when the fuzzy numbers have the same
support but different cores, (2) when they are crisp-valued but have differ-
ent heights, (3) when the expectation values of the centroid points for the
fuzzy numbers to be compared are the same, and (4) when the fuzzy num-
bers are used to represent compensation of areas. As explained in the next
section, even a recent technique created by [13] fails to offer effective dif-
ferentiating in certain fuzzy numbers. To overcome these restrictions and
shortcomings, new techniques are required. The purpose of this research
is to try to overcome the limits and shortcomings of present approaches.
[20] and [18] for their discovery that the fuzzy quantities “value” and “ambi-
guity” have a substantial impact on ordering fuzzy numbers. Furthermore,
[2] discovered that the quantities “fuzziness” and “vagueness” of fuzzy
numbers have a substantial influence on fuzzy number ranking. As a result,
fuzzy numbers are ordered based on these quantities. This paper presents a
novel defuzzification procedure that ranks fuzzy numbers according to the
notions of “value,” “ambiguity,” “fuzziness,” and “vagueness.” Furthermore,
the inclusion-exclusion functions θ1 and θ2 combine value, ambigu-
ity, fuzziness, and vagueness. This aspect of the suggested methodology
demonstrates a high level of discrimination and may effectively rank fuzzy
numbers that are difficult to separate using other methods.
The ensuing sections make up the remaining portions of this paper. The
content is introduced in Section 5.2 with a many delineations of fuzzy fig-
ures and other terms that are applicable. The ranking approach suggested
in section 5.3 makes advantage of ill-defined amounts values, inscruta-
bility, fuzziness and vagueness of the fuzzy figures. Also some important
theorems has been proved. The suggested system’s rationality confirma-
tion has been demonstrated in section 5.4. Numerical exemplifications
are handed in Section 5.5, pressing the suggested system’s bettered perfor-
mance. Also, using the suggested system, section 5.6 has carried out a fuzzy
threat analysis for fiscal investment. The findings and important factors of
the suggested approach are stressed in section 5.7.
5.2 Preliminaries
5.2.1 Definitions and Concepts
A quick overview of some fundamental terms and ideas connected to the
current discussion is provided in this section.
Definition 5.2.1. [44] “A fuzzy number µ = (µ1, µ2, µ3, µ4) is a fuzzy subset
of the real line R such that fµ : R −→ [0, 1] which satisfies the following
properties:
(1) f µ is upper semi-continuous,

(2) fµ = 0 outside some interval [µ1, µ4], where µ1, µ2 ∈ R,
(3) There are real numbers µ2, µ3 such that µ1 ≤ µ2 ≤ µ3 ≤ µ4,
(4) fµ(x) is monotonic increasing on [µ1, µ2],
(5) fµ(x) is monotonic decreasing on [µ3, µ4],
(6) fµ(x) = 1, µ2 ≤ x ≤ µ3.”
Definition 5.2.2. [51] “A fuzzy number µ = (µ1, µ2, µ3, µ4) is described as
any fuzzy subset of the real line R with membership function defined as
(k 1 , k 2 , k 3 , k 4 ) k 0
k (5.1)
(k 4 , k 3 , k 2 , k 1 ) k 0
where f µL : [ µ1 , µ 2 ] − → [ 0,1] is strictly increasing and continuous function

and f µR : [ µ3 , µ 4 ] − → [ 0,1] is strictly decreasing and continuous func-
tion. The inverse functions of f µL and f µR are denoted and defined as
g µL : [ 0,1] − → [ µ1 , µ 2 ] and g µR : [ 0,1] − → [ µ3 , µ 4 ] respectively.”
Definition 5.2.3. [51] “A trapezoidal fuzzy number µ = (µ1, µ2, µ3, µ4) is
described as any fuzzy subset of the real line R with membership function
defined as
x 1
, if 1 x 2 ,
2 1
1, if 2 x 3 ,
f x (5.2)
4 x
, if 3 x 4 ,
4 3
0, otherwise,
If followed by the condition µ2 = µ3, then it is called a normal triangular

fuzzy number. For the fuzzy number µ, the inverse functions of f µL ( x ) and
f µR ( x ) are defined as
g µL ( y ) = µ1 + y( µ 2 − µ1 ) (5.3)
and
g µR ( y ) = µ 4 − y( µ 4 − µ3 ) (5.4)
respectively.”
Definition 5.2.4. [49] “Assume that µ = (µ1, µ2, µ3, µ4) and ν = (ν1, ν2, ν3, ν4)
are two TrFNs where µi, νi, i = 1, 2, 3, 4 are real values. The arithmetic oper-
ation for these TrFNs are defined as follows:
(1) Addition: µ ⊕ ν = (µ1, µ2, µ3, µ4) ⊕ (ν1, ν2, ν3, ν4) = (µ1 + ν1,
µ2 + ν2, µ3 + ν3, µ4 + ν4),
(2) Subtraction: µ 0 ν = (µ1, µ2, µ3, µ4) 0 (ν1, ν2, ν3, ν4) = (µ1 − ν4,
µ2 − ν3, µ3 − ν2, µ4 − ν1),
(3) Multiplication: µ ⊗ ν = (µ1, µ2, µ3, µ4) ⊗ (ν1, ν2, ν3, ν4) =
(µ1 × ν1, µ2 × ν2, µ3 × ν3, µ4 × ν4),
(4) Division: µ 0 ν = (µ1, µ2, µ3, µ4) 0 (ν1, ν2, ν3, ν4) = (µ1 ÷ ν4,
µ2 ÷ ν3, µ3 ÷ ν2, µ4 ÷ ν1),
(5) Scalar multiplication:
(k 1 , k 2 , k 3 , k 4 ) k 0
k
(k 4 , k 3 , k 2 , k 1 ) k 0.
where µi for i = 1, 2, 3, 4 are any real numbers and ν1, ν2, ν3, ν4
are any non zero positive real numbers.”
5.3 The Designed Method

The proposed method is thoroughly explained in this section. With the
help of these four parameters: value, ambiguity, fuzziness, and vagueness, a
new ranking system has been developed. We rank the fuzzy numbers first
using the parameter ‘value (V)’, where V can be thought of as a core value
that indicates the value of a fuzzy number. When the values of the fuzzy
numbers are equivalent, the parameter ‘ambiguity (A)’ is used to rank the
fuzzy numbers, with A being the global spread of the fuzzy numbers’ mem-
bership functions. When the ambiguities of the fuzzy numbers are equal,
rank the fuzzy numbers using the parameter ‘fuzziness (F)’, where F is a
global difference between the fuzzy numbers and their complement. When
the fuzziness of the fuzzy numbers is equal, rank the fuzzy numbers using
the parameter ‘vagueness (T)’, where T is the measure of the fuzzy num-
bers’ vagueness.
Let the following membership function for the arbitrary fuzzy number
µ = (µ1, µ2, µ3, µ4), which is defined on the real line R.
f L (x ), if 1 x 2,
1, if 2 x 3,
f (x ) R
(5.5)
f (x ), if 3 x 4 ,
0, otherwise,
and let the inverse functions of f µL and f µR be g µL : [ 0,1] − → [ µ1 , µ 2 ] and

g µR : [ 0,1] − → [ µ3 , µ 4 ] , respectively. Let the reducing function be s: [0,1] −
→ [0,1], then
(1) The following is a definition of the value of the fuzzy number

µ with regard to s:
V(µ ) = ∫ 0 s( y )( g µL ( y ) + g µR ( y )) dy
1
(5.6)
(2) The following is a definition of the ambiguity of the fuzzy

number µ with regard to s:
1
A( ) 0 s( y ) g R ( y ) g L ( y ) dy (5.7)
(3) The fuzziness in µ is defined as
1
1
F( ) 2
0 s( y ) g R ( y ) g L ( y ) dy 1 s( y ) g L ( y ) g R ( y ) dy (5.8)
2
(4) The vagueness in µ is defined as
1
1
T( ) 2
0 s( y ) g L ( y ) g R ( y ) dy 1 s( y ) g L ( y ) g R ( y ) dy (5.9)
2
In particular, for a TrFN µ = (µ1, µ2, µ3, µ4) with the member-
ship function defined in Eq. 5.5 with the inverse functions defined as
g µL ( y ) = µ1 + y( µ 2 − µ1 ) and g R ( y ) 4 y( 4 3 ) and the reducing

function s(y) = y for [0, 1], then value V(µ) is given by
1 1
V( ) 0 s( y ) g L ( y ) g R ( y ) dy ( 1 2 2 2 3 4 ) (5.10)
6
and ambiguity A(µ) is given by
1 1
A( ) 0 s( y ) g R ( y ) g L ( y ) dy ( 4 2 3 2 2 1 ) (5.11)
6
similarly, fuzziness F(µ) is given by
1
1 1
F( ) 2
0 s( y ) g R ( y ) g L ( y ) dy 1 s( y ) g L ( y ) g R ( y ) dy ( 4 3 2 1 )
2 4
(5.12)
similarly, vagueness T(µ) is given by
1
1
T( ) 2
0 s( y ) g L ( y ) g R ( y ) dy 1 s( y ) g L ( y ) g R ( y ) dy ( 1 2 3 4 )
2
(5.13)
Let µ, ν ∈ X represent two arbitrary fuzzy numbers, where µ = (µ1, µ2,

µ3, µ4) and ν = (ν1, ν2, ν3, ν4) respectively. The average α level at zero is cal-
1 1
culated as t0 = ( µ1 + µ 4 ) and t0 = ( µ1 + µ 4 ), respectively. The ranking
2 2
of TrFNs may be done using the definition described below based on the
aforementioned quantities.
R( µ ,θ1 ,θ 2 ) = V ( µ ) + θ1 { A( µ ) + F ( µ )} + θ 2T ( µ ) (5.14)
1 2 0, if V( ) V( ),
1 1, 2 0, if V( ) V( ) and A( ) A( ) or F( ) F( ) and t 0 0,
where
1 1, 2 0, if V( ) V( ) and A( ) A( ) or F( ) F( ) and t 0 0,
1 0, 2 1 if V( ) V( ), A( ) A( ) and F( ) F( ).
Consider the fuzzy numbers µ and ν, then the following decisions can
be made
• If R(µ, θ1, θ2) > R(ν, θ1, θ2), then µ > ν,

• If R(µ, θ1, θ2) < R(ν, θ1, θ2), then µ ≺ ν,
• If R(µ, θ1, θ2) = R(ν, θ1, θ2), then µ ∼ ν.
For the ranking index R, the ordering of the fuzzy numbers µ and ν is
µ ≾ ν if and only if µ ≺ ν or µ ∼ ν, and the ordering µ ≿ ν follows if and
only if µ > ν or µ ∼ ν.
Following, a few significant and crucial theorems are addressed, which
will assist to further develop a few characteristics of the recommended
technique.
Theorem 5.3.1. If there are two TrFNs µ = (µ1, µ2, µ3, µ4) and ν = (ν1, ν2, ν3,
ν4) in R, then V(µ + ν) = V(µ) + V(ν).
Proof. Let µ = (µ1, µ2, µ3, µ4) and ν = (ν1, ν2, ν3, ν4) be two TrFNs, then by
arithmetic of TrFNs µ + ν = (µ1 + ν1, µ2 + ν2, µ3 + ν3, µ4 + ν4). As such,
1
V( ) 0 s( y ) g L ( y ) g R ( y ) dy
Consequently, it follows from Remark 5.3.1
1
V( ) 0 s( y ) g L ( y ) g L ( y ) g R ( y ) g R ( y ) dy
1 1
0 s( y ) g L ( y ) g R ( y ) dy 0 s( y ) g L ( y ) g R ( y ) dy
As a result, the outcome V(µ + ν) is equal to V(µ) + V(ν).
Remark 5.3.1. Let us consider two arbitrary fuzzy numbers be µ and ν and
the inverse functions of f µL ( x ), f µR ( x ), fνL ( x ) and fνR ( x ) are denoted as
g µL ( y ), g µR ( y ), g νL ( y ) and g νR ( y ) respectively. Then
g L ( y) g L ( y ) g L ( y ) and g L ( y ) g L ( y ) g R ( y ),
g R ( y) g R ( y ) g R ( y ) and g R ( y ) g R ( y ) g L ( y ).
Theorem 5.3.2. If there are two TrFNs µ = (µ1, µ2, µ3, µ4), ν = (ν1, ν2, ν3, ν4)
in R. Then A(µ +ν) = A(µ) + A(ν).
Proof. Let µ = (µ1, µ2, µ3, µ4) and ν = (ν1, ν2, ν3, ν4) be two TrFNs, then by the
1
A( ) 0 s( y ) g R ( y ) g L ( y ) dy

1
A( ) 0 s( y ) g R ( y ) g R ( y ) g L ( y ) g L ( y ) dy
1 1
0 s( y ) g R ( y ) g L ( y ) dy 0 s( y ) g L ( y ) g L ( y ) dy
As a result, the outcome A(µ + ν) = A(µ) + A(ν).
Theorem 5.3.3. If µ = (µ1, µ2, µ3, µ4), ν = (ν1, ν2, ν3, ν4) be two TrFNs, then
F(µ + ν) = F(µ) + F(ν).
Proof. Let µ = (µ1, µ2, µ3, µ4) and ν = (ν1, ν2, ν3, ν4) be two TrFNs, then by the
1
1
F( ) 2
0 s(y ) g R (y ) g L (y ) dy 1 s(y ) g L (y ) g R (y ) dy
2

1 1
F( ) 2
0 s( y ) g R ( y ) g R ( y ) g L ( y ) g L ( y ) dy 1 s( y ) g L ( y ) g L ( y ) g R ( y ) g R ( y ) dy
2
1 1 1
( 2
0 s( y ) g R ( y ) g L ( y ) dy 1 s( y ) g L ( y ) g R ( y ) dy ) ( 2
0 s( y ) g R ( y ) g L ( y ) dy
2
1
1 s( y ) g L ( y ) g R ( y ) dy )
2
As a result, the outcome F(µ + ν) is equal to F(µ) + F(ν).
Theorem 5.3.4. If there are two TrFNs µ = (µ1, µ2, µ3, µ4), ν = (ν1, ν2, ν3, ν4)
in R, then T(µ + ν) = T(µ) + T(ν).
Proof. If µ = (µ1, µ2, µ3, µ4) and ν = (ν1, ν2, ν3, ν4) be two TrFNs, then by the
arithmetic of IT2FNs µ + ν = (µ1 + ν1, µ2 + ν2, µ3 + ν3, µ4 + ν4). As such,
1
1
T( ) 2
0 s(y ) g L (y ) g R (y ) dy 1 s(y ) g L (y ) g R (y ) dy
2

1 1
T( ) 2
0 s( y ) g L ( y ) g L ( y ) g R ( y ) g R ( y ) dy 1 s( y ) g L ( y ) g L ( y ) g R ( y ) g R ( y ) dy
2
1 1 1
( 2
0 s( y ) g L ( y ) g R ( y ) dy 1 s( y ) g L ( y ) g R ( y ) dy ) ( 2
0 s( y ) g L ( y ) g R ( y ) dy
2
1
1 s( y ) g L ( y ) g R ( y ) dy )
2
As a result, the outcome T(µ + ν) = T(µ) + T(ν).
Remark 5.3.2. Let µ = (µ1, µ2, µ3, µ4) be an TrFN and g µL ( y ) and g µR ( y )
are the inverse functions of f µL ( y ) and f µR ( y ) respectively. Let k be a real
number, then
g kLµ ( y ) = kg µL ( y ) and g kRµ ( y ) = kg µR ( y ) if k > 0,

g kRµ ( y ) = kg µR ( y ) and g kRµ ( y ) = kg µL ( y ) if k < 0.
Theorem 5.3.5. Let µ = (µ1, µ2, µ3, µ4) be an TrFN. Then V(kµ) = kV(µ),
A(kµ) = kA(µ), F(kµ) = kF(µ) and T(kµ) = kT(µ) for k > 0 and V(kµ) =
kV(µ), A(kµ) = −kA(µ), F(kµ) = −kF(µ) and T(kµ) = kT(µ) for k < 0.
Proof. Let µ = (µ1, µ2, µ3, µ4) be an TrFN and k ∈ R, then by the arithmetic
of TrFNs,
(k 1 , k 2 , k 3 , k 4 ) k 0
k
(k 4 , k 3 , k 2 , k 1 ) k 0
Let µ = (µ1, µ2, µ3, µ4) be an TrFN and k > 0, then by the arithmetic of
TrFNs, kµ = (kµ1, kµ2, kµ3, kµ4).
Hence using the Eq. 5.10, it follows that
1
V(k ) 0 s( y ) g kL ( y ) g kR ( y ) dy
1
V(k ) 0 s( y ) kg L ( y ) kg R ( y ) dy
1
k ( 0 s( y ) g L ( y ) g R ( y ) d y )
kV( )
Similarly using the Eq. 5.11, it follows that
1
A(k ) 0 s( y ) g kR ( y ) g kL ( y ) dy
Then by Remark 5.3.2, get that
1
A(k ) 0 s( y ) kg R ( y ) kg L ( y ) dy
1
k ( 0 s( y ) g R ( y ) g L ( y ) d y )
kA( )
1
1
F(k ) 2
0 s( y ) g kR ( y ) g kL ( y ) dy 1 s( y ) g kL ( y ) g kR ( y ) dy
2
1
1
F(k ) 2
0 s( y ) kg R ( y ) kg L ( y ) dy 1 s( y ) kg L ( y ) kg R ( y ) dy
2
1
1
F(k ) k( s( y ) g R ( y ) g L ( y ) dy
2
0 1 s( y ) g L ( y ) g R ( y ) dy )
2
F(k ) kF( )
Again using the Eq. 5.13, it follows that
1
1
T(k ) 2
0 s( y ) g kL ( y ) g kR ( y ) dy 1 s( y ) g kL ( y ) g kR ( y ) dy
2
1
1
T(k ) 2
0 s( y ) kg L ( y ) kg R ( y ) dy 1 s( y ) kg L ( y ) kg R ( y ) dy
2
1
1
T(k ) k( s( y ) g L ( y ) g R ( y ) dy
2
0 1 s( y ) g L ( y ) g R ( y ) dy )
2
T(k ) kT( )
Again, let µ = (µ1, µ2, µ3, µ4) be an TrFN and k < 0, then by the arithmetic
of TrFNs, kµ = (kµ4, kµ3, kµ2, kµ1).
Now using the Eq. 5.10, it follows that
1
V(k ) 0 s( y ) g kL ( y ) g kR ( y ) dy

1
V(k ) 0 s( y ) kg R ( y ) kg L ( y ) dy
1
k ( 0 s( y ) g L ( y ) g R ( y ) d y )
kV( )

1
A(k ) 0 s( y ) g kR ( y ) g kL ( y ) dy
1
A(k ) 0 s( y ) kg L ( y ) kg R ( y ) dy
1
k ( 0 s( y ) g L ( y ) g R ( y ) d y )
kA( )
1
1
F(k ) 2
0 s( y ) g kR ( y ) g kL ( y ) dy 1 s( y ) g kL ( y ) g kR ( y ) dy
2
1
1
F(k ) 2
0 s( y ) kg L ( y ) kg R ( y ) dy 1 s( y ) kg R ( y ) kg L ( y ) dy
2
1
1
F(k ) k( 02 s( y ) g R ( y ) g L ( y ) dy 1 s( y ) g L ( y ) g R ( y ) dy )
2
F(k ) kF( )
Also using the Eq. 5.13, it follows that
1
1
T(k ) 2
0 s( y ) g kL ( y ) g kR ( y ) dy 1 s( y ) g kL ( y ) g kR ( y ) dy
2
1
1
T(k ) 2
0 s( y ) kg R ( y ) kg L ( y ) dy 1 s( y ) kg R ( y ) kg L ( y ) dy
2
1
1
T(k ) k( 02 s( y ) g L ( y ) g R ( y ) dy 1 s( y ) g L ( y ) g R ( y ) dy )
2
T(k ) kT( )
Theorem 5.3.6. The relations ≻ and ∼ will meet the following axioms for
order relations if µ, ν, ρ ∈ X are three arbitrary fuzzy numbers:
(1) Reflexivity: µ ≿ ν,
(2) Antisymmetricity: if µ ≿ ν and ν ≿ µ both are true if and
only if µ ∼ ν,
(3) Laws of trichotomy: µ ≻ ν or ν ≿ µ
(4) µ = ν if and only if µ ∼ ν.
Proof.
(1) It can be proof trivially.

(2) Let, µ ≿ ν, then R(µ, θ1, θ2) ≥ R(ν, θ1, θ2); also let ν ≿ µ, then
R(ν, θ1, θ2) ≥ R(µ, θ1, θ2). Hence, R(µ, θ1, θ2) = R(ν, θ1, θ2) leads
to µ ∼ ν. As a result, claim (2) is proved.
(3) Since the order relations > and = of real numbers serve as
the foundation for the order relations ≻ and ∼ of fuzzy num-
bers, the assertion (3) was simple to make.
(4) If µ ∼ ν, then R(µ, θ1, θ2) = R(ν, θ1, θ2). As a result, the asser-
tion (4) is proven.
Theorem 5.3.7. If µ ≿ ν and ν ≿ ρ, then µ ≿ ρ when θ1 and θ2 are same for

both the inequalities µ ≿ ν and ν ≿ ρ.
Proof. Let, µ ≿ ν and ν ≿ ρ occur for R(µ, 0, 0) ≥ R(ν, 0, 0) and R(ν, 0, 0)

≥ R(ρ, 0, 0) respectively. Following that, V(µ) ≠ V(ν) and V(ν) ≠ V(ρ) are
obtained, resulting in V(µ) > V(ν) and V(ν) > V(ρ). As a result, V(µ) >
V(ρ). The result is that R(µ, 0, 0) > R(ρ, 0, 0). The resulting value of µ ≿ ρ
was obtained.
Again, assume that µ ≿ ν and ν ≿ ρ for R(µ, ±1, 0) ≥ R(ν, ±1, 0) and
R(ν, ±1, 0) ≥ R(ρ, ±1, 0), respectively. As a result, V(µ) = V(ν) and V(ν) =
V(ρ).
Then for θ1 = ±1 and θ2 = 0,
[A( ) F( )] [A( ) F( )] (5.15)
[A( ) F( )] [A( ) F( )] (5.16)
Now from Eqs. 5.15, 5.16 get that, ±[A(µ) + F(µ)] > ±[A(ρ) + F(ρ)]. This
results in the inequity V(µ) ± [A(µ) + F(µ)] > V(ρ) ± [A(ρ) + F(ρ)]. Thus,
R(µ, ±1, 0) > R(ρ, ±1, 0). The outcome is µ ≻ ρ.
Suppose that, µ ≿ ν and ν ≿ ρ happen for R(µ, 0, 1) ≥ R(ν, 0, 1) and
R(ν, 0, 1) ≥ R(ρ, 0, 1). Following that, V(µ) = V(ν) and V(ν) = V(ρ).
Then for θ1 = 0 and θ2 = 1,
T( µ ) ≥ T(ν ) (5.17)
T( ) T( ) (5.18)
Now from Eqs. 5.21, 5.22 get that, T(µ) ≥ T(ρ). As a result, there is ineq-
uity V(µ) + T(µ) ≥ V(ρ) + T(ρ). Thus, R(µ, 0, 1) ≥ R(ρ, 0, 1). Thus, the
outcome is µ ≿ ρ.
Theorem 5.3.8. If µ ≻ ν and ν ≻ ρ, then µ ≻ ρ when θ1 and θ2 are same for

both the inequalities µ ≻ ν and ν ≻ ρ.
Proof. Let, µ ≻ ν and ν ≻ ρ occur for R(µ, 0, 0) > R(ν, 0, 0) and R(ν, 0, 0) >
R(ρ, 0, 0). Then, V(µ) ≠ V(ν) and V(ν) ≠ V(ρ), resulting in V(µ) > V(ν) and
V(ν) > V(ρ). As a result, V(µ) > V(ρ). Eventually, R(µ, 0, 0) > R(ρ, 0, 0). As
a consequence, the result is µ ≻ ρ.
Again let, µ ≻ ν and ν ≻ ρ occur for R(µ, ±1, 0) > R(ν, ±1, 0) and R(ν, ±1,
0) > R(ρ, ±1, 0). Then, V(µ) = V(ν) and V(ν) = V(ρ).
[A( ) F( )] [A( ) F( )] (5.19)
[A( ) F( )] [A( ) F( )] (5.20)
Now, take it from Eqs. 5.19, 5.20 get that, ±[A(µ) + F(µ)] > ±[A(ρ) +
F(ρ)]. As a result, there is inequity V(µ) ± [A(µ) + F(µ)] > V(ρ) ± [A(ρ) +
F(ρ)]. Therefore, R(µ, ±1, 0) > R(ρ, ±1, 0). The outcome is µ ≻ ρ.
Let, µ ≻ ν and ν ≻ ρ occur for R(µ, 0, 1) > R(ν, 0, 1) and R(ν, 0, 1) >
R(ρ, 0, 1) respectively. Consequently, V(µ) = V(ν) and V(ν) = V(ρ).
T( µ ) > T(ν ) (5.21)
T( ) T( ) (5.22)
Now from Eqs. 5.21, 5.22 get that, T(µ) > T(ρ). This result in the ineq-
uity V(µ) + T(µ) > V(ρ) + T(ρ). Therefore, R(µ, 0, 1) > R(ρ, 0, 1). The out-
come is µ ≿ ρ.
Theorem 5.3.9. Given two arbitrary fuzzy numbers be µ, ν ∈ X, then

R(µ + ν, θ1, θ2) = R(µ, θ1, θ2) + R(ν, θ1, θ2), where θ1 and θ2 are invariant.
Proof. Given two arbitrary fuzzy numbers be µ = (µ1, µ2, µ3, µ4) and ν = (ν1, ν2,
ν3, ν4), then it is follows from the Theorems 5.3.1, 5.3.2, 5.3.3 and 5.3.4 get that,
V(µ + ν) = V(µ) + V(ν)
A(µ + ν) = A(µ) + A(ν)
F(µ + ν) = F(µ) + F(ν)
T(µ + ν) = T(µ) + T(ν).

Thus, the result follows as, R(µ + ν, θ1, θ2) is equal to V(µ + ν) +
θ1{A(µ + ν) + F(µ + ν)} + θ2T(µ + ν). This leads to V(µ) + V(ν) + θ1{A(µ) +
A(ν) + F(µ) + F(ν)} + θ2{T(µ) + T(ν)}. The following equation holds true:
V(µ) + θ1{A(µ) + F(µ)} + θ2T(µ) + V(ν) + θ1{A(ν) + F(ν)} + θ2T(ν). Hence,
the result is R(A, θ1, θ2) + R(B, θ1, θ2).
5.4 Validate the Reasonableness of the Suggested

Ranking Algorithm
This section illustrates and discusses the [45, 46]’s reasonable properties of
ordering fuzzy numbers to support the logic of the proposed technique. In
light of the fact that Q is a finite subset of X, let X and X1 be the sets of fuzzy
numbers that the suggested method can be applied to.
Proposition 5.4.1. If Q and µ in X are fuzzy numbers, then µ ≿ µ by V, A,

F and T on Q.
Proof. The proof is insignificant.
Proposition 5.4.2. If Q ∈ X and (µ, ν) ∈ Q2 be the order of two fuzzy num-

bers, where µ ≿ ν and ν ≿ µ by V, A, F and T on Q, then µ ∼ ν by V, A, F
and T on Q.
Proof. The proof simply follows from the Theorem 5.3.7.
Proposition 5.4.3. If Q ∈ X and (µ, ν, ρ) ∈ Q3, then µ ≿ ν and ν ≿ ρ by V,

A, F and T on Q, then µ ≿ ρ by V, A, F and T on Q.
Proof. The proof is directly related to the Theorem 5.3.7.
Proposition 5.4.4. If Q is contained in X and (µ, ν) is contained in X1.

Currently, if inf supp(µ) ≥ sup supp(ν), then µ ≻ ν is achieved by V, A, F,
and T on Q.
Proof. Let µ and ν be fuzzy numbers with non-crisp values. Inferentially,

V(µ) ∈ supp(µ) and V(ν) ∈ supp(ν). Given that inf supp(µ) ≥ sup supp(ν),
V(µ) > V(ν). Thus, µ ≻ ν by V on Q.
Proposition 5.4.5. If X and X/ are the two sets, which contain the fuzzy
numbers where V, A, F, and T may be used, and (µ, ν) ∈ (X ∪ X1)2, then
µ ≻ ν by V, A, F and T on X if and only if µ ≻ ν by V, A, F on X1.
Proof. The quantities V, A, F, and T determine the ranking order of µ and ν

and are independent of X and X1. Thus if µ ≻ ν by V, A, F and T on X, then
µ ≻ ν by V, A, F and T on X1.
Proposition 5.4.6. Let µ, ν, µ + ρ, and ν + ρ be components of X such that

µ, ν, and ρ heights are equal. If µ ≿ ν by R on {µ, ν}, which follows that µ +
ρ ≿ ν + ρ by R on {µ + ρ, ν + ρ}.
Proof. Claim: θ1 and θ2 have the same value when ordering µ and ν, and
they are invariant when ordering µ + ρ and ν + ρ. The proof of the claim
as: Let θ1 = 0 and θ2 = 0 in ordering µ and ν. Due to this, V(µ) ≠ V(ν), and
trivially V(µ + ρ) ≠ V(ν + ρ). Consequently, θ1 = 0 and θ2 = 0 in ordering
µ + ρ and ν + ρ. Let θ1 = ±1 and θ2 = 0 in ordering µ and ν. In light of the fact
that, V(µ) = V(ν), it follows that V(µ + ρ) = V(ν + ρ). Hence, θ1 = ±1 and
θ2 = 0 in ordering µ + ρ and ν + ρ. Again, let θ1 = 0 and θ2 = 1 in ordering
µ and ν. Hence, V(µ) = V(ν), A(µ) = A(ν) and F(µ) = F(ν), then trivially
V(µ + ρ) = V(ν + ρ), A(µ + ρ) = A(ν + ρ) and F(µ + ρ) = F(ν + ρ). As a result,
in the ordering of µ + ρ and ν + ρ, θ1 = 0 and θ2 = 1. Hence, the assertion.
Given that µ ≿ ν, there are three possible outcomes: i) R(µ, 0, 0) ≥
R(ν, 0, 0), ii) R(µ, ±1, 0) ≥ R(ν, ±1, 0) and iii) R(µ, 0, 1) ≥ R(ν, 0, 1).
Case 1: If R(µ, 0, 0) ≥ R(ν, 0, 0). Inequality R(µ, 0, 0) + R(ρ, 0, 0) ≥ R(ν, 0,

0) + R(ρ, 0, 0) results from this. Consequently, the inequality R(µ + ρ, 0, 0)
≥ R(ν + ρ, 0, 0) holds according to the Theorem 5.3.9. As a result, µ + ρ
≿ ν + ρ.
Case 2: if R(µ, ±1, 0) ≥ R(ν, ±1, 0). Inequality R(µ, ±1, 0) + R(ρ, ±1, 0) ≥
R(ν, ±1, 0) + R(ρ, ±1, 0) results from this. The inequality R(µ + ρ, ±1, 0) ≥
R(ν + ρ, ±1, 0) now holds according to Theorem 5.3.9. Consequently, the
outcome is µ + ρ ≿ ν + ρ.
Case 3: if R(µ, 0, 1) ≥ R(ν, 0, 1). As a result, R(µ, 0, 1) + R(ρ, 0, 1) ≥ R(ν, 0,

1) + R(ρ, 0, 1) is an inequality. Therefore, the inequality R(µ + ρ, 0, 1) ≥ R(ν
+ ρ, 0, 1) is true according to the Theorem 5.3.9. As a result, µ + ρ ≿ ν + ρ.
Proposition 5.4.7. Let µ, ν, µ + ρ, ν + ρ be elements of X such that heights

of µ, ν and ρ are equal. If µ ≻ ν by R on {µ, ν}, then µ + ρ ≻ ν + ρ via R on
the variables {µ + ρ, ν + ρ}.
Proof. The proof is a simple derivation from the proposition 5.4.6.
Proposition 5.4.8. Let µ, ν, µ + ρ, ν + ρ be components of X. If µ + ρ ≿ ν + ρ

by R on {µ + ρ, ν + ρ}, then µ ≿ ν on {µ, ν}.
Proof. If µ + ρ ≿ ν + ρ by R, then R(µ + ρ, θ1, θ2) ≥ R(ν + ρ, θ1, θ2). Theorem

5.3.7 states that R(µ, θ1, θ2) + R(ρ, θ1, θ2) ≥ R(ν, θ1, θ2) + R(ρ, θ1, θ2). Therefore,
R(µ, θ1, θ2) ≥ R(ν, θ1, θ2), which comes just after that µ ≿ ν.
Proposition 5.4.9. Let µ, ν, µ + ρ, ν + ρ be components of X. If µ + ρ ≻ ν +

ρ by R on {µ + ρ, ν + ρ}, which follows that µ ≻ ν on {µ, ν}.
Proof. If µ + ρ ≻ ν + ρ by R, then R(µ + ρ, θ1, θ2) > R(ν + ρ, θ1, θ2). Theorem
5.3.7 states that R(µ, θ1, θ2) + R(ρ, θ1, θ2) > R(ν, θ1, θ2) + R(ρ, θ1, θ2). Therefore,
R(µ, θ1, θ2) > R(ν, θ1, θ2), which comes just after that µ ≻ ν.
Proposition 5.4.10. For every arbitrary finite subset Q of X, (µ, ν) in Q2,

and k in R, where kµ, kν ∈ Q. If µ c: ν, which follows that kµ ≿ kν for k > 0
and kµ = kν for k < 0.
Proof. Let, µ ≿ ν, then R(µ, θ1, θ2). This results in V(µ) + θ1[A(µ) +F(µ)] +
θ2T(µ) ≥ V(ν) + θ1[A(ν) + F(ν)] + θ2T(ν). The fact that for k > 0, k{V(µ) +
θ1[A(µ) + F(µ)] + θ2T(µ)} ≥ k{V(ν) + θ1[A(ν) + F(ν)] + θ2T(ν)} is true at
this time. Now, using the Theorem 5.3.5, the inequality {V(kµ) + θ1[A(kµ)
+ F(kµ)] + θ2T(kµ)} ≥ {V(kν) + θ1[A(kν) + F(kν)] + θ2T(kν)} holds. The
inequality resulting from this is R(kµ, θ1, θ2) ≥ R(kν, θ1, θ2). As a result, the
outcome is kµ ≿ kν.
The following situations arise if we assume that k = −m < 0 for k < 0.
Case 1: Let µ ≿ ν and V(µ) ≠ V(ν), which follows θ1 = θ2 = 0. As a result,

V(−µ) ≠ V(−ν) and R(µ, 0, 0) ≥ R(ν, 0, 0) are equivalent. This results in the
inequality V(µ) ≥ V(ν). Thus, it is obvious that −mV(µ) ≤ −mV(ν). Now,
from the Theorem 5.3.5, the inequality V(−mµ) ≤ V(−mν) holds. Thus
R(−mµ, 0, 0) ≤ R(−mν, 0, 0). This leads to −mµ = −mν. Hence, the result.
Case 2: Let µ ≿ ν and V(µ) = V(ν) and t0 ≥ 0, which follows that θ1 = −1,
θ2 = 0. As a result, V(−mµ) = V(−mν) and then R(µ, −1, 0) ≥ R(ν, −1,
0). This leads to the inequality −[A(µ)+F(µ)] ≥ −[A(ν)+F(ν)]. Then
m[A(µ)+F(µ)] ≤ m[A(ν)+F(ν)]. Now, t0 ≥ 0, then −t0 ≤ 0 and θ1 = 1, θ2 =
0; hence, it follows that 5.3.5[A(−mµ)+F(−mµ)] ≤ [A(−mν)+F(−mν)].
This leads to the inequality that V(−mµ) + [A(−mµ) + F(−mµ)] ≤ V(−mν)
+ [A(−mν) + F(−mν)]. Then, it is true that R(−mµ, 1, 0) ≤ R(−mν, 1, 0).
This leads to −mµ = −mν. Hence the result.
Case 3: Let µ ≿ ν and V(µ) = V(ν) and t0 ≤ 0, which follows that θ1 = 1,

θ2 = 0. As a result, V(−mµ) = V(−mν) and then R(µ, 1, 0) ≥ R(ν, 1, 0). This
results in the inequality [A(µ) + F(µ)] ≥ [A(ν) + F(ν)]. Then −m[A(µ) +
F(µ)] ≤ −m[A(ν) + F(ν)]. Hence, it follows that 5.3.5 [A(−mµ) + F(−mµ)] ≤
[A(−mν) + F(−mν)]. This leads to the inequality that V(−mµ) + [A(−mµ)
+ F(−mµ)] ≤ V(−mν) + [A(−mν) + F(−mν)]. Then, it is true that R(−mµ, 1,
0) ≤ R(−mν, 1, 0). This leads to −mµ = −mν. Hence the result.
Case 4: Let µ ≿ ν and V(µ) = V(ν), A(µ) = A(ν) and F(µ) = F(ν), which fol-
lows that θ1 = 0, θ2 = 1. As a result, V(−µ) = V(−ν) and R(µ, 0, 1) ≥ R(ν, 0, 1).
This leads to the inequality T(µ) ≥ T(ν). Thus, it is obvious that −mT(µ) ≤
−mT(ν). Now from Theorem 5.3.5, the inequality T(−mµ) ≤ T(−mν) holds,
which leads to V(−mµ) + T(−mµ) ≤ V(−mν) + T(−mν). Thus, R(−mµ, 0, 1)
≤ R(−mν, 0, 1). This leads to −mµ = −mν.
Proposition 5.4.11. Assume that X is a universal set and that Q is any cho-
sen finite subset of X. Also {µ, ν} ∈ Q2 and k ∈ R, where kµ, kν ∈ Q. Then
kµ ≻ kν follows if µ ≻ ν, for k > 0.
Proof. The proof simply follows from the Proposition 5.4.10.
Proposition 5.4.12. Let µ, ν, µ − ρ, and ν − ρ be components of X. If µ ≿ ν

by R on {µ, ν}, which follows that µ − ρ ≿ ν − ρ by R on {µ − ρ, ν − ρ}.
Proof. In order to prove this fact, a claim on the invariance of θ1 and θ2 in

the orders µ, ν, and µ − ρ, ν − ρ must be stated. The following is the claim:
The value of θ1 and θ2 when ordering µ and ν is constant for ordering µ − ρ
and ν − ρ, respectively. The following information demonstrates the claim:
Let θ1 = 0 and θ2 = 0 in ordering µ and ν. As a result, V(µ) ≠ V(ν), then triv-

ially V(µ − ρ) ≠ V(ν − ρ). Consequently, θ1 = 0 and θ2 = 0 in ordering µ − ρ
and ν − ρ. In order µ and ν, let θ1 = ±1 and θ2 = 0. As a result, V(µ) = V(ν),
then trivially V(µ − ρ) = V(ν − ρ). Hence, θ1 = ±1 and θ2 = 0 in ordering
µ − ρ and ν − ρ. Again, let θ1 = 0 and θ2 = 1 in ordering µ and ν. Hence,
V(µ) = V(ν), A(µ) = A(ν) and F(µ) = F(ν), then trivially V(µ − ρ) = V(ν − ρ),
A(µ − ρ) = A(ν − ρ) and F(µ − ρ) = F(ν − ρ). As a result, when µ − ρ and
ν − ρ are ordered, θ1 = 0 and θ2 = 1. Therefore the claim.
In light of the fact that µ ≿ ν, three scenarios are possible: i) R(µ, 0, 0) ≥
R(ν, 0, 0), ii) R(µ, ±1, 0) ≥ R(ν, ±1, 0) and iii) R(µ, 0, 1) ≥ R(ν, 0, 1).
Case 1: If R(µ, 0, 0) ≥ R(ν, 0, 0). Inequality R(µ, 0, 0) + R(−ρ, 0, 0) ≥ R(ν, 0,

0) + R(−ρ, 0, 0) results from this. Consequently, the inequality R(µ − ρ, 0,
0) ≥ R(ν − ρ, 0, 0) is true according to the Theorem 5.3.7. As a result, µ − ρ
≿ ν − ρ.
Case 2: If R(µ, ±1, 0) ≥ R(ν, ±1, 0). Inequality R(µ, ±1, 0)+R(−ρ, ±1, 0) ≥
R(ν, ±1, 0)+R(−ρ, ±1, 0) results as a result of this. The inequality R(µ − ρ, ±1,
0) ≥ R(ν − ρ, ±1, 0) is still valid in light of the Theorem 5.3.7. Consequently,
the outcome is µ − ρ ≿ ν − ρ.
Case 3: If R(µ, 0, 1) ≥ R(ν, 0, 1). As a result, R(µ, 0, 1) + R(−ρ, 0, 1) ≥ R(ν, 0, 1)

+ R(−ρ, 0, 1) is an inequality. Therefore, the inequality R(µ − ρ, 0, 1) ≥ R(ν − ρ,
0, 1) is true according to the Theorem 5.3.7. As a result, µ − ρ ≿ ν − ρ.
☐
Proposition 5.4.13. Let µ, ν, µ − ρ, and ν − ρ be components of X. If µ ≻

ν by R on {µ, ν}, then µ − ρ ≻ ν − ρ by R on {µ − ρ, ν − ρ} is also followed.
Proof. The proof is evident from the proposition 5.4.12 alone. ☐
Proposition 5.4.14. Let µ, ν, µ + ρ, ν + ξ be components of X. If µ ≿ ν and

ρ ≿ ξ by R on {µ, ν, ρ, ξ}, then µ + ρ ≿ ν + ξ by R on {µ + ρ, ν + ξ} when θ1
and θ2 are same for both the inequalities µ ≿ ν and ρ ≿ ξ.
Proof. Given that, µ ≿ ν then there are three situations that can happen:
i) R(µ, 0, 0) ≥ R(ν, 0, 0), ii) R(µ, ±1, 0) ≥ R(ν, ±1, 0) and iii) R(µ, 0, 1) ≥
R(ν, 0, 1). Again, ρ ≿ ξ then there are three situations that can happen:
i) R(ρ, 0, 0) ≥ R(ξ, 0, 0), ii) R(ρ, ±1, 0) ≥ R(ξ, ±1, 0) and iii) R(ρ, 0, 1) ≥
R(ξ, 0, 1). Combining all the situations 3 cases arise:
Case 1: Let, µ ≿ ν and ρ ≿ ξ hold, which followed by R(µ, 0, 0) ≥ R(ν, 0, 0)

and R(ρ, 0, 0) ≥ R(ξ, 0, 0) respectively. Now, the inequalities V(µ) ≠ V(ν)
and V(ρ) ≠ V(ξ) hold, which leads to V(µ) > V(ν) and V(ρ) > V(ξ). Thus,
V(µ) + V(ρ) > V(ν) + V(ξ). Eventually, it follows that R(µ + ρ, 0, 0) > R(ν +
ξ, 0, 0). Hence, the result µ + ρ ≻ ν + ξ.
Case 2: Let, µ ≿ ν and ρ ≿ ξ hold, which followed by R(µ, ±1, 0) ≥ R(ν, ±1,
0) and R(ρ, ±1, 0) ≥ R(ξ, ±1, 0) respectively. Now, the inequalities V(µ) =
V(ν) and V(ρ) = V(ξ) hold. Then for θ1 = ±1 and θ2 = 0,
[A( ) F( )] [A( ) F( )] (5.23)
±[A( ρ ) + F( ρ )] > ±[A(ξ ) + F(ξ )] (5.24)
Now from Eqs. 5.23, 5.24 get that, ±[A(µ) +F(µ)] +±[A(ρ) +F(ρ)] >
±[A(ν) +F(ν)] +±[A(ξ) +F(ξ)]. This leads to the inequality V(µ) ± [A(µ) +
F(µ)] + V(ρ) ± [A(ρ) + F(ρ)] > V(ν) ± [µ(ν) + F(ν)] + V(ξ) ± [A(ξ) + F(ξ)].
Thus, R(µ + ρ, ±1, 0) > R(ν + ξ, ±1, 0). Hence, the result µ + ρ ≻ ν + ξ.
Case 3: Let, µ ≿ ν and ρ ≿ ξ hold, which followed by R(µ, 0, 1) ≥ R(ν, 0, 1)

and R(ρ, 0, 1) ≥ R(ξ, 0, 1). Now, the inequalities V(µ) = V(ν) and V(ρ) =
V(ξ) hold.
T( ) T( ) (5.25)
T( ) T( ) (5.26)
Now from Eqs. 5.25, 5.26 get that, T(µ) + T(ρ) ≥ T(ν) + T(ξ). This results
in the inequity {V(µ) + T(µ)} + {V(ρ) + T(ρ)} ≥ {V(ν) + T(ν)} + {V(ξ) +
T(ξ)}. Thus, R(µ, 0, 1) + R(ρ, 0, 1) ≥ R(ν, 0, 1) + R(ξ, 0, 1). Hence, the result
µ + ρ ≿ ν + ξ.
Proposition 5.4.15. Let µ, ν, µ + ρ, ν + ξ be components of X. If µ ≻ ν and

ρ ≻ ξ by R on {µ, ν, ρ, ξ}, then µ + ρ ≻ ν + ξ by R on {µ + ρ, ν + ξ} when θ1
and θ2 are same for both the inequalities µ ≻ ν and ρ ≻ ξ.
Proof. Given that, µ ≻ ν, in light of this, three cases come into play, these
are: i) R(µ, 0, 0) > R(ν, 0, 0), ii) R(µ, ±1, 0) > R(ν, ±1, 0) and iii) R(µ, 0, 1)
> R(ν, 0, 1). Once more, given that ρ ≻ ξ, then there are three cases that
come up: i) R(ρ, 0, 0) > R(ξ, 0, 0), ii) R(ρ, ±1, 0) > R(ξ, ±1, 0) and iii) R(ρ,
0, 1) > R(ξ, 0, 1). The following three cases are revealed when all the cases
are combined:
Case 1: Let, µ ≻ ν and ρ ≻ ξ occur for R(µ, 0, 0) > R(ν, 0, 0) and R(ρ, 0, 0)
> R(ξ, 0, 0) respectively. It follows that V(µ) ≠ V(ν) and V(ρ) ≠ V(ξ). This
means that V(µ) > V(ν) and V(ρ) > V(ξ). Thus, V(µ) + V(ρ) > V(ν) + V(ξ).
Finally, it is evident that R(µ + ρ, 0, 0) > R(ν + ξ, 0, 0). The outcome is µ +
ρ ≻ ν + ξ.
Case 2: Let, µ ≻ ν and ρ ≻ ξ occur for R(µ, ±1, 0) > R(ν, ±1, 0) and R(ρ, ±1,
0) > R(ξ, ±1, 0), which follows that V(µ) = V(ν) and V(ρ) = V(ξ).
[A( ) F( )] [A( ) F( )] (5.27)
[A( ) F( )] [A( ) F( )] (5.28)
Now from Eqs. 5.27, 5.28 get that, ±[A(µ) + F(µ)] + ±[A(ρ) + F(ρ)] >
±[A(ν) + F(ν)] + ±[A(ξ) + F(ξ)]. This results in the inequity V(µ) ± [A(µ) +
F(µ)] + V(ρ) ± [A(ρ) + F(ρ)] > V(ν) ± [A(ν) + F(ν)] + V(ξ) ± [A(ξ) + F(ξ)].
Thus, R(µ + ρ, ±1, 0) > R(ν + ξ, ±1, 0). Hence, the result µ + ρ ≻ ν + ξ.
Case 3: Let, µ ≻ ν and ρ ≻ ξ occur for R(µ, 0, 1) > R(ν, 0, 1) and R(ρ, 0, 1) >
R(ξ, 0, 1), which follows that V(µ) = V(ν) and V(ρ) = V(ξ).
T( ) T( ) (5.29)
T( ) T( ) (5.30)
Now from Eqs. 5.29, 5.30 get that, T(µ) + T(ρ) > T(ν) + T(ξ). This leads
to the inequality {V(µ) + T(µ)} + {V(ρ) + T(ρ)} > {V(ν) + T(ν)} + {V(ξ) +
T(ξ)}. Thus, R(µ, 0, 1) + R(ρ, 0, 1) > R(ν, 0, 1) + R(ξ, 0, 1). Hence, the result
µ + ρ ≻ ν + ξ.
Proposition 5.4.16. Let µ and ν represent the components of X. If µ ≿ ν by

R on {µ, ν}, which follows that −µ ≿ −ν by R on {−µ, −ν}.
Proof. When k = −1 is used in Property 5.4.10, the proof follows

immediately.
Proposition 5.4.17. Let µ and ν be the components of X. If µ succeeds ν by

R on {µ, ν}, which follows that −µ succeeds −ν by R on {−µ, −ν}.
Proof. The evidence is exactly the same as the proof for Property 5.4.16.
5.5 Comparative Analysis and Numerical Examples

This section evaluates the performance of the suggested method using sev-
eral sets of fuzzy numbers. Seven numerical examples are used to demon-
strate the recommended method’s capacity to discern between several
current approaches.
Example 5.5.1. Consider Two symmetric fuzzy numbers from [18], with
the same cores but different spreads, µ = (0.3, 0.6, 0.6, 0.9) and ν = (0.5,
0.6, 0.6, 0.7), are shown in Fig. 5.1. In comparison to ν, spreads for µ are
larger. As a result, the case is V(µ) < V(ν). As a result, the numerical data
in Table 5.1 show that µ < ν. The proposed method, as well as those by [17],
[15], [48], [13], [12], [6], [39], [20] and [9] give similar result. However, the
approaches of [15], [48], [41], [39], [39] rank the fuzzy numbers or images
of the fuzzy numbers inconsistently. Table 5.1 displays the outcomes of
various strategies.
µ
1.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
µ = (0.1, 0.6, 0.6, 0.8)

ν = (0.3, 0.6, 0.6, 0.7)
Figure 5.1 Fuzzy numbers μ and ν are represented graphically in Example 5.5.1.
Table 5.1 Rank order of fuzzy numbers in Example 5.5.1.

Methods µ ν Result −µ −ν Result
[17] 0.2619 0.2747 µ≺ν -0.2619 -0.2747 −µ ≻ −ν
[15] 0.7282 0.7426 µ≺ν 0.7186 0.7393 −µ ≺ −ν
[48] 0.6009 0.6289 µ≺ν 0.6009 0.6289 −µ ≺−ν
[13] 0.5250 0.5500 µ≺ν -0.5250 -0.5500 −µ ≻ −ν
[12] 0.3581 0.4318 µ≺ν -0.3795 -0.4509 −µ ≻ −ν
[6] 0.5250 0.5500 µ≺ν -0.5250 -0.5500 −µ ≻ −ν
[41] 0.0730 0.0686 µ ≻ν 0.0730 0.0686 −µ ≻ −ν
[39] 1.4447 1.5622 µ≺ν -0.6553 -0.6378 −µ ≺ −ν
[9] 0.1917 0.1944 µ≺ν -0.1917 -0.1944 −µ ≻ −ν
[49]
optimistic α = 1.0 0.3500 0.4500 µ≺ν 0.4500 0.3500 −µ ≻ −ν
moderate α = 0.5 0.5250 0.5500 µ≺ν 2.2205 0.1132 −µ ≻ −ν
pessimistic α = 0.0 0.7000 0.6500 µ ≻ν 9.8462 2.7308 −µ ≻ −ν
[20]
optimistic α = 0.9 0.1126 0.1131 µ≺ν -0.1126 -0.1131 −µ ≻ −ν
moderate α = 0.5 0.4250 0.4333 µ≺ν -0.4250 -0.4333 −µ ≻ −ν
pessimistic α = 0.1 0.5454 0.5616 µ≺ν -0.5454 -0.5616 −µ ≻ −ν
Proposed index
value 0.5500 0.5667 µ≺ν -0.5500 -0.5667 −µ ≻ −ν
Example 5.5.2. Two symmetric fuzzy numbers from [18], with the same
cores but different spreads, µ = (0.3, 0.6, 0.6, 0.9) and ν = (0.5, 0.6, 0.6,
0.7), are shown in Fig. 5.2. In comparison to ν, spreads for µ are larger. As
a result, it is clear that V(µ) = V(ν), and that under the given case A(μ) ≠
A(ν). As a result, Table 5.2 numerical data demonstrate that µ < ν. The pro-
posed method yields equivalent results to those of [12]’s, [39]’s, [9]’s and
[20]’s techniques. Other methods, which are unable to differentiate these
fuzzy numbers are [17], [15], [48], [13], and [6]. Other methods that rank
fuzzy numbers or images of fuzzy numbers inconsistently include [12],
[41], [39], [20], [49], and [9]. Table 5.2 displays the outcomes of various
methods.
µ
1.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
µ = (0.3, 0.6, 0.6, 0.9)

ν = (0.5, 0.6, 0.6, 0.7)
Figure 5.2 Graphical representation of fuzzy numbers µ and ν in Examples 5.5.2.
Table 5.2 Rank order of fuzzy numbers in Example 5.5.2.

[17] 0.3000 0.3000 µ ~ν -0.3000 -0.3000 −µ ~ −ν
[15] 0.7810 0.7810 µ~ν 0.7810 0.7810 −µ ~ −ν
[48] 0.6834 0.6834 µ~ν 0.6834 0.6834 −µ ~ −ν
[13] 0.6000 0.6000 µ~ν -0.6000 -0.6000 −µ ~ −ν
[12] 0.4456 0.5379 µ≺ν 0.4456 0.5379 −µ ≺ −ν
[6] 0.6000 0.6000 µ~ν -0.6000 -0.6000 −µ ~ −ν
(Continued)
Table 5.2 Rank order of fuzzy numbers in Example 5.5.2. (Continued)

[41] 0.1051 0.0896 µ ≻ν 0.1051 0.0896 −µ ≻ −ν
[39] 1.6174 1.6901 µ≺ν -0.7826 -0.7099 −µ ≺ −ν
[9] 0.0167 0.0056 µ≺ν 0.0167 0.0056 −µ ≺ −ν
[49]
optimistic α = 1.0 0.4500 0.5500 µ≺ν 0.4500 0.5500 −µ ≺−ν
moderate α = 0.5 0.6000 0.6000 µ~ν 1.0000 1.0000 −µ ~ −ν
pessimistic α = 0.0 0.7500 0.6500 µ ≻ν 8.5000 13.500 −µ ≺ −ν
[20]
optimistic α = 0.9 0.0028 0.0009 µ≺ν 0.0028 0.0009 −µ ≺ −ν
moderate α = 0.5 0.0500 0.0167 µ≺ν 0.0500 0.0167 −µ ≺ −ν
pessimistic α = 0.0 0.0972 0.0324 µ≺ν 0.0972 0.0324 −µ ≺ −ν
Proposed method -0.0100 0.1567 µ≺ν 0.0100 -0.1567 −µ ≻ −ν
.
Example 5.5.3. Consider the two typical fuzzy numbers µ = (4.0, 6.0,
9.0, 10.0) and ν = (2.0, 7.0, 8.0, 12.0), which are depicted in Fig. 5.3. They
have distinct cores and spreads. Furthermore, the spread of ν is wider
than the spread of µ. Therefore, it is clear that given the current situation,
µ ν
1.0
0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0
µ = (4.0, 6.0, 9.0, 10.0)

ν = (2.0, 7.0, 8.0, 12.0)
Figure 5.3 Graphical representation of TrFNs µ and ν in Example 5.5.3.

V(µ) = V(ν), A(µ) = A(ν), and F(µ) ≠ F(ν). The information in Table 5.3
thus proves that µ >- ν. The outcomes of the proposed method are com-
parable to those found in [17], [13], and [12]. The fuzzy numbers cannot
be ranked using [6]’s and [9]’s, respectively. Other approaches, including
[15]’s, [48]’s, [41]’s, [39]’s, [49]’s, and [20]’s are inconsistent in how they
rank the fuzzy numbers or the pictures of the fuzzy numbers. Table 5.3
presents the findings of several techniques.
Table 5.3 Rank the order of fuzzy numbers in Example 5.5.3.

[18] 3.6526 3.6322 µ≻ν -3.6526 -3.6322 −µ ≺ −ν
[15] 7.2400 7.1998 µ≻ν 7.2400 7.1998 −µ ≻ −ν
[48] 7.2359 7.1910 µ≻ν 7.2359 7.1910 −µ ≻ −ν
[13] 0.7250 0.6042 µ≻ν -0.7250 -0.6042 −µ ≺ −ν
[12] 0.5103 0.4033 µ≻ν -1.4607 -1.0837 −µ ≺ −ν
[6] 7.2500 7.2500 µ~ν -7.2500 -7.2500 −µ ~ −ν
[41] 1.2810 16.3350 µ≺ν 1.2810 16.3350 −µ ≺ −ν
[39] 14.342 14.048 µ≻ν -14.658 -14.952 −µ ≻ −ν
[9] 0.3333 0.3333 µ~ν 0.3333 0.3333 −µ ~ −ν
[49]
optimistic α = 1.0 5.0000 4.5000 µ≻ν 0.0037 0.0388 −µ ≺ −ν
moderate α = 0.5 7.2500 7.2500 µ~ν -0.0984 6.7971 −µ ≺ −ν
[20]
optimistic α = 0.9 0.2990 0.1370 µ≺ν 0.2990 0.1370 −µ ≺ −ν
moderate α = 0.5 1.3750 1.1250 µ≺ν 1.3750 1.1250 −µ ≺ −ν
Proposed method 4.5833 3.0833 µ≻ν -4.5833 -3.0833 −µ ≺ −ν
Example 5.5.4. As illustrated in Fig. 5.4, two arbitrary normal fuzzy num-
bers with separate spreads and cores are µ = (1.0, 2.0, 5.0, 9.0) and ν = (−1.0,
3.0, 6.0, 7.0). In light of this, it is evident that V(µ) = V(ν), A(µ) = A(ν), F(µ)
= F(ν), and T(µ) ≠ (ν). The numerical information in Table 5.4 thus shows
that µ >- ν. The suggested approach produces results that are comparable
to those found in the works of [17], [15], [15], [12], [6], [6], [39], and [49].
Other methods, such as those by [14], [9], and [22], order the fuzzy num-
bers in an inconsistent manner. Furthermore, the fuzzy number images are
inconsistently interpreted by [15], [48], [41], and [39]. Table 5.4 shows the
outcomes of different approaches.
µ ν
1.0
–1.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0
µ = (1.0, 2.0, 5.0, 9.0)

ν = (−1.0, 3.0, 6.0, 7.0)
Figure 5.4 Graphical representation of TrFNs µ and ν in Example 5.5.4.
Table 5.4 Rank the fuzzy numbers in Example 5.5.4 in order.

[17] 2.0535 1.9394 µ ≻ν -2.0535 -1.9394 −µ ≺ −ν
[15] 4.3862 3.6760 µ ≻ν 4.3862 3.6760 −µ ≻ −ν
[48] 4.3842 3.6610 µ ≻ν 4.3842 3.6610 −µ ≻ −ν
[13] 0.4722 0.5357 µ≺ν -0.4722 -0.5357 −µ ≻ −ν
[12] 0.3064 0.2969 µ ≻ν -0.6409 -0.6389 −µ ≺ −ν
[6] 4.2500 3.7500 µ ≻ν -4.2500 -3.7500 −µ ≺ −ν
[41] 8.7629 0.3741 µ ≻ν 8.7629 0.3741 −µ ≻ −ν
(Continued)
Table 5.4 Rank the fuzzy numbers in Example 5.5.4 in order. (Continued)

[39] 8.2428 7.2428 µ ≻ν -8.7572 -7.7572 −µ ≺ −ν
[9] 1.2500 1.4167 µ≺ν -1.2500 -1.4167 −µ ≻ −ν
[49]
optimistic α = 1.0 1.5000 1.0000 µ ≻ν 0.0040 0.4244 −µ ≺ −ν
moderate α = 0.5 4.2500 3.7500 µ ≻ν 0.0134 74.5764 −µ ≺ −ν
pessimistic α = 0.0 7.0000 6.5000 µ ≻ν 2.3564 249.305 −µ ≺ −ν
[20]
optimistic α = 0.9 0.6790 0.8410 µ≺ν -0.6790 -0.8410 −µ ≻ −ν
moderate α = 0.5 2.8750 3.1250 µ≺ν -2.8750 -3.1250 −µ ≻ −ν
pessimistic α = 0.1 3.9510 3.9690 µ≺ν -3.9510 -3.9690 −µ ≻ −ν
Proposed method 12.500 11.500 µ >- ν -12.500 -11.500 −µ ≺ −ν
Example 5.5.5. As illustrated in Fig. 5.5, the three arbitrary normal fuzzy
numbers µ = (−1.0, −1.0, −1.0, −1.0), ν = (0.1, 0.2, 0.2, 0.6), and ρ = (0.25,
0.275, 0.275, 0.3) have separate cores and spreads. Thus, it can be observed
that in the situation, V(µ) < V(ν) < V(ρ). Table 5.5’s demonstrates that
µ < ν < ρ. The suggested method yields results that are comparable to those
µ ν ρ
1.0
–1.0 –0.9 –0.8 –0.7 –0.6 –0.5 –0.4 –0.3 –0.2 –0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
µ = (−1.0, −1.0, −1.0, −1.0)

ν = (0.1, 0.2, 0.2, 0.6)
ρ = (0.25, 0.275, 0.275, 0.3)
Figure 5.5 Fuzzy numbers µ, ν, and ρ are graphically represented in Example 5.5.5.
Methods µ ν ρ Result −µ −ν −ρ Result

[17] - 0.1364 0.1375 ν ≺ρ - -0.1364 -0.1375 −ν ≻ −ρ
[15] - 0.5505 0.5706 ν ≺ρ - 0.5505 0.5706 −ν ≺ −ρ
[48] - 0.4485 0.4321 ν ≻ρ - 0.4485 0.4321 −ν ≻ −ρ
[13] -1.0000 0.2750 0.2750 µ≺ν ~ρ 1.0000 -0.2750 -0.2750 −µ ≻ −ν ~ −ρ
[12] -1.0000 0.2152 0.2673 µ≺ν ≺ρ 1.0000 -0.2152 -0.2673 −µ ≻ −ν ≻ −ρ
[6] -1.0000 0.2750 0.2750 µ≺ν ~ρ 1.0000 -0.2750 -0.2750 −µ ≻ −ν ~ −ρ
[41] - 0.0419 0.0185 ν ≻ρ - 0.0419 0.0185 −ν ≻ −ρ
[39] -1.5000 0.9935 1.0494 µ≺ν ≺ρ 2.5000 -0.1065 -0.0506 −µ ≻ −ρ ≻ −ν
[9] -0.3333 0.0750 0.0917 µ≺ν ≺ρ 0.3333 -0.0750 -0.0917 −µ ≻ −ν ≻ −ρ
(Continued)

[49]
optimistic α = 1.0 - 0.2593 0.0360 ν ≻ρ - -0.01004 0.0525 −ν ≺ −ρ
moderate α = 0.5 - 20.706 0.6960 ν ≻ρ - 0.0483 1.4367 −ν ≺ −ρ
pessimistic α = 0.0 - 69.78 4.6845 ν ≻ρ - 3.8571 27.786 −ν ≺ −ρ
[20]
optimistic α = 0.9 -0.1900 0.0394 0.0522 µ≺ν ≺ρ 0.1900 -0.0394 -0.0522 −µ ≻ −ν ≻ −ρ
moderate α = 0.5 -0.7500 0.1750 0.2063 µ≺ν ≺ρ 0.7500 -0.1750 -0.2063 −µ ≻ −ν ≻ −ρ
pessimistic α = 0.1 -0.9900 0.2466 0.2722 µ≺ν ≺ρ 0.9900 -0.2466 -0.2722 −µ ≻ −ν ≻ −ρ
Proposed method -1.0000 0.2500 0.2750 µ≺ν ≺ρ 1.0000 -0.2500 -0.2750 −µ ≻ −ν ≻ −ρ
of [12], [6], [39], [9], and [20]. These fuzzy numbers are ranked as ν < ρ
in other systems, such as the [17]’s and [15]’s, which do not rank the fuzzy
number C. Additionally, the fuzzy numbers are ranked as ν ≻ ρ and the
fuzzy number C is not ranked by [48], [41] and [49]. On the other hand,
[13] rank these fuzzily defined numbers as µ < ν < ρ. Furthermore, the
fuzzy number images are inconsistently interpreted by [15], [48], [41], and
[39]. Table 5.5 shows the outcomes of different approaches.
Example 5.5.6. As illustrated in Fig. 5.6, the three arbitrary normal fuzzy
numbers µ = (5.0, 6.0, 6.0, 7.0), ν = (5.9, 6.0, 6.0, 7.0), and ρ = (6.0, 6.0, 6.0,
7.0) have the same core but different spreads. As a result, it can be seen that
under the given circumstance, V(µ) < V(ν) < V(ρ). Table 5.6 numerical
data therefore demonstrates that µ < ν < ρ. The recommended method, as
well as those by [48], [12], [6], [41], [39], [13], [49], [9] and [20] give simi-
lar result. Other techniques, like those employed by [15] and [17], arrange
the fuzzy numbers as µ < ν and are unable to rank the fuzzy number C.
Additionally, [41], [48] and [15] give the inconsistent result to the images
of the fuzzy numbers. Table 5.6 displays the results of various methods.
ν
1.0
ρ
µ
1.0 2.0 3.0 5.0 6.0 7.0 8.0
µ = (5.0, 6.0, 6.0, 7.0)

ν = (5.9, 6.0, 6.0, 7.0)
ρ = (6.0, 6.0, 6.0, 7.0)
Figure 5.6 Fuzzy numbers µ, ν, and ρ are graphically represented in Example 5.5.6.

[17] 3.0000 3.1120 - µ≺ν -3.0000 -3.1120 - −µ ≻ −ν
[15] 6.0208 6.3193 - µ≺ν 6.0208 6.3193 - −µ ≺ −ν
[48] 6.0093 6.3088 6.3421 µ≺ν ≺ρ 6.0093 6.3088 6.3421 −µ ≺ −ν ≺ −ρ
[13] 0.8571 0.8893 0.8929 µ≺ν ≺ρ -0.8571 -0.8893 0.8929 −µ ≻ −ν ≻ −ρ
[12] 0.7358 0.8150 0.8210 µ≺ν ≺ρ -2.7846 -3.3494 -3.3937 −µ ≻ −ν ≻ −ρ
[6] 6.0000 6.2250 6.2500 µ≺ν ≺ρ -6.0000 -6.2250 -6.2500 −µ ≻ −ν ≻ −ρ
[41] 8.9553 10.1836 10.3310 µ≺ν ≺ρ 8.9553 10.1836 10.3310 −µ ≺ −ν ≺ −ρ
[39] 12.000 12.7667 12.8536 µ≺ν ≺ρ -12.000 -12.7667 -12.8536 −µ ≻ −ν ≻ −ρ
[9] 2.0000 2.0250 2.0278 µ≺ν ≺ρ -2.0000 -2.0250 -2.0278 −µ ≻ −ν ≻ −ρ
(Continued)
[49]
optimistic α = 1.0 0.0199 0.1519 0.5787 µ≺ν ≺ρ 0.5787 0.0117 0.0100 −µ ≻ −ν ≻ −ρ
moderate α = 0.5 0.0532 11.3937 36.854 µ≺ν ≺ρ 18.7936 0.0878 0.0271 −µ ≻ −ν ≻ −ρ
pessimistic α = 0.0 1.7282 85.3932 99.5437 µ≺ν ≺ρ 50.2718 6.5825 1.7282 −µ ≻ −ν ≻ −ρ
[20]
optimistic α = 0.9 1.1400 1.1442 1.1447 µ≺ν ≺ρ -1.1400 -1.1442 -1.1447 −µ ≻ −ν ≻ −ρ
moderate α = 0.5 4.5000 4.5750 4.5833 µ≺ν ≺ρ -4.5000 -4.5750 -4.5833 −µ ≻ −ν ≻ −ρ
pessimistic α = 0.1 5.9400 6.0858 6.1020 µ≺ν ≺ρ -5.9400 -6.0858 -6.1020 −µ ≻ −ν ≻ −ρ
Proposed method 6.0000 6.1500 6.1667 µ≺ν ≺ρ -6.0000 -6.1500 -6.1667 −µ ≻ −ν ≻ −ρ
Example 5.5.7. Let µ = (0.4, 0.5, 1.0), ν = (0.4, 0.7, 1.0), and ρ = (0.4, 0.9,
0.9, 1.0) are three randomly generated normal fuzzy numbers, which have
the same spreads but a different core, as shown in Fig. 5.7. As a result, it
is clear that under the current situation, V(µ), V(ν), and V(ρ). As a result,
the numerical data in Table 5.7 show that µ < ν < ρ. The suggested method,
along with those by [17], [15], [13], [20], [48], [12], [6], [39], [49] and [9]
give similar result. Other approaches, such as [43]’s unable to differentiate
these fuzzy numbers. Further, [48] give an inconsistent result to the images
of the fuzzy numbers. The results of several approaches are displayed in
Table 5.7.
5.6 Application
Fuzzy risk analysis Take into account the following n alternatives, Ai,
where 1 ≤ i ≤ n. The p criterion Cik, 1 ≤ k ≤ p, may be used to evaluate
each alternative. The alternative Ai is assessed using two parameters, the
likelihood of failure (Rik ) and the magnitude of the loss (w
ik ) in terms of
1 ≤ k ≤ p and 1 ≤ i ≤ n, respectively. The framework for risk analysis in an
uncertain world.
Step 1. For each alternative Ai, take into account the likelihood of failure
( Rik ) and magnitude of loss (w
ik ), which are linguistic expressions with the
values 1 ≤ k ≤ p and 1 ≤ i ≤ n, where n is the number of alternatives and p
is the number of criteria in each option.
ν
1.0
µ ρ
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
µ = (0.4, 0.5, 0.5, 1.0)

ν = (0.4, 0.7, 0.7, 1.0)
ρ = (0.4, 0.9, 0.9, 1.0)
Figure 5.7 Fuzzy numbers µ, ν and ρ are graphically represented in Example 5.5.7.

[17] 0.2991 0.3500 0.3993 µ≺ν ≺ρ -0.2991 -0.3500 -0.3993 −µ ≻ −ν ≻ −ρ
[15] 0.7783 0.8602 0.9276 µ≺ν ≺ρ 0.7913 0.8602 0.9241 −µ ≻ −ν ≻ −ρ
[48] 0.7157 0.7753 0.8360 µ≺ν ≺ρ 0.7157 0.7753 0.8360 -µ ≻ -ν ≻ -ρ
[13] 0.6000 0.7000 0.8000 µ≺ν ≺ρ -0.6000 -0.7000 -0.8000 −µ ≻ −ν ≻ −ρ
[12] 0.4489 0.5200 0.5557 µ≺ν ≺ρ -0.4168 -0.5200 -0.5985 −µ ≻ −ν ≻ −ρ
[6] 0.6000 0.7000 0.8000 µ≺ν ≺ρ -0.6000 -0.7000 -0.8000 −µ ≻ −ν ≻ −ρ
[41] 0.1367 0.1367 0.1367 µ≺ν ≺ρ 0.1367 0.1367 0.1367 −µ ~ −ν ~ −ρ
[39] 1.6227 1.8174 2.0227 µ≺ν ≺ρ -0.7773 -0.9826 -1.1773 −µ ≻ −ν ≻ −ρ
[9] 0.1778 0.2333 0.2889 µ≺ν ≺ρ -0.1778 -0.2333 -0.2889 −µ ≻ −ν ≻ −ρ
(Continued)

[49]
optimistic α = 1.0 0.0398 0.1176 0.8515 µ≺ν ≺ρ 0.8515 0.1176 0.0398 −µ ≻ −ν ≻ −ρ
moderate α = 0.5 0.0832 0.1000 12.016 µ≺ν ≺ρ 12.016 1.0000 0.0832 −µ ≻ −ν ≻ −ρ
pessimistic α = 0.0 1.1744 8.5000 25.1279 µ≺ν ≺ρ 25.128 8.5000 1.1744 −µ ≻ −ν ≻ −ρ
[20]
optimistic α = 0.9 0.0969 0.1330 0.1691 µ≺ν ≺ρ -0.0969 -0.1330 -0.1691 −µ ≻ −ν ≻ −ρ
moderate α = 0.5 0.4083 0.5250 0.6417 µ≺ν ≺ρ -0.4083 -0.5250 -0.6417 −µ ≻ −ν ≻ −ρ
pessimistic α = 0.1 0.5598 0.6930 0.8262 µ≺ν ≺ρ -0.5598 -0.6930 -0.8262 −µ ≻ −ν ≻ −ρ
Proposed method 0.5667 0.7000 0.8333 µ≺ν ≺ρ -0.5667 -0.7000 -0.8333 −µ ≻ −ν ≻ −ρ
Step 2. Using the fuzzy weighted mean technique, determine the overall
risk Ri of the alternative options Ai by integrating the Rik and w
ik of each
citation Cik. The total risk, denoted by Ri, is given as,
n
Rik Wik
R k 1
n
(ri1 , ri 2 , ri 3 , ri 4 ) (5.31)
k 1 Wik
Step 3. Then rank these fuzzy numbers Ri’s by the proposed ranking method.
However, in today’s society, savings may not be enough to provide

financial stability. Money sitting in lockers or even in a bank account may
not be sufficient. Capital appreciation might help investors beat inflation.
Compounding also aids in wealth building. Investing can also help some-
one achieve long-term goals like buying a house, taking a foreign vacation,
or preparing his/her retirement. Here, we’ll look at a possible application of
the methodology we just described.
Assume that there are five investment avenue(alternates) Ai : i = 1, 2,
· · · 5 in which a man want to invest his money, like A1: The commodity
market, A2: Stock market, A3: Gold, A4: Real easte and A5: Long term bond.
There are some criteria Cik : k = 1, 2, · · · 4 in which these investment ave-
nues are depend, like C1: The possibility of financial loss, C2: The sensitiv-
ity of capital sum to changes in inflation, i.e. the risks caused by external/
exogenous shocks to system of production, distribution and consumption,
C3: The amount of interest received (profit), C4: Liquidity i.e. the ease with
which a commodity may be swiftly bought or sold on the market for a price
that accurately reflects its market worth. On the basis of discussions with
several economists, the language phrases for the alternative are allocated
Table 5.8 Linguistic variables for ratings and

weights [42].
Linguistic variables Fuzzy numbers
Very Low (VL) (0.01, 0.02, 0.04, 0.07)
Low (L) (0.04, 0.10, 0.18, 0.23)
Medium (M) (0.32, 0.41, 0.58, 0.65)
High (H) (0.72, 0.78, 0.92, 0.97)
Very High (VH) (0.93, 0.98, 1.00, 1.00)
Table 5.9 Decision matrix of candidates versus decision makers

opinions and their aggregated fuzzy ratings of candidates.
Probability of
Criteria Suppliers failure (Rik) Severity of loss (Wik)
A1 C1 L L
C2 M M
C3 M M
C4 L VL
A2 C1 H VH
C2 VH VH
C3 L L
C4 H H
A3 C1 L VL
C2 L VL
C3 L VL
C4 VL VL
A4 C1 VL VL
C2 VL VL
C3 VL VL
C4 L L
A5 C1 L L
C2 L VL
C3 M M
C4 VH VH
from Table 5.8. The words allocated in terms of language are shown in
Table 5.9 and explained.
1. Commodity Market: A commodities market is a location where pri-

mary goods or raw materials can be purchased, sold, or traded.
2. Stock Market: The stock market is a network of exchanges and other

venues where shares of publicly traded companies can be purchased, sold,
and issued.
3. Gold: As the name implies, a gold fund invests in numerous types of

gold. It might take the form of actual gold or gold mining company equi-
ties. Gold funds invest in real gold to provide investors with the ease of
purchasing pure gold at a reasonable price. There is no risk of theft, and
these units can be sold at any moment at market value.
4. Real easte: Real estate is a type of real property. It’s not the same as
personal property. Real estate refers to land and anything endearingly
connected to it, whether created by nature or by man. Real estate can be
divided into five basic categories: residential, commercial, industrial, raw
land, and special use.
5. Long Term Bond: The Treasury’s longest maturity bond issue is known
as long term bonds. It can also be used to regular bond markets in order to
incorporate an issuer’s longest-term bond. A bond with a one-year matu-
rity is far more predictable and consequently less dangerous than one with
a 20-year maturity. Long-term bond treasury and other corporate long
bonds require a concentration on investing for long-term income, which
comes with its own set of risks and benefits.
Therefore, the total risk are
4
Rik Wik
R k 1
4
Wik
k 1
L L M M M M VL M
L M M VL
(0.1293, 0.2523, 0.7579, 1.3246)
4
Rik Wik
R k 1
4
Wik
k 1
VH H VL VL L L H H
VH VH L H
(0.6420, 0.7559, 0.9855, 1.1312)
4
Rik Wik
R k 1
4
Wik
k 1
L VL L VL L VL VL VL
VL VL VL VL
(0.0046, 0.0400, 0.2900, 1.3300)
4
Rik Wik
R k 1
4
Wik
k 1
VL VL VL VL VL VL L L
VL VL VL L
(0.0043, 0.0373, 0.2325, 0.9657)
4
Rik Wik
R k 1
4
Wik
k 1
L L VL L M M VH VH
L VL M VH
(0.4971, 0.6336, 0.9113, 1.1473)
Table 5.10 Fuzzy numbers for the five investment avenues.

Investment avenues Fuzzy numbers
R1 (0.1293,0.2523,0.7579,1.3246)
R2 (0.6420,0.7559,0.9855,1.1312)
R3 (0.0046,0.0400,0.2900,1.3300)
R4 (0.0043,0.0373,0.2325,0.9657)
R5 (0.4971,0.6336,0.9113,1.1473)
Now, we get that, the five fuzzy numbers for the five investment avenues
as shown in Table 5.10. Further, it is observed that, the suggested technique
ranks the investment options in the following order: R2 > R5 > R1 > R3 > R4,
which results in the inequality A2 > A5 > A1 > A3 > A4. It is seen that, the
investment avenue A2: Stock market has the highest risk and the investment
avenue A4: Real estate has the lowest risk among all the investment avenues.
5.7 Conclusions
Most current ranking methods, according to the evaluations, failed to rate
some of the equivocal figures. As a consequence, a new ranking system
based on fuzzy numbers’ value, ambiguity, fuzziness, and vagueness has
been proposed. Two new quantities, θ1 and θ2, are included in the suggested
method. As a result, the proposed approach is used to express and prove a
number of important theorems. Then all the characteristics of Wang and
Kerre were studied and shown in detail using theorems. To highlight the
benefits of the suggested technique, a comparison study was also con-
ducted. Studies also demonstrate that the methods previously described
failed to consistently rank the fuzzy number images, whereas the methods
put forth are successful in doing so. A real-world risk analysis application
based on investment avenue selection must also be addressed. The invest-
ment avenue A2: Stock market has the most risk, while the investment ave-
nue A4:Real esate has the lowest risk to invest someone’s money, according
to this discussion.
References
1. Abbasbandy, S. and Hajjari, T., A new approach for ranking of trapezoidal
fuzzy numbers. Comput. Math. Appl., 57, 3, 413–419, 2009.
2. Ahmad, N., Jacob, K., Mamat, M., Amir Hamzah, N.S., Solving a system of
fuzzy polynomials by ranking method. Far East J. Appl. Math., 43, 7–20, 01
2010.
3. Allahviranloo, T., Abbasbandy, S., Saneifard, R., A method for ranking of
fuzzy numbers using new weighted distance. Math. Comput. Appl., 16, 2,
359–369, 2011.
4. Allahviranloo, T. and Saneifard, R., Defuzzification method for ranking
fuzzy numbers based on center of gravity. Iran. J. Fuzzy Syst., 9, 57–67, 12
2012.
5. Arya, A. and Yadav, S.P., A new approach to rank the decision making units
in presence of infeasibility in intuitionistic fuzzy environment. Iran. J. Fuzzy
Syst., 17, 2, 183–199, 2020.
6. Asady, B., The revised method of ranking L-R fuzzy number based on devia-
tion degree. Expert Syst. Appl., 37, 7, 5056–5060, 2010.
7. Asady, B. and Zendehnam, A., Ranking fuzzy numbers by distance minimi-
zation. Appl. Math. Modell., 31, 11, 2589–2598, 2007.
8. Bortolan, G. and Degani, R., A review of some methods for ranking fuzzy
subsets. Fuzzy Sets Syst., 15, 1, 1–19, 1985.
9. Botsa, D., Rao, P., Boddu, V., Ranking parametric form of fuzzy numbers by
defuzzification based on centroids value and ambiguity. J. Intell. Fuzzy Syst.,
41, 1–15, 05 2021.
10. Chen, C.-C. and Tang, H.-C., Ranking nonnormal p-norm trapezoidal fuzzy
numbers with integral value. Comput. Math. Appl., 56, 9, 2340–2346, 2008.
11. Chen, S.M., New methods for subjective mental workload assessment and
fuzzy risk analysis. Cybern. Syst., 27, 5, 449–472, 1996.
12. Chen, S.-M. and Chen, J.-H., Fuzzy risk analysis based on ranking general-
ized fuzzy numbers with different heights and different spreads. Expert Syst.
Appl., 36, 3, Part 2, 6833–6842, 2009.
13. Chen, S.-M. and Sanguansat, K., Analyzing fuzzy risk based on a new fuzzy
ranking method between generalized fuzzy numbers. Expert Syst. Appl., 38,
3, 2163–2171, 2011.
14. Chen, Z., Huang, G., Chakma, A., Hybrid fuzzy-stochastic modeling
approach for assessing environmental risks at contaminated groundwater
systems. J. Environ. Eng., 129, 1, 79–88, 2003.
15. Cheng, C.-H., A new approach for ranking fuzzy numbers by distance
method. Fuzzy Sets Syst., 95, 3, 307–317, 1998.
16. Choobineh, F. and Li, H., An index for ordering fuzzy numbers. Fuzzy Sets
Syst., 54, 3, 287–294, 1993.
17. Chu, T.-C. and Tsao, C.-T., Ranking fuzzy numbers with an area between the
centroid point and original point. Comput. Math. Appl., 43, 1–2, 111–117,
2002.
18. Chutia, R., Ranking of fuzzy numbers by using value and angle in the epsi-
lon-deviation degree method. Appl. Soft Comput., 60, 706–721, 2017.
19. Chutia, R., Ranking of z-numbers based on value and ambiguity at levels of
decision making. Int. J. Intell. Syst., 36, 1, 313–331, 2021.
20. Chutia, R. and Chutia, B., A new method of ranking parametric form of
fuzzy numbers using value and ambiguity. Appl. Soft Comput., 52, 1154–
1168, 2017.
21. Chutia, R., Gogoi, M.K., Firozja, M.A., Smarandache, F., Ordering single-
valued neutrosophic numbers based on flexibility parameters and its reason-
able properties. Int. J. Intell. Syst., 36, 4, 1831–1850, 2021.
22. Chutia, R., Gogoi, R., Datta, D., Ranking p-norm generalised fuzzy numbers
with different left height and right height using integral values. Math. Sci., 9,
1, 1–9, 2015.
23. Chutia, R. and Saikia, S., Ranking intuitionistic fuzzy numbers at levels of
decision-making and its application. Expert Syst., 35, 5, e12292, 2018.
24. Chutia, R. and Saikia, S., Ranking of interval type-2 fuzzy numbers using
value and ambiguity, in: 2020 International Conference on Computational
Performance Evaluation (ComPE), pp. 305–310, 2020.
25. Darehmiraki, M., A novel parametric ranking method for intuitionistic fuzzy
numbers. Iran. J. Fuzzy Syst., 16, 1, 129–143, 2019.
26. Delgado, M., Vila, M., Voxman, W., On a canonical representation of fuzzy
numbers. Fuzzy Sets Syst., 93, 1, 125–135, 1998.
27. Dubois, D. and Prade, H., Fuzzy sets and systems: Theory and applications,
Academic Press, Inc., Orlando, FL, USA, 1980.
28. Ezzati, R., Allahviranloo, T., Khezerloo, S., Khezerloo, M., An approach for
ranking of fuzzy numbers. Expert Syst. Appl., 39, 690–695, 01 2012.
29. Ezzati, R., Enayati, R., Mottaghi, A., Saneifard, R., A new method for ranking
fuzzy numbers without concerning of real numbers. TWMS J. Pure Appl.
Math., 2, 256–270, 01, 2011.
30. Ezzati, R., Khezerloo, S., Ziari, S., Application of parametric form for ranking
of fuzzy numbers. Iran. J. Fuzzy Syst., 12, 59–74, 02 2015.
31. Ezzati, R. and Saneifard, R., A new approach for ranking of fuzzy numbers
with continuous weighted quasi-arithmetic means. Math. Sci. Q. J., 4, 143–
158, 06, 2010.
32. Jain, R., Decisionmaking in the presence of fuzzy variables. IEEE Trans. Syst.
Man Cybern., SMC-6, 10, 698–703, Oct 1976.
33. Jain, R., A procedure for multiple-aspect decision making using fuzzy sets.
Int. J. Syst. Sci., 8, 1, 1–7, 1977.
34. Kim, K. and Park, K.S., Ranking fuzzy numbers with index of optimism.
Fuzzy Sets Syst., 35, 2, 143–150, 1990.
35. Kumar, A., Singh, P., Kaur, A., Kaur, P., A new approach for ranking nonnor-
mal p-norm trapezoidal fuzzy numbers. Comput. Math. Appl., 61, 4, 881–
887, 2011.
36. Liou, T.-S. and Wang, M.-J.J., Ranking fuzzy numbers with integral value.
Fuzzy Sets Syst., 50, 3, 247–255, 1992.
37. Liu, P., Chen, S.-M., Wang, Y., Multiattribute group decision making based
on intuitionistic fuzzy partitioned maclaurin symmetric mean operators. Inf.
Sci., 512, 830–854, 2020.
38. Liu, P. and Wang, P., Multiple attribute group decision making method based
on intuitionistic fuzzy einstein interactive operations. Int. J. Fuzzy Syst., 22,
790–809, 02, 2020.
39. Nasseri, S., Zadeh, M., Kardoost, M., Behmanesh, E., Ranking fuzzy quanti-
ties based on the angle of the reference functions. Appl. Math. Modell., 37, 22,
9230–9241, 2013.
40. Nayagam, V.L.G., Jeevaraj, S., Dhanasekaran, P., An improved ranking
method for comparing trapezoidal intuitionistic fuzzy numbers and its
applications to multicriteria decision making. Neural Comput. Appl., 30, 2,
671–682, 2018.
41. Rezvani, S., Ranking generalized exponential trapezoidal fuzzy numbers

based on variance. Appl. Math. Comput., 262, 191–198, 2015.
42. Schmucke, K.J., Fuzzy sets: Natural language computations, and risk analysis,
Rockville, Md.: Computer Science Press, Incorporated, 1984.
43. Shakouri, B., Abbasi Shureshjani, R., Daneshian, B., Lotfi, F., A parametric
method for ranking intuitionistic fuzzy numbers and its application to solve
intuitionistic fuzzy network data envelopment analysis models. Complexity,
2020, 25, 09 2020.
44. Shureshjani, R.A. and Darehmiraki, M., A new parametric method for rank-
ing fuzzy numbers. Indagationes Math., 24, 3, 518–529, 2013.
45. Wang, X. and Kerre, E.E., Reasonable properties for the ordering of fuzzy
quantities (i). Fuzzy Sets Syst., 118, 3, 375–385, 2001.
46. Wang, X. and Kerre, E.E., Reasonable properties for the ordering of fuzzy
quantities (ii). Fuzzy Sets Syst., 118, 3, 387–405, 2001.
47. Wang, Y.-J. and Lee, H.-S., The revised method of ranking fuzzy numbers
with an area between the centroid and original points. Comput. Math. Appl.,
55, 9, 2033–2042, 2008.
48. Wang, Y.-M., Yang, J.-B., Xu, D.-L., Chin, K.-S., On the centroids of fuzzy
numbers. Fuzzy Sets Syst., 157, 7, 919–926, 2006.
49. Yu, V.F., Chi, H.T.X., wen Shen, C., Ranking fuzzy numbers based on
epsilon-deviation degree. Appl. Soft Comput., 13, 8, 3621–3627, 2013.
50. Yu, V.F. and Dat, L.Q., An improved ranking method for fuzzy numbers with
integral values. Appl. Soft Comput., 14, Part C, 603–608, 2014.
6
Evacuation of Attributes to Translucent
TNSET in Mathematics Using
Rough Topology
Kala Raja Mohan*, R. Narmada Devi, Nagadevi Bala Nagaram,
Sathish Kumar Kumaravel and Regan Murugesan
Department of Mathematics, Vel Tech Rangarajan Dr. Sagunthala R&D Institute of

Science and Technology, Avadi, Chennai, TamilNadu, India
Abstract
Nowadays, clearance of competitive examinations is the major challenge faced
by many educators. The method of selecting toppers, based on the rank or grade
or percentage of marks in an examination, is the task involved in such competi-
tive examination. This paper aims at bringing out an analysis on determining the
salient factors involved in getting through TNSET in mathematics using the con-
cept of rough topology.
Keywords: TNSET, rough set, rough topology, lower approximation, upper

approximation, basis
6.1 Introduction
Selection of relevant candidates in any field based on their knowledge,
aptitude, skill, and other factors is made through suitable assessment. The
assessment conducted lists out the marks or grades. The required num-
ber of candidates are chosen from the prepared list and given the credit of
clearing the exam. This procedure of selecting the candidates is referred as
competitive exams.
99
Competitive exams are conducted in all fields with various purposes.

Few exams like NEET are conducted with the purpose of providing admis-
sion for higher studies. Few exams like CSIR NET aim at giving certifica-
tion to people to get through it for using it as a tool for jobs prescribed.
These types of exams are conducted either with objective or descriptive
type of questions. Assessment of questions is done based on the number of
questions with the specified time management technique.
In the analysis of this examination, the concept of rough topology is
applied. The concept of rough set theory was introduced by Pawlak in the
year 1982. It is an extension of set theory which was designed to sort out
a solution in decision making problems. The theory is applied in various
fields and has gained the interest of many researchers. Few of them are as
stated below.
Yao and Zhao discussed about attribute reduction using rough topology
[1]. Nasiri and Mashinchi made a study on decision tables [2]. Thivagar
et al. discussed about medical events [3]. Tao Yan and Chongzhao Han
proposed a rough topology method of attribute selection using conditional
entropy [4]. Zhang et al. made a survey on applications of rough set topol-
ogy [5]. Paul made a medical diagnosis analysis using rough topology [6].
Chaturveri et al discussed about the various concepts involved in rough set
topology [7]. Kanchana and Rekha discussed about corona virus diagnosis
[8]. Gomathi et al. have made a study fuzzy rough set using its properties
with local closed sets and local compactness [9, 10].
In this chapter, section 6.2 describes the standard definitions applied in
rough topology. Section 6.3 presents the algorithm involved in the process.
Section 6.4 demonstrates the information system applied in this process.
Section 6.5 depicts the working rule applied with an example. Section 6.6
is about the conclusion followed by references.
6.2 Basic Concepts of Rough Topology

The following standard definitions are applied in the analysis of the attri-
butes applied in TNSET examination.
6.2.1 Conditional Attribute

The set of condition required to bring out the output is defined by condi-
tional attribute.
TNSET Attributes Evacuation by Rough Topology 101
6.2.2 Decision Attribute

The output, which is aimed for the study, is named as decision attribute.
6.2.3 Rough Topology

The universal set and null set together with the lower approximation, upper
approximation, boundary region, and the outer region are defined as the
rough topology.
6.2.4 Lower Approximation

The non empty set chosen from the target set, which is the union of equiv-
alent classes is lower approximation.
6.2.5 Upper Approximation

The collection of equivalent classes, which has non empty intersection with
the target set, is upper approximation.
6.2.6 Boundary Region

The difference set between lower and upper approximation is referred as
boundary region.
6.2.7 Basis
The collection of the universal set, null set together with the lower approx-
imation and the boundary region are the basis.
6.2.8 Information System

The data collected, with respect to the attributes, from various persons
referred as objects, are arranged in the form of a matrix. The matrix is
called information system.
6.2.9 Core
The rough topology and the basis set removing each attribute are framed.
The attributes, which corresponds to different basis corresponding to the
basis of the target set, are collectively known as the core.
6.3 Algorithm
The step-by-step procedure to be carried out in this analysis is as follows:
Step 1: the set of condition and decision attributes are defined with
respect to the analysis to be made.
Step 2: the data related to the attributes are collected and represented
as an information table. The attributes are represented by column and
the objects corresponding to the attributes are represented by rows.
Step 3: the universal set U and the subset X of U are framed.
Step 4: lower approximation, upper approximation, boundary region
are found.
Step 5: rough topology and basis are identified.
Step 6: the abovementioned two steps are repeated removing each
attribute one by one.
Step 7: identifying the attributes for which the basis are different is
done.
Step 8: the core set is framed using the results of step 6.
6.4 Information System

In the process of forming information system, eight condition attribute
required to achieve the decision attribute are framed. The decision attri-
bute is to refer whether the candidate has cleared TNSET. In the process of
achieving the target of clearing the exam, the following eight attributes are
identified as condition attributes.
Q1) Whether the candidate has prepared for the examination personally?
Q2) Whether the candidate has adopted group study in the preparation of
the examination?
Q3) Has the candidate learnt through any private coaching centers?
Q4) Is the candidate preparing for CSIR NET exam too? It is the country
level competitive examination, whereas TNSET is the state level examina-
tion conducted by Tamil Nadu State Government.
Q5) Has the candidate has prepared for the exam going through the entire
syllabus of TNSET?
Q6) Has the candidate involved in revision and practice on a regular basis?
Q7) Is the candidate employed?
Q8) Has the candidate faced stress while preparing for the examination?
The questionnaire raised with respect to the decision attribute is

“Whether the candidate has cleared the examination?” which is referred as
D in the information system.
All these attributes are set to be with two choices of answers, like yes
or no. The questions with respect to the above mentioned attributes are
prepared in Google form and been circulated to academicians to get
their response. Mathematicians from districts of Chennai, Thiruvallur,
Thoothukudi, Thirunelveli, and Trichy were considered for this survey.
Around 60 data were collected. These data were collected based on the
aspirants who attended the examination during the year 2016 to 2019.
This year has been identified since the examination has not been con-
ducted during the year 2020 and 2021. The result obtained from this
work will be helpful for the aspirants to clear the examinations in the
future years. With a motive of this, from the responses received, eight
of the respondents’ answers on a random basis are taken into consid-
eration for this analysis and the corresponding information system is
formed. The eight respondents are referred as P1, P2, P3, P4, P5, P6, P7, P8.
The options chosen as yes are referred as A and the options chosen as
no are referred as B in Table 6.1.
Table 6.1 Options recorded by respondents corresponding to the

attributes referred.
Object Q1 Q2 Q3 Q4 Q5 Q6 Q7 Q8 D
P1 A B B A A A A B A
P2 A B B A A B A A A
P3 A A B A A A A A A
P4 A A B B A A A B A
P5 A A B A A B A A B
P6 B B B B B B A B A
P7 B A B A A A A A B
P8 A A B A A B A A B
6.5 Working Procedure

The universal set is U = {P1, P2, P3, P4, P5, P6, P7, P8}
The target set X = {P1, P2, P3, P4, P6}
Equivalence Class is {P5, P8} {P1} {P2} {P3} {P4} {P6} {P7}
Lower Approximation is RL = {P1, P2, P3, P4, P6}
Upper Approximation is RU = {P1, P2, P3, P4, P6}
Boundary Region RNR = {Ø}
Outside Region RNR = {P5, P7, P8}
Rough Topology is {U, Ø, {P1, P2, P3, P4, P6}, {P1, P2, P3, P4, P6}, {Ø},
{P5, P7, P8}}
Basis is {U, Ø, {P1, P2, P3, P4, P6}, {Ø}}
The above working procedure is repeated by removing each attribute. It

is found that the basis formed by removing the attributes Q3, Q4, Q5, Q7, Q8
is the same as the above mentioned basis. The basis obtained by removing
the attributes Q2, Q6 are respectively {U, Ø {P1, P3, P4, P6}}, {Ø}} and
{U, Ø, {P3}, {P5, P7, P8}} which are different from the above mentioned basis.
6.6 Conclusion
Rough set topology has attracted researchers in bringing out solution to
decision making problems. In this chapter, various aspects involved in
clearing TNSET in mathematics are discussed. Using the method of rough
topology, the predominant attributes required for clearing the examina-
tion is identified. This work helps as a source for the aspirants of TNSET
to know the important factors that should be followed to get through the
examination.
References
1. Yao, Y. and Zhao, Y., Attribute reduction in decision- theoretic rough set
models. Inf. Sci., 178, 3356–3373, 2008.
2. Nasiri, J.N. and Mashinchi, M., Rough set and data analysis in decision
tables. J. Uncertain Syst., 3, 232–240, 2009.
3. Lellis Thivagar, M., Richard, C., Paul, N.R., Mathemcatical innovations of a
modern topology in medical events. J. Inf. Sci., 2, 33–36, 2012.
4. Yan, T. and Han, C., A normal approach of rough conditional entropy-based

attribute selection for incomplete decision system. Math. Probl. Eng., 2014,
1–15, 2014.
5. Zhang, Q., Xie, Q., Wang, G., A survey on rough set theory and its applica-
tions. CAAI Trans. Intell. Technol., 1, 323–333, 2016.
6. Paul, N.R., Rough topology based decision making in medical diagnosis. Int.
J. Math. Trends Technol., 18, 40–43, 2016.
7. Gomathi, G., Narmada Devi, R., Praba, B., Nx locally compactness in fuzzy
rough topological spaces. Int. J. Pure Appl. Math., 109, 302–310, 2016.
8. Chaturveri, P., Daniel, A.K., Khusboo, K., Concept of rough set theory
and its applications in decision making processes. Int. J. Adv. Res. Comput.
Commun. Eng., 6, 43–46, 2017.
9. Kanchana, M. and Rekha, S., Decision making using rough topology and
indiscernibility for corna virus diagnosis. Int. J. Sci. Res. Sci. Eng. Technol., 7,
31–33, 2020.
10. Gomathi, G., Narmada Devi, R., Sophia Ponmalar, D., A study on properties
of fuzzy rough topological spaces. Int. J. Adv. Res. Manage. Architecture Tech.
Eng., 7, 13–17, 2021.
7
Design of Type-2 Fuzzy Controller for
Hybrid Multi-Area Power System
Susmit Chakraborty1, Arindam Mondal2* and Soumen Biswas2
1
Department of Electrical & Electronics Engineering, Pailan College of Management
& Technology, Kolkata, India
2
Department of Electrical Engineering, Dr. B C Roy Engineering College,
Durgapur, India
Abstract
In this chapter, a new fuzzy method named as interval type-2 fuzzy inference sys-
tems (IT2FIS) is proposed along with Fractional order PID (FOPID) controller
to control tie-bar power error and frequency error of power system containing
multi area. Two area system containing conventional as well as non-conventional
sources like solar-thermal and wind-hydroelectric systems are studied in this
chapter. Type-2-fuzzy parameters are trained by Levenberg–Marquardt algorithm
(LMA). A comparison is done for the responses of the systems when controlled by
PID-type-2-fuzzy combined controller and FOPID-type-2-fuzzy combined con-
troller. PID parameters (Kp, Ki, Kd) are optimized using metaheuristic Chaotic
Atomic Search Optimization (CASO) algorithm. Simulation results are analysed
though the obtained parameters such as settling time, overshoot and undershoot
when the system is controlled by both the complex controllers with and without
disturbances and the same is compared with the results obtained using conven-
tional PID controller. The mathematical formulations along with the simulation
results are illustrated using MATLAB to prove the efficacy of the proposed con-
troller design methodology for hybrid multi-area power system using type-2 fuzzy
FOPID controller.
Keywords: Automatic generation control (AGC), chaotic atomic search
optimization (CASO), fractional order controller, Levenberg–Marquardt
algorithm, power system optimization, type-2 fuzzy inference systems
107
7.1 Introduction
Nowadays interconnected power system becomes popular due to heavy
demand of the load [1]. The main prominent drawback [2] of this intercon-
nected power system is the mismatching of tie line power and frequency
between the inter connected systems. This problem can be minimized
or even solved with the help of Load Frequency Control (LFC) [3] or
Automatic Generation Control (AGC) [4, 5]. Many controllers [6, 7] are
already being used to control Load frequency, but being proved inefficient.
Normal PI controller [8] is simple but frequency deviation is very wide to
tolerate. In conventional PI controller, controlling gain is constant, so it is
failed to provide the constant frequency and power to the load in different
conditions. So, self-tuning gain controller [9] is required to provide the
optimum performance in wide range of operation. Fuzzy [10, 11] has a
power to choose different values of controlling gain as per the frequency
deviation between inter connected systems [12]. Type-1 fuzzy [10, 11] is
a knowledge base controller where membership functions are static and
controlling action is done as per the rule base. Type-2-fuzzy system [13,
14] is a kind of fuzzy system where grading to membership function is
provided and can be used along with conventional PID and FOPID con-
trollers. Fractional order PID controller are more suitable to control any
system than the conventional PID controller due to its much number of
degrees of freedom [15]. Moreover, renewable resources like wind, solar,
ocean thermal, etc. produce high fluctuations in generation of electric
power and also give frequency error [16]. In order to reduce the mismatch-
ing of tie line power and frequency between the interconnected systems
load frequency controller is in demand particularly in the connected sys-
tem with renewable sources. In this chapter the advantages of the type-2
fuzzy theory along with the conventional PID and FOPID controllers are
utilized to control the network for optimum functioning. In the literature,
there are many nature inspired metaheuristic algorithms to optimize the
controller parameters like PSO [17], CSA [18], BFA [19]. But due to fast
converging behaviors of Chaotic Atomic Search Optimization (CASO)
[20] it is being utilized to get the optimum values of controller parameters
to control the hybrid multi-area power system.
7.2 Plant Model

In this chapter, a two-area interconnected power system is considered as
the plant to be controlled where two different units are present. Each area
Type2 Fuzzy Contr for Hyb Mult Area Power Sys 109
(Thermal Power Plant)

1 1 1 + S KrTr
+
+ KT
1 + STg 1 + STt 1 + S Tr
ACE1 TYPE 2 FUZZY CONTROLLER 1 1 ∆F1

∆u R1
∆r + KPS β1 +
+ ACE1
+
1 + S TPS
+ a + Sb
+
S2 + cS + d
(Solar Plant) ∆Ptie12
+
+
(Hydro-Electric Power Plant)

1 1 + S TR 1 + S TW
+
+ KT α
1 + S TRH 1 + S TGH 1 + 0.5 S TW
ACE2 1
∆F2
TYPE 2 FUZZY CONTROLLER 1
∆u R2
∆r KPS β2 ++ ACE2
+
+
1 + S TPS
1
++
1 + S TWT
(Wind Power Plant)
Figure 7.1 Block diagram of the hybrid two area interconnected power system [21].
contains a combination of renewable and conventional source. The first

area is comprised of a solar photovoltaic unit and thermal power plant [19,
21] and the second area contains a wind power unit and a hydro-electric
power unit [19, 21]. The interconnected power system is called a hybrid
system as conventional as well as renewable energy sources are incorpo-
rated in both the areas to develop the model. The block diagram of the
model is shown in the Figure 7.1.
Thermal power plant is comprising of governor block, turbine block and
reheat block. Each of the blocks are represented by their own transfer func-
tions [8] as demonstrated below.
1
Transfer function of generator block: ; Transfer function of turbine
1 STg
1 1 SK r
block: ; Transfer function of reheat block: .
1 STt 1 STr
Solar power plant consists of solar panel, inverter and filter units and over-
all plant is represented by the following transfer function [21].
a bS
Transfer function of solar power plant: 2
S cS d
Hydro-Electric Power Plant comprises of mechanical hydraulic governor
and hydro turbine [10]. The transfer function representation of the hydro
1 1 STR
electric power plant:
1 STRH 1 STGH
7.3 Controller Design

In this work, two types of controllers are used to eliminate the tie line
power and frequency between the inter connected systems and type-2
fuzzy system is associated with the controllers to find the applicability of

type-2 fuzzy system in power system generation. Conventional PID con-
troller as well as FOPID controllers are studied in cascading with type-2-
Fuzzy controller and their results are compared.
7.3.1 Proportional Integral Derivative (PID) Controller

It is the most common type of controller used in industrial applications
[23] though it produces certain range of error [24] and for such reason
are not used efficiently for load frequency controller design in power sys-
tem network. The schematic diagram of PID controller along with Plant is
Generally, PID controllers are having proportional, integral and deriv-
ative sections. Three gains Kp, Kd and Ki are proportional gain, derivative
controller gain and integral controller gain respectively. The transfer func-
tion of a PID controller is given by (7.1).
Y (S) KI
GPID K p SK d (7.1)
E(S) S
Where, Y(S) is the output of the PID controller and E(S) is the error
signal generated. Type-2-Fuzzy aided ACE output signal is fed to the con-
troller for the controlling operation. For getting the optimum result, three
(Kp, Kd and Ki) parameters of the PID controller needs to be tuned. The
parameters are tuned by CASO algorithm.
Kp
Set Point Error + Output

+ Ki 1/S + Plant
– +
Kd S
Figure 7.2 Structure of PID controller.

7.3.2 Fractional Order Proportional Integral Derivative

(FOPID) Controller
FOPID Stands for Fractional Order Proportional Integral Derivative con-
troller. It was firstly implemented by Podlubny in [25]. The FOPID con-
troller outperforms the PID controller due to its larger number of degrees
of freedom and are more robust [15]. However, controller parameters are
more complex to tune than that of the conventional PID controller. The
different designing schemes are developed for FOPID controllers [26–29].
Other than the three parameters (Kp, Kd and Ki), there are two more param-
eters (λ and µ) are to be considered to represent a FOPID controller. The
transfer function of a FOPID controller is given by (7.2). The schematic
representation of FOPID controller is depicted in Figure 7.3. The parame-
ters are tuned by CASO algorithm.
YF (S) KI
GFOPID K p S Kd (7.2)
EF (S) S
where, µ = degree of derivative, λ = degree of integration.

The schematic representation of FOPID controller is depicted in Figure
7.3. The parameters are tuned by CASO algorithm.
7.3.3 Type-2-Fuzzy Logic

Fuzzy is very powerful multi valued logical operation that acts on impre-
cise granular information from a set of collection of data. Fuzzy Logic
System comprises of a set of rules. It is very common practice that the
knowledge of building these rules is uncertain. These uncertain cases
can be handled by using Fuzzy Logic Controllers (FLC) [10–14]. There
Kp
Set Point Error + Output

+ Ki 1/Sλ + Plant
– +
Kd Sµ
Figure 7.3 Structure of FOPID controller.

are two different approaches for FLC design: type-1 FLCs (T1FLCs) and
type-2 FLCs (T2FLCs). Fuzzy logic was invented by Zadeh [30]. Fuzzy
Controlling action has an ability to handle uncertainties and imprecisions.
Using some statistical data related to system activity, it works. Any inter-
mediate state of input can be successfully handled by using Fuzzy logic
controller (FLCs). Hence, the FLC has established as a powerful technique
for various applications [31]. A Type-1 fuzzy controller deals with ordinary
fuzzy sets (OFSs) to interference system. But, type-2 Fuzzy controller has
a membership grade which itself is a fuzzy. So, Type-2 fuzzy Controller
is known as Fuzzy-Fuzzy set, whereas crisp type memberships [32] are
belonged to Type-1 fuzzy set.
Type-2 fuzzy system also contains two different membership functions
called Primary and Secondary Membership Functions (PMF & SMF).
PMF is a subset in [0,1] which determines a membership grade of type-2
and SMF is basically a probability of occurrence of PMF. Mathematically,
type-2 fuzzy set can be denoted as U  that is characterized by a type-2
membership function U (a,u), where a ∈ X and u ∈ Ga ⊆ [0, 1].
i.e, U {((a, u), U (a, u)) | a X , u Ga [0,1]}, in which 0 

U (a, u) 1.
(7.3)
For this case the above equation can be expressed as follow:
 
U (a, u)
U , Ga [0, 1] (7.4)
a X u Ga (a, u)
Double integration in the above equation denotes union of a and u

available. Ga denotes the primary membership set of the fuzzy type-2 con-
troller which is the membership grade or equivalent to type-1 Fuzzy. U  is
the union of all primary memberships, also called footprint of uncertainty
(FOU). When U  (a,u) 1 for ∀u ∈ G ⊆ [0,1] a type-2fuzzy controlling set
a
(IT2FLC) is obtained. Figure 7.4 shows a typical type-2 fuzzy triangular
type set. PMF is shown in Figure 7.4a and its associated membership func-
tions of triangular type and interval type are represented in the Figure 7.4b
and Figure 7.4c respectively. The Interval Type-2 Fuzzy logic controlling
set (IT2FLC) when SMF is considered as shown in Figure 7.4c obtained.
Two important aspects like uncertainties and the description of the entire
secondary grades of a type-2 fuzzy logic controller membership function
[32] is identified by the shaded region of Figure 7.4a. The complete block
diagram representing the whole IT2FLC is represented in Figure 7.5.
IT2FLC consists of i) a type-2 fuzzifier that converts all real-life val-
ues into crisp values, ii) rule base that is used to process the fuzzifying
operations, iii) type reducer and iv) de-fuzzifier. Real life parameters like
Area Control Error (ACE) along with its derivatives go into the fuzzifier
first. Fuzzy sets are formed that is processed with the help of rule base of
the type-2 fuzzy sets. Then outputs in terms of type-2 fuzzy sets are pro-
cessed by the type-reducer performing centroid calculation. Type reduced
sets can be represented as LTR = [L1,Lr], L1 and Lr are the two end points of
type reduced fuzzy sets. In last stage, de-fuzzifier unit de-fuzzify the type-1
fuzzy sets to give crisp outputs [33]. The rule base that is used in this fuzzy
controller for Load Frequency Control is given as follows:
(a)
1
_
µū(x) Ga
µū(x)
0
X
(b) (c)
1 1
0 _ 0 _
µū(x) µū(x) 1 u µū(x) µū(x) 1 u
Figure 7.4 Illustration of (a) type-2 fuzzy membership function (b) triangular secondary
membership function (c) interval secondary membership function.
OUTPUT
PROCESSING Crisp
RULES
Output
Crisp DEFUZZIFIER
Input
FUZZIFIER
TYPE-REDUCER
INFERENCE
Type-2 Input Type-2 Output
Fuzzy Sets Fuzzy Sets
Figure 7.5 Block diagram representation of IT2FLC.

If ACE U  1i and ACE U  2i then u = C Where U  1i and U

 2i are
I
interval type-2 fuzzy sets, CI = singleton and I = 1,2,3…integer values.
In this chapter, two inputs as Area Control Error (ACE) and its deriva-
tives (∆ACE) are fuzzified by positive and negative intervals type-2 fuzzy
as shown in Figure 7.1. Notations used for representing PMF of positive
ACE, negative ACE, positive derivative of ACE and negative derivative of
ACE are respectively μp[ACE], μn[ACE], μp[ΔACE], μn[ΔACE]. SMF of
type-2 fuzzy controller is kept constant. Type-1 fuzzy set that is used in
final defuzzification process, can be mathematically represented by (7.5).
0 ( , L1 )
(L1 ACE)
p[ ACE] ( L1 , L1 ) (7.5)
2L1
1 (L1 , )
Corresponding rules of type-1 fuzzy controller is consolidated in Table

7.1.
There are some prominent Merits of this type-2 Fuzzy controller as
listed below.
• Easy application for linear and non-linear type load.

• Capability to handle internal as well as external disturbances
of the system.
• Dynamic response is better compared to conventional type
controllers.
Table 7.1 Rule base for type

reduction.
ΔACE
ACE --- N Z P
N P N N
Z N P P
P N N N
7.4 Levenberg–Marquardt Algorithm

Fuzzy system can be trained by using Levenberg Marquardt Algorithm
(LMA) [34] which is very much efficient for its high-speed converging rate.
For this optimization technique the cost function is considered as given
by (7.6).
1
E (error )2 (7.6)
2
Gradient is taken as G = JT * (error), where J denotes the Jacobian matrix.

LMA employs the following equations in different stages to complete
the process.
Hessian Matrix [35]: Q(k) = (JTJ + μI)−1
1
Adjusted Weight: w t 1 w t JTt J t I J t error
Start
Initialize the weights and

the parameters
ωt, m = 1
Error evaluation
Et
µ = µ/β ωt = ωt+1
0<β<10
ωt = ωt+1 Compute the Jacobian matrix
T m=m+1
ωt+1 = ωt – (Jt Jt + µI)–1JtTe
m≤5 m>5
Et+1 ≤ Et Compute: Et+1 Et+1 > Et
Et+1 ≤ Emax
End
Figure 7.6 Flow chart of Levenberg Marquardt algorithm.

“I” is identity matrix and µ represents the coefficient of combination.

The flowchart of the LMA is demonstrated in Figure 7.6.
7.5 Optimization of Controller Parameters Using

CASO Algorithm
CASO [20] method is very efficient in the optimization process for the
parameters of PID and FOPID controllers in AGC system.
Step 1: Initialize the control parameters (Kp, Kd and Ki for PID and µ, λ
additionally for FOPID) along with population size (F) and maximum
Iteration number (T).
Step 2: Set iteration T = 0 and calculate the fitness value for the mth particle.
d
Step 3: Compute Fmm using the equation Fmn rand n Fmn
nk
Step 4: Find the constrain force Vmd ( dbest m
d )
Step 5: Take the accelerating factor using
13 7
1
2 2 Hmn Hmn d
n
d
m
2 best m
d d d
p m 1 e T rand j Qe T
T j k
nj m , n 2 nj
Step 6: Take the Chaotic mapping using
xk
xk 1 mod(rx k (1 x k ) (4 r)sin( ,1)); r (1, 4)
4
Step 7: Update the control parameters in mth position.

Step 8: Preserve the best fitness value of the objective function.
Step 9: T = T+1.
Step 10: Extract the best Kp, Kd, and Ki for PID and µ, λ additionally for
FOPID after finding optimal solution otherwise return to step 2.
7.6 Result and Analysis

7.6.1 Without Disturbances
Hybrid Two Area Power System (shown in Figure 7.1) containing one
renewable energy source along with one conventional type unit in each area
Table 7.2 Performance analysis of two area hybrid system using IT2FLC based
PID and IT2FLC FOPID controllers without disturbances.
Two area test system
Functions Parameters PID IT2FLC-PID IT2FLC-FOPID
Δf1(Hz) OS 0.0098 0.0007962 0.0009518
US -0.0185 -0.008381 -0.008390
ST 32.17 15.61 20.56
Δf2(Hz) OS 0.00512 0.0006668 0.0007523
US -0.015 -0.006460 -0.006403
ST 31.19 13.77 20.12
ΔPtie(puMW) OS 0.00188 0.00056 0.0007111
US -0.00689 -0.004936 -0.005139
ST 30.47 14.08 20.83
has been simulated by using MATLAB. A small perturbation of demand

ΔPD1 and ΔPD2 are applied in 1st and 2nd area respectively. Frequency errors
ΔF1 and ΔF2 are seen in respective areas along with a tie-bar power devia-
tion of ΔPtie. These three terms are tabulated in Table 7.2. The Area1 con-
sists of Thermal Power and Solar plant whereas Wind and Hydro-Electric
Power Plant belong to the Area2.
In this study, two different combinations of controlling units are used
in the same system and comparisons of different system results like
Overshoot, Undershoot and Settling time are shown in Figure 7.7a, Figure
7.7b and Figure 7.7c respectively. Step change is applied both the areas and
is observed that the three observing parameters ΔF1, ΔF2 and ΔPtie come
back to zero within a few seconds. The performance parameters are listed
below in the Table 7.2.
Figure 7.7 represents the simulation snaps of 2 area hybrid power system.
Figure 7.7a, Figure 7.7b and Figure 7.7c show the frequency error of 1st area
(Δf1), frequency error of 2nd Area (Δf2) and tie-bar power error between
these two areas. In each of the three graphs response of the controllers are
denoted using different columns as illustrated in legend. For all cases, blue
identifies IT2FLC based FOPID controlling of the system, green one rep-
resents the responses obtained after controlling by IT2FLC based PID con-
troller and pink one is the responses obtained using PID controller only.
×10–3
Frequency Error (puHz)

0.01 (a) 5 (b)
0.005
0
0
–0.005 ×10–4 –5
×10–4
–0.01 5
0 –10 0
–5 : IT2FLC-FOPID –5 : IT2FLC-FOPID
–0.015 –10 : IT2FLC-PID : IT2FLC-PID
–15 –10
–0.02 19 20 21 : PID –15 20 22 : PID
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
Time (s) Time (s)
×10–3
Tie-Bar Power Error (puMw)
2 (c)
0
–2 ×10–4
5
–4 0
–5 : IT2FLC-FOPID
–6 : IT2FLC-PID
16 18 20 22 : PID
–8
0 10 20 30 40 50 60 70
Time (s)
Figure 7.7 Two area hybrid system controlled by PID, IT2FLC based PID and IT2FLC
based FOPID controllers. (a) Frequency error in 1st area, (b) Frequency error in 2nd area,
(c) Tie-bar power error between two areas.
The detail parameters are scripted in Table 7.2 where it has been found that
type-2 fuzzy can control the errors of the power system in a very efficient
way. According to the overshoot (OS) condition, the combination of type-2
fuzzy and PID shows better responses compared to FOPID for Δf1(IT2FLC-
PID: 0.0007962Hz, IT2FLC-FOPID: 0.0009518 Hz), Δf2(IT2FLC-PID:
0.0006668 Hz, IT2FLC-FOPID: 0.0007523 Hz) and for ΔPtie (IT2FLC-
PID: 0.00056puMW, IT2FLC-FOPID: 0.0007111puMW). Similar effect
has been seen in Undershoot (US) (IT2FLC-PID: 0.0007962puMW,
IT2FLC-FOPID: 0.0009518puMW) for Δf1 and for ΔPtie (IT2FLC-PID:
-0.004936 Hz, IT2FLC-FOPID: -0.005139 Hz) but IT2FLC-FOPID gives
better response of US for Δf2 only. (IT2FLC-PID: -0.006460 Hz, IT2FLC-
FOPID: -0.006403 Hz). From the view of the Settling time (ST) IT2FLC-
FOPID produces better result as it gets back the error exactly to zero after
few seconds after (Δf1: 20.56 Hz, Δf2: 20.12 Hz, ΔPtie: 20.83puMW) but
IT2FLC-PID produces a minimum constant error that is very close to zero
with a faster settlement (Δf1: 15.61 Hz, Δf2: 13.77 Hz, ΔPtie: 14.08puMW).
Both the IT2FLC-PID and IT2FLC-FOPID controllers produce satisfac-
tory controlling operations in the power system network than the result
obtained with only PID controller.
CASO gives optimized values for different controllers used in the simu-
lation. Table 7.3 shows the optimized parameters for both PID and FOPID
controller without disturbances.
Table 7.3 CASO tuned PID and FOPID

controller parameters.
Parameters PID FOPID
Kp1 0.01 5.845
Kp2 0.01 0.01
Kd1 0.46773 5.856
Kd2 0.92378 0.01
KI1 3.6593 3.6586
KI2 0.01 0.01
μ1 -- 0.7
μ2 -- 0.7
λ1 -- 0.965
λ2 -- 0.965
7.6.2 With Disturbances

In this case, a sinusoidal disturbance as ΔPD = |sin(t)| is applied in both
the areas of the interconnected power system. Frequency error and tie-
bar error of the closed-loop power system with IT2FLC-PID and IT2FLC-
FOPID controller in area 1 and area 2 are illustrated in Figure 7.8. Results
obtained from the power system network controlled by those two control-
lers are compared with the results obtained after controlling of the same
network system using only PID controller are illustrated in Figure 7.8.
Results obtained after successful simulation of the system with sinu-
soidal disturbances are tabulated in the Table 7.4. It is observed that con-
ventional PID controller does not produce stable output when sinusoidal
disturbances are introduced in the system. Fuzzy type 2 enabled PID con-
troller works efficiently to make the system errors back to zero. IT2FLC-
FOPID also performs successfully and controlling to nullify the errors. In
comparison with the responses made without disturbances, the settling
time is larger in the performance made by the system when disturbances
are present with it.
Table 7.4 shows an infinite settling time of each error for PID controlling
action. IT2FLC-PID produces OS for Δf1 = 0.00538 Hz, Δf2 = 0.00425
Hz, ΔPtie = 0.00211 puMW which are slightly higher than the responses
0.02

(a) 0.01 (b)
0.01 0.005
0
0
–0.005
–0.01
–0.01
–0.02 : IT2FLC-FOPID –0.015 : IT2FLC-FOPID
: IT2FLC-PID : IT2FLC-PID
–0.03 : PID –0.02 : PID
0 10 20 30 40 50 60 70 0 10 20 30 40 50 60 70
Time (s) Time (s)
×10–3
5
Tie-Bar Power Error (puMw)
(c)
0
–5
: IT2FLC-FOPID
: IT2FLC-PID
: PID
–10
0 10 20 30 40 50 60 70
Time (s)
Figure 7.8 Two area hybrid system controlled by PID, IT2FLC based PID and IT2FLC
based FOPID controllers with disturbance (a) Frequency error in 1st Area, (b) Frequency
error in 2nd area, (c) Tie-bar power error between two areas.
obtained without disturbances. Similar effect has been observed for other
parameters of the system: US for Δf1 = -0.0191 Hz, Δf2 = -0.0149 Hz,
ΔPtie = -0.00742 puMW and ST for Δf1 = 34.45s, Δf2 = 33.81s, ΔPtie =
Table 7.4 Performance results of 2 area hybrid system using IT2FLC based PID
and FOPID controllers with sinusoidal disturbances.
Two area test system (with sinusoidal disturbances)
Functions Parameters PID IT2FLC-PID IT2FLC-FOPID
Δf1(Hz) OS 0.0176 0.00538 0.0077
US -0.0282 -0.0191 -0.0197
ST -- 34.45 34.47
Δf2(Hz) OS 0.0093 0.00425 0.0049
US -0.0218 -0.0149 -0.0158
ST -- 33.81 33.82
ΔPtie(puMW) OS 0.0046 0.00211 0.00223
US -0.0089 -0.00742 -0.00736
ST -- 34.08 34.09
34.08s. Settling time has been delayed in this case due to the presence of
disturbances in the system, though can easily be nullified. Type-2 Fuzzy
based FOPID also produces the stable responses for all cases: OS for Δf1 =
0.0077 Hz, Δf2 = 0.0049 Hz, ΔPtie = 0.00223 puMW, US for Δf1 = -0.0197
Hz, Δf2 = -0.0158 Hz, ΔPtie = -0.00736 puMW, STs are Δf1 = 34.47s, Δf2 =
33.82s, ΔPtie = 34.09s.
7.7 Conclusion
This chapter provides an analysis of the application of Type-2 Fuzzy
controller in the field of inter-connected hybrid power system network.
Initially this type of power system network was controlled by only PID
controller. From the results obtained as can be shown from Figure 7.8, it
is observed that a significant level of oscillations in the response graph
though there are no applied disturbances. With the artful introduction of
Type-2 Fuzzy logic in association with PID controller, the oscillations are
drastically reduced making the system more stabilized. IT2FLC-FOPID
also performs efficiently in different conditions and outperforms the oper-
ation of IT2FLC-PID in certain cases as tabulated in Table 7.2. The simi-
lar controller works suitably for the system with injected disturbances in
it and making the system stable in respect to the desired parameters as
has been depicted in Table 7.4. The application of Type-2 Fuzzy controller
along with the PID/FOPID, thus can be successfully applied in the field of
hybrid multi area power system.
Appendix
• For Thermal Power Plant [19]:
Tg 0 08s Tt 0 3s Kr 0 3 Tr 10s KT 0 33
• For Solar Power Unit [21]:
a 900 b 18 c 50 d 50
• For Hydro-Electric Power Plant [10]:
TRH 48 7s TR 5s TGH 0 513s Tw 1s KT 0 99
• For Wind Power Unit [22]:
ωT = 0.5
• For Other Parameters [10]:
Hz 2.4Hz puMW
K PSi 120 , TPSi 20s, R1 R 2 , 1 2 0.425
puMW pu Hz
References
1. Kristov, L., The bottom-up (R)evolution of the electric power system: The
pathway to the integrated-decentralized system. IEEE Power Energy Mag.,
17, 2, 42–49, March-April 2019.
2. Xu, S., Xue, Y., Chang, L., Review of power system support functions for
inverter-based distributed energy resources-standards, control algorithms,
and trends. IEEE Open J. Power Electron., 2, 88–105, 2021.
3. Shakibjoo, A.D., Moradzadeh, M., Din, S.U., Mohammadzadeh, A., Mosavi,
A.H., Vandevelde, L., Optimized type-2 fuzzy frequency control for multi-
area power systems. IEEE Access, 10, 6989–7002, 2022.
4. Panda, A.M., Automatic generation control of multi area interconnected
power system considering non-linearity due to governor dead band. Arch.
Control Sci., 7, 3-4, 285–99, 1998.
5. Hidalgo, D., Castillo, O., Melin, P., Type-l and type-2 fuzzy inference systems
as integration methods in modular neural networks for multimodal biome-
try and its optimization with genetic algorithms. J. Inf. Sci., 179, 13, 2123–45,
2009, [June].
6. Zribi, AI-Rashed, M., Alrifai, M., Adaptive decentralized load frequency
control of multi-area power systems. Int. J. Electr. Power Energy Syst., 27, 8,
575–83, 2005.
7. Gayadhar, P., Sidhartha, P., Ardil, C., Hybrid neuro fuzzy approach for auto-
matic generation control of two-area interconnected power system. Int. J.
Comput. Intell., 5, 1, 80–4, 2009.
8. Sahu, Kumar, R., Panda, S., Chandra Sekhar, G.T., A novel hybrid PSO-PS
optimized fuzzy PI controller for AGC in multi area interconnected power
systems. Int. J. Electr. Power Energy Syst., 64, 880–893, 2015.
9. Rodríguez-Abreo, O. et al., Self-tuning neural network PID with dynamic
response control. IEEE Access, 9, 65206–65215, 2021.
10. Arya, Y., New optimized fuzzy FOPI-FOPD controller for automatic genera-
tion control of electric power systems, 356, 11, 5611–5629, 2019.
11. Sudha, K.R. and Santhi, R.V., Robust decentralized load frequency control
of interconnected power system with generation rate constraint using type-2
fuzzy approach. Int. J. Electr. Power Energy Syst., 33, 3, 699–707, March 2011.
12. Kocaarslan, et al., Application of fuzzy reasoning to load? Frequency control
in two area power system. 1999 European Control Conference.
13. Shakibjoo, A.D. et al., Load frequency control for multi-area power systems:
A new type-2 fuzzy approach based on Levenberg–Marquardt algorithm.
ISA, 121, 40–52, 2021, Published by Elsevier.
14. Venkatraman, E., Velusami, S., Mohamed Thameem Ansari, M., Ganapathy,
S., Design of type- II fuzzy logic load frequency controller for two area ther-
mal reheat power systems. 2012 International Conference on Computing,
Electronics and Electrical Technologies (ICCEET), pp. 180–185, 2012.
15. Nayak, Ranjan, J., Shaw, B., Sahu, B.K., Implementation of hybrid SSA–SA
based three-degree-of-freedom fractional-order PID controller for AGC
of a two-area power system integrated with small hydro plants. IET Gener.
Transm. Distrib., 14, 13, 2430–2440, 2020.
16. Jurasz, J. et al., A review on the complementarity of renewable energy
sources: Concept, metrics, application and future research directions. Sol.
Energy, 195, 703–724, 2020.
17. Sanki, P. and Basu, M., New approach in two-area interconnected AGC
including various renewable energy sources using PSO. Turk. J. Elec. Eng.
Comp. Sci., 26, 3, 1491–1504, 2018.
18. Chaine, S. and Tripathy, M., Performance of CSA optimized controllers
of DFIGs and AGC to improve frequency regulation of a wind integrated
hydrothermal power system. Alexandria Eng. J., 58, 2, 579–590, 2019.
19. Hakimuddin, N., Nasiruddin, I., Bhatti, T.S., Generation-based automatic
generation control with multisources power system using bacterial foraging
algorithm. Eng. Rep., 2, e12191, 2020.
20. Biswas, S., Roy, P.K., Chatterjee, K., Development of MADB of P-I controller
using LMI technique in a renewable energy based AGC system and study its
application in a deregulated environment including energy storage device.
Optim. Control Appl. Methods, 44, 1–26, 2021.
21. Biswas, S., Roy, P.K., Chatterjee, K., Renewable energy-based multi-source
system under deregulated environment using COKHA algorithm. IETE J.
Res., 1–19, 2021.
22. Bhongade, S., Tyagi, B., Gupta, H.O., Multi-area automatic generation con-
trol scheme including renewable energy sources, TELKOMNIKA Indonesian
Journal of Electrical Engineering (IJEE), 12, 7, 5052-5070, 2014.
23. Panwar, A. and Chahar, S., Automatic load frequency control of three area
power system using artificial intelligence, IEEE, 2016, 978-1-5090-3411-6/16
$31.00.
24. Gozde, H., Cengiz Taplamacioglu, M., Kocaarslan, I., Comparative perfor-
mance analysis of Artificial Bee Colony algorithm in automatic generation
control for interconnected reheat thermal power system. Int. J. Electr. Power
Energy Syst., 42, 1, 167–178, 2012.
25. Podlubny, I., Fractional-order systems and PI/sup/spl lambda//D/sup/spl
mu//-controllers. IEEE Trans. Automat. Control, 44, 1, 208–214, 1999.
26. Biswas, A., Das, S., Abraham, A., Dasgupta, S., Design of fractional-order
PIλ Dμ controllers with an improved differential evolution. Eng. Appl. Artif.
Intell., 22, 343–50, 2009.
27. Zamani, M., Karimi-Ghartemani, M., Sadati, N., Parniani, M., Design of
fractional order PID controller for an AVR using particle swarm optimiza-
tion. Control Eng. Pract., 17, 1380–7, 2009.
28. Monje, C.A., Vinagre, B.M., Calderon, A.J., Feliu, V., Chen, Y.Q., On frac-
tional PI controllers: Some tuning rules for robustness to plant uncertainties.
Nonlinear Dyn., 38, 369–81, 2004.
29. Barbosa, R.S., Tenreiro Machado, J.A., Ferreira, I.M., Tuning of PID con-
trollers based on Bode’s ideal transfer function. Nonlinear Dyn., 38, 305–21,
2004.
30. Zadeh, L.A., The concept of a linguistic variable and its application to approx-
imate reasoning-I. Inf. Sci., 8, 199–249, 1975.
31. Hagras, H., Type-2 FLCs: A new generation of fuzzy controllers. IEEE
Comput. Intell. Mag., 2, 1, 30–43, 2007.
32. Sudha, K.R. and Vijaya Santhi, R., Robust decentralized load frequency con-
trol of interconnected power system with Generation Rate Constraint using
type2fuzzy approach. Int. J. Electr. Power Energy Syst., 33, 3, 699–707, March
2011.
33. Mendel, J.M., Uncertain rule-based fuzzy logic: Introduction and new direc-
tions, Prentice Hall, USA, 2000.
34. Dokht Shakibjoo, A., Moradzadeh, M., Moussavi, S.Z. et al., Load frequency
control for multi-area power systems: A new type-2 fuzzy approach based on
Levenberg–Marquardt algorithm. ISA Trans., 121, 40–52, 2021, https://doi.
org/10.1016/j.isatra.2021.03.044.
35. Wilamowski, Bogdan, M., David Irwin, J. (Eds.), Intelligent systems, CRC
Press, Boca Raton, 2018.
8
Alzheimer’s Detection and
Classification Using Fine-Tuned
Convolutional Neural Network
Anooja Ali1*, Sarvamangala D. R.2, Meenakshi Sundaram A.1 and Rashmi C.2
1
School of Computer Science and Engineering, REVA University, Bengaluru, India
2
School of Computing and Information Technology, REVA University,
Bengaluru, India
Abstract
Alzheimer’s disease (AD) is a neurocognitive disorder and it evolves into the
death of nerve cells. After the age of 60, the risk of developing the illness dou-
bles every five years, with estimates that by 2050, the number will have risen to
135 million. Brain structural image with magnetic resonance imaging (MRI)
has been extensively utilized to recognize AD as it can detect morphometric
variations and cerebral congenital malformations. Convolutional neural net-
works (CNNs) are extensively used for image receptions and analysis because
of their capacity to handle enormous amounts of unstructured data and retrieve
significant characteristics automatically. A new approach involving pretrained
CNN model, VGG16 with fine tuning has been proposed for automatic detec-
tion and classification of brain MRI images for AD. The results show that the
performance of the proposed modeling terms of accuracy, f1-score, recall and
precision is above 90%.
Keywords: ADNet, Alzheimer’s disease, dementia, transfer learning
8.1 Introduction
The human brain is a sophisticated organ with millions of neurons that
communicate data via electrical and chemical impulses. Alzheimer’s
125
disease (AD) involves brain tissue deterioration and may progress to nerve
cell dying, resulting in memory loss and interruptions in basic human func-
tions. AD initially affects hippocampus region of brain, and later spread to
cerebral cortex. Hippocampus is the portion of the brain that is in com-
mand of episodic and visual memory. It also serves as a conduit between
body and minds. The loss of cells and damage to synapses and neuron will
occurs when the hippocampus shrinks. As a result, neurons can no longer
connect through synapses. In AD, an abnormal level of beta–amyloid pro-
tein cause cell damage, thereby damaging the neurons and thereby causing
brain atrophy in the final stage.
Normal brain and AD brain with changes in cerebral cortex and hippo-
campus have been depicted in Figures 8.1 and 8.2, respectively.
AD consist of three phases, preclinical, mild cognitive, and dementia.
Preclinical has unnoticeable change with high risk of AD, mild cogni-
tive has mere symptoms and interference in daily routines. Presently, AD
affects around 46.8 million people globally. By the year 2050, it is expected
that the number of Alzheimer’s sufferers would have risen to 300 mil-
lion and nearly 60% of dementia cases are due to AD [1]. If AD can be
Cerebral cortex
Healthy neurons
Normal brain Hippocampus
Figure 8.1 Normal brain.
Shrinkage of
cerebral cortex Enlarged ventricles
Tau neurofibrillary tangles

Shrinkage of
Alzheimer disease hippocampus
brain Aβ plaques
Figure 8.2 Alzheimer disease brain.

Detect and Classify Alzheimer’s Using CNN 127
detected earlier, then medication or therapy can avoid brain cell damage.
An early investigation can control the rate of progression of AD. Loss of
memory, anxiety, and speech problems, problems in reading, and writing
are involved in AD. All of these make AD patients nervous and hostile. In
dementia, the levels of AD are visible in patient. The four stages of demen-
tia include moderate dementia, mild dementia, very mild dementia, and
no dementia. The different stages of dementia are depicted in Figure 8.3.
Early researchers found that the social environment, particularly social
interactions, can have an impact on behavior and mental health of humans.
Researchers examined the link between loneliness and the chance of fall-
ing into AD, and it is found that lonely people had a greater risk of having
the disease than others. Dementia develops social isolation of patients [2].
When people begin to have difficulties in remembering things, it is recom-
mended that they seek medical care for a diagnosis and potential therapy.
For early-stage identification, many methodologies, such as biomark-
ers, brain imaging, cerebrospinal examinations, and so on, were used.
Diagnostic biomarkers and tests based on amyloid give inadequate insight
on disease process and unable to detect patients with the illness before con-
siderable amyloid-beta deposition in the brain occurs. Examining genetic
code or deoxyribonucleic acid (DNA) has the benefit of revealing a variety
of human disorders [3]. Uncovering and medicament of AD by DNA anal-
ysis techniques like genome sequencing and nucleotide conglomerations in
wet lab research is often time consuming. Genome sequencing approaches
with support vector machines (SVM) were inaccurate in detecting cancer
due to the predominance of wrong positives.
(a) (b) (c) (d)
Figure 8.3 (a) Nondemented. (b) Moderately demented. (c) Mild demented. (d) Very
mild demented.
Biomarkers are useful in every step of a patient’s treatment [4]. The

detection of biomarkers for AD with decision tree did not provide any
intriguing instance on how data mining tools might assist genome-wide
association studies for AD models [5]. Automated systems are not prone
to human mistake, they are more reliable than human evaluations and can
be employed in smart healthcare systems. Hence, artificial intelligence
and pattern recognition techniques can guide the doctors to forecast ill-
ness development without human interventions [6]. By learning compli-
cated patterns from the data, machine learning (ML) allows multivariable
records to be fitted to a model [7].
The classical ML classifiers like Naive Bayesian [8], Gaussian mixture
(GM), and decision tree [9] were used to diagnose AD. ML has been used
in several researches to construct classifiers that can distinguish between
AD patients and healthy controls. The summarized linear model classified
the different stages of AD with an accuracy of 88.24%. Feature vectors are
often required to train the supervised models. Extracting these vectors is
often found to be tedious and inaccurate [10]. Moreover, these models face
difficulty of learning deep, and the important aspects of clinical data in
detail remains a barrier for typical ML algorithms.
In ML, data must be adequately processed in analysis before evaluation.
This includes deleting undesired attributes, null and missing values, and
duplicated records. Extraction of discriminatory features is required for
ML approaches, and these characteristics are typically unknown, making
it a difficult undertaking, particularly for image-based applications. Few
researchers identified that dementia frequently coexist in the brain and
share some changeable risk factors, it has been hypothesized that AD can
be avoided by focusing on modifiable vascular risk factors [11].
To address these challenges, raw imaging data can be subjected to
effective processing through deep learning (DL) strategies. Recent studies
state that DL approaches have attained state-of-the-art efficiency in sev-
eral areas, including medical image processing. DL does not involve pre-
processing and is efficient to infer an ideal description from MRI images
without the need for feature extraction, resulting in a more objective and
bias-free procedure [12]. DL has recently been used to the AD neuroimag-
ing initiative with great success (ADNI).
CNN have the ability to handle large volumes of unstructured inputs
[13]. Evidences shows that medical characteristics, in combination with
statistical and DL approaches, succeeds in the categorization of AD. With
the invent of DL, researchers can extract characteristics automatically from
images without the involvement of human specialists. These deep models
produced notable findings in the fields of pathology, lung, retina, brain,
heart, breast, abdomen, bone, and others for tissue, organ, and substruc-
ture segmentation, cancer detection, and classification. The DL-based
CNN model, facilitates computer-aided affirmation on the morphological
aspect of brain and/or hippocampus inclined on the textural specifications.
Multiple CNN with segmentation algorithm is a recursive Otsu thresh-
olding procedure is used to segment an MRI brain image and to detect
the level of brain degradation. The consideration was Alexnet architecture
with Dropout and Batch Normalization. The overall system performed
well, but the accuracy of few individual CNN went below 80% [14]. CNN
frameworks (toolkits) make deep learning approaches more efficient.
Researchers have suggested a variety of CNN architectures depending on
the task at hand [15].
The most promising field for detecting Alzheimer’s disease is brain
imaging. Imaging is a well-established technology in drug development,
and it’s becoming more used in clinical trials as an informative criterion,
a safety indicator, and an evaluation measure. Functional, structural, and
molecular imagings are the major technologies in AD detection. MRI anal-
ysis can insight the long-term structural alterations induced by the dis-
ease along with the disease’s development stages [16]. When compared to
CT scans, MRI allows for a better view of the brain’s structural parts. By
detecting typical signatures of brain structure and function changes, imag-
ing provides positive evidence for clinical assessment of AD patients.

The various methods for diagnosis of AD are discussed in literature, com-
prising of biomarker-based approaches, ML- and DL-based techniques.
Various forms of CNN architecture and activation functions are also
analyzed.
The biochemical processes including metabolism of amyloid protein,
tau protein phosphorylation, dysfunction of mitochondria, decreased
energy, regulatory issues of membrane lipids and neurotransmitter path-
ways are affected by AD. A biomarker would be capable to reflect these
changes. Hence, metabolomics analysis (MA) through biomarker would
early diagnosis of AD [17]. MA shows transcriptomic, genomic, and pro-
teomic alterations. Transcription factors, biomarkers, and protein are suit-
able for phenotype research, and they are biochemical reaction executors.
Amyloid-based biomarkers were involved in the early detection of AD
with the constraint of limited knowledge regarding the disease’s patho-
logical aetiology and causes [18]. So, there arise the necessity of potential
biomarkers that can contribute to a better knowledge of the condition, as

well as the identification of patients in the early stages of the disease and the
development of new therapies [19]. However, a biomarker must always be
validated before it can be used in clinical trials and requires several studies
in broad and different sections of the population to show that it accurately
and reliably signals the presence of illness.
In this section, we discuss the various ML-based approaches for AD
detection. ML is currently being superseded by DL in most classification
applications, despite its superior performance. The main distinction among
ML and DL is the technique for feature extraction. DL’s classification per-
formance is considerably superior than ML’s classification, which depends
on crafted features, and the features extracted from many non-linear hid-
den layers. DL is a set of approaches based on autonomous feature engi-
neering and driven by neural data [20].
The commonly used ML-based classifiers include DT, SVM, KNN, and
so on. ML prediction model for early AD diagnosis based on patient neu-
ropathological abnormalities was not suitable for clinical use with accuracy
of 77%, but it could be a step toward precision treatment in AD. However,
the selection of features based on their relative relevance reduces the size
of feature space.
Alzheimer’s disease neuroimaging initiative (ADNI) portal’s blood
proteomic data for their work to identify non-amyloid biomarker panels
and developed ML models for the early detection of AD [21]. They deter-
mined a combination of six non-amyloid proteins, alpha-2 macro glob-
ulin (A2M), apolopoprotein E (ApoE), brain natriuretic peptide (BNP),
Eotaxin3 (Eot3), receptor1 for advanced glycosylation end (RAGE) and
serum glutamic oxaloacetic transaminase (SGOT) in the blood plasma as
prominently responsible for the early detection of AD.
The SVM model classifies people into prodromal AD (PAD), AD
dementia (ADD) and normal controls (HC) [22]. By learning compli-
cated patterns from data, ML allows multivariable statistics to be fitted to
a model. Brute force search is performed with the preselection of feature
subset and feature panel with kernelized SVM could discriminate ADD
and HC. The SVM-based model obtained specificity and sensitivity of 0.70
and 0.80 respectively. Studies with SVM classified dementia and healthy
cases with area under operating curve of 0.78.
Later, for multiclass classifications, CNN-based approaches were found
to be more promising. This is because the proliferation of image content
on the web has the potential to develop more comprehensive and reliable
indexing, retrieval, organization, and interaction models and algorithms
for images and multimedia data. Imagenet, a hierarchial database covers
around images of 5,247 categories with ontology with wordnet as a back-

bone [23]. For document recognition, LeCun et al., created the CNN
framework (LeNet). The system was created for a recognition of handwrit-
ten character detection and tested on the MNIST standard dataset, achiev-
ing an error margin of 0.8%. The system was later described as the core
method for Graph Transformer systems [24].
In 2012, AlexNet, an improved version of LeNet that outperformed
standard machine learning approaches, won the Imagenet competition
[25]. Different forms of CNN architectures, including as LeNet, AlexNet,
ResNet [26], InceptionNet [27], Xception [28] have been suggested. These
CNN architectures became the custom models for AD diagnosis. The goal
of image classification is to categorize an instance into different groups. In
order to effectively employ internal representations, architectural design
must take into account factors, such as features fusing and attentiveness
mechanisms, in addition to network scale.
Due to the residual and dense connection patterns, approaches that con-
centrate on feature aggregation, like the ResNet and DenseNet, have had
considerable success. To make training a deep network with fewer parame-
ters easier, ResNet initially introduces identity mapping. However, because
many layers in ResNet are redundant, only one of the representations of
each layer is reused by a subsequent layer, lowering the learning efficiency
dramatically. Various CNN frameworks have been created and utilized for
classification as a result of the fast development and efficiency of CNNs.
Xian et al. employed feature pyramid networks (FPN) to boost the
detection of tiny objects via a top-down hierarchical fashion and lateral
connections using a Faster R-CNN, SSD, and YOLOv3 architectures [29].
The work was carried out without applying any MRI preprocessing tech-
nique to the ADNI and OASIS data sets which otherwise would tamper
the MRI slices, increasing the diagnosis complexity [30]. Authors sug-
gest a ternary intermediate classification between AD and NC as Mild
Cognitive Impairment (MCI). These DL-based object detection methods
and the algorithm achieved detection accuracy of 0.938 for YOLOv3, 0.922
for SSD, and 0.928 for Faster R-CNN, all surpassing the 0.75 Intersection
Over Union (IoU) criterion, after being trained on 1000 raw, unprocessed
individuals’ data from the ADNI database. For any inference input images
below 6002, faster R-CNN failed to give any localization and classification
input.
Few researchers employed layer wise tuning by gathering the data via
an open-access series of imaging studies (OASIS). In order to determine
the impact of each section of layers on classification results, the authors
performed shallow tuning and fine tuning of three pretrained models
(Alexnet, Googlenet, and Resnet50) in a number of layers and detected

NC and AD patients [31]. After collecting all of the slices, each model was
trained, validated, and tested in a 6:2:2 ratio on a random selection basis.
It was evaluated that pretrained CNN models with layer wise tuning works
as good as scratch trained models.
The main building elements of CNN are convolutional layers, activation
functions and pooling, and then the fully-connected layers as represented
in Figure 8.4. A convolutional layer is made up of many filters that perform
convolutional action. Convolution Layers are employed in the first few lay-
ers of CNN for feature extraction. Filters are used with image data, which
is essentially grids or matrix, for feature extraction. A filtered image is cre-
ated by passing it through a filter, and the technique is known as convolu-
tion. To produce the feature map, convolution of all filters with input image
occurs. Here, dot product operation occurs between inputs and filters,
and the result of each convolutional layer becomes simpler. Hence, fewer
shared parameters are generated instead of a fully connected neural net-
work, which minimizes the model’s complexity and improves its efficiency.
The tensors are then max pooled, and output is subsampled to produce
a smaller image. The number of parameters decreases as a result of the
aforementioned approach, lowering the complexity. Pooling layer pro-
duces an abstract representation of features by decreasing the spatial size
and by considering the region of overlap with kernel. Hence, the downs-
ampling of convolved features results in dimensionality reduction and
thereby, decreasing the computational power [32]. Convolution layer learn
the parameters and pooling layer applies function and partition the input
into rectangular patches.
This section gives an overview on activation functions. Non-linearity is
introduced to the model through activation functions, which lets model
Fully
Connected
Convolution
Pooling Output
Input
Figure 8.4 CNN architecture.

Table 8.1 Summary of various activation functions.

Activation function Specification
Binary Used with binary classifiers
Linear Proportional to input
Sigmoid Nonlinear activation with values ranging from 0 to 1
Tanh Symmetric on origin with values ranging from -1 to +1
ReLU Non-linear, used in DL domain, Neurons will not be
activated in case of negative input.
Leaky ReLU For negative input, ReLU function is redefined to a
smaller linear component of input.
Softmax Multiclass classification, Combination of sigmoids
to train functional mappings between response variables and input. The

various activation function and their usage are summarized in Table 8.1.
The proposed methodology is a deep learning-based system comprises
of feature extraction and multi class classifier, which classify the input MRI
data into four clinically confirmed conditions of Alzheimer’s disease, mild
demented, moderate demented, non-demented, and very mild demented.
8.3 Methodology
The proposed methodology for classification of dementia is illustrated in
Figure 8.5.
AD Images Preprocess Data Augmentation
No Dementia
Moderate
VGG16
Mild Dementia
Very Mild
Figure 8.5 Proposed architecture ADNET.

8.3.1 Dataset
The dataset was taken from kaggle. The dataset consisted of train and test
directories. The train directory consisted of 5,121 images of which 2,560
images were nondemented, 1,792 images were very mild demented, 717
images were mild demented and 52 images were moderate demented. The
test directory consisted of 1,279 images of which 640 images were non-
demented, 448 images were very mild demented, 179 images were mild
demented and 12 images were moderate demented.
8.3.2 Pre-Processing
The images were pre-processed by histogram-based intensity standard-
ization. This was to address the variabilities in intensities of the pixels.
Moreover, the intensity values were also brought in the range of 0 to 255.
Since the number of training samples was limited, data augmentation was
performed to increase the training samples. The images were randomly
rotated in the range of 0 degree to 30 degree and were also randomly
flipped horizontally. Variations to brightness was done by +10% to −10%
and the images were zoomed in and out in the range of +1 to −1%. To
make the training set more robust to noise, random jitter was added to
the images. As the training and test images were imbalanced, synthetic
minority over-sampling technique (SMOTE) was applied. SMOTE is a
balanced statistical strategy for expanding overall number of examples
in the dataset. It creates new instances using the current minority cases
provided as input. It seeks to balance the class distribution by recreating
minority class cases at random.
8.4 Implementation and Results

The proposed model ADNet was implemented using ImageNet pretrained
CNN model VGGNet16. In order to make it easier to train deep net-
works with fewer parameters, ResNet first introduces identity mapping.
However, because many of the layers in ResNet are redundant, only one
of the representations of each layer is reused by a subsequent layer, signifi-
cantly decreasing the learning efficiency. DenseNet suggests dense feature
reuse to get around this restriction. Despite producing good performance,
DenseNet has a very high density to skip connections, which increases
complexity and hinders convergence.
A pretrained model was used, as it had already learnt the basic features
of the image. It was fine-tuned on AD data set. For fine tuning the model,
the last layer of VGGNet16 was replaced by a fully connected layer hav-
ing four neurons with softmax as the activation function. The optimizer
used was adam with learning rate of 0.1. L2 regularizer with dropout of 0.5
was used to overcome over fitting. The model was trained for 1000 epochs
with a batch size of 32. The performance of the model was evaluated using
accuracy, precision, recall, f1-score, and AUC. Visual modeling for perfor-
mance just in time is another central instructional approach for effective
mobile learning. This instructional strategy often involves reducing extra-
neous cognitive load on the spot, providing worked examples right before
performing a critical task, and visualizing learning object in three dimen-
sional presentations.
The initial implementation was done without SMOTE. Due to imbal-
anced data set, the performance of the model was not very efficient. The
model arrived at an overall accuracy of 0.57. The other performances mea-
sures obtained are depicted in Table 8.2.
SMOTE was applied to the dataset to overcome imbalance. The obtained
images were used to train and test the ADNet. ADNet have efficient fea-
ture reuse capabilities of Resnet and Densenet. Application of SMOTE
enhanced the performance of the model, as the model had learnt on a
Table 8.2 Performance metrics of ADNet without SMOTE.

Performance metric Precision Recall F1-Score Support
Non-demented 0.85 0.22 0.35 179
Very mild demented 0.33 0.17 0.22 12
Mild demented 0.82 0.43 0.57 640
Moderate demented 0.46 0.91 0.61 448
Accuracy 0.64 0.76 0.57 1279

Micro average 0.61 0.43 0.44 1279
Macro average 0.62 0.44 0.46 1279
Weighted average 0.59 0.45 0.42 1279
Samples average 0.61 0.43 0.44 1279
Model acc
train
0.9 val
0.8
0.7
acc
0.6
0.5
0.4
0 20 40 60 80 100
Epochs
Model auc
1.00
0.95
0.90
0.85
auc
0.80
0.75
0.70
train
val
0.65
0 20 40 60 80 100
Epochs
Model loss
1.4 train
val
1.2
1.0
loss
0.8
0.6
0.4
0.2
0 20 40 60 80 100
Epochs
Figure 8.6 Training and validation accuracy, AUC and loss for 100 epochs.
balanced dataset. The ADNet model was able achieve an accuracy of over
90%. The training and validation accuracy, and AUC loss obtained are
above 90%, and loss decreased with increase in the number of epochs. The
same is illustrated in Figure 8.6. The intensity of feature reuse is automati-
cally taught and data-adaptive.
Table 8.3 Various performance metrics for ADNet.

Performance metric Precision Recall F1-score Support
Nondemented 0.94 0.95 0.94 639
Very mild demented 1 1 1 635
Mild demented 0.85 0.79 0.82 662
Moderate demented 0.78 0.83 0.81 624
Accuracy 0.89 0.89 0.905 2560
Microaverage 0.89 0.89 0.89 2560
Macroaverage 0.89 0.89 0.89 2560
Weighted average 0.89 0.89 0.89 2560
Samples average 0.89 0.89 0.89 2560
Alzheimer’s Disease Diagnosis
600
NonDemented 636 0 0 3
500
VeryMildDemented 0 635 0 0 400

Truth
300
MildDemented 13 0 592 57
200
100
ModerateDemented 14 1 44 565
0
NonDemented VeryMildDemented MildDemented ModerateDemented
Prediction
Figure 8.7 Confusion matrix of ADNet with SMOTE.

The values of various performance metrics obtained is depicted in

Table 8.3
The confusion matrix of ADNet obtained is depicted in Figure 8.7.
8.5 Conclusion
The research has been motivated by the wide success of CNN algorithms
to solve the classification of AD. The detection of various types of AD has
been treated as four class classification challenge. A popular CNN model
VGG16 is used to solve the challenge. Transfer learning with fine tuning
model is used to build the model.
A brief overview of AD, different types of AD, and CNN is provided.
The results reported in the chapter reveals that the CNN model ADNet
can properly handle classification of AD. The performance of the proposed
ADNet in terms of various performance metrics of accuracy, AUC score,
recall, precision and f1-score has been shown. All the performance metrics
are nearly around 90%, which denotes the efficiency of the proposed model.
The research can be taken in a number of different areas. To improve the
quality of outcomes and resource efficiency, variants and hybrids of CNN
models can be designed. CNN models can also be utilized to solve a num-
ber of unsolved medical image processing tasks.
References
1. Zhan, L., Zhou, J., Wang, Y., Jin, Y., Jahanshad, N., Prasad, G., Nir, T.M.,
Leonardo, C.D., Ye, J., Thompson, P.M., Comparison of nine tractography
algorithms for detecting abnormal structural brain networks in Alzheimer’s
disease. Front. Aging Neurosci., 7, Apr, 1–19, 2015, https://doi.org/10.3389/
fnagi.2015.00048.
2. Rashmi, C. and Kodabagi, M.M., A review on overlapping community detec-
tion methodologies. International Conference on Smart Technologies For
Smart Nation (SmartTechCon), pp. 1296–1300, 2017, https://ieeexplore.ieee.
org/document/8358576.
3. Ali, A., Viswanath, R., Patil, S.S., Venugopal, K.R., A review of aligners for
protein protein interaction networks. RTEICT 2017 - 2nd IEEE International
Conference on Recent Trends in Electronics, Information and Communication
Technology, Proceedings, 2018-Janua, pp. 1651–1655, 2017, https://doi.
org/10.1109/RTEICT.2017.8256879.
4. Blennow, K. and Zetterberg, H., Biomarkers for Alzheimer’s disease: Current

status and prospects for the future. J. Intern. Med., 284, 6, 643–663, 2018,
https://doi.org/10.1111/joim.12816.
5. Ali, A., Hulipalled, V.R., Patil, S.S., Abdulkader, R., DPEBic: Detecting essen-
tial proteins in gene expressions using encoding and biclustering algorithm.
J. Ambient Intell. Humaniz. Comput., 2021, 2011 Mar 11:1-8, https://doi.
org/10.1007/s12652-021-03036-9.
6. Ali, A., Hulipalled, V.R., Patil, S.S., Centrality measure analysis on pro-
tein interaction networks. 2020 IEEE International Conference on
Technology, Engineering, Management for Societal impact using Marketing,
Entrepreneurship and Talent (TEMSMET), IEEE, 2020.
7. Lama, R.K., Gwak, J., Park, J.S., Lee, S.W., Diagnosis of Alzheimer’s dis-
ease based on structural MRI images using a regularized extreme learning
machine and PCA features. J. Healthc. Eng., 2017, 1, 1–14, 2017, https://doi.
org/10.1155/2017/5485080.
8. Wei, W., Visweswaran, S., Cooper, G.F., The application of naive Bayes
model averaging to predict Alzheimer’s disease from genome-wide data.
J. Am. Med. Inf. Assoc., 18, 4, 370–375, 2011, https://doi.org/10.1136/
amiajnl-2011-000101.
9. Naganandhini, S. and Shanmugavadivu, P., Effective diagnosis of Alzheimer’s
disease using modified decision tree classifier. Proc. Comput. Sci., 165, 548–
555, 2019.
10. Ali, A., Hulipalled, V.R., Patil, S.S., A novel semantic similarity score for pro-
tein data analysis. Comput. Technol. Res. J., 1, 1, 1–4, 2022.
11. Sana, T. and Barber, P.A., Dementia risk and prevention by targeting mod-
ifiable vascular risk factors. J. Neurochem., 144, 565–581, 2018, https://doi.
org/10.1111/jnc.14132.
12. Liang, S. and Gu, Y., Computer-aided diagnosis of Alzheimer’s disease
through weak supervision deep learning framework with attention mech-
anism. Sensors (Switzerland), 21, 1, 1–15, 2021, https://doi.org/10.3390/
s21010220.
13. Islam, J. and Zhang, Y., Brain MRI analysis for Alzheimer’s disease diagnosis
using an ensemble system of deep convolutional neural networks. Brain Inf.,
5, 2, 2018.
14. Slim, A., Melouah, A., Layachi, S., Alzheimer’ s disease diagnosis using par-
allel convolutional neural networks. 2021 International Conference on Recent
Advances in Mathematics and Informatics (ICRAMI), 2021.
15. Sarvamangala, D.R. and Kulkarni, R.V., Grading of knee osteoarthritis using
convolutional neural networks. Neural Process. Lett., 53, 2985–3009, 2021,
https://doi.org/10.1007/s11063-021-10529-3.
16. Kavitha, C., Mani, V., Srividhya, S.R., Khalaf, O.I., Tavera Romero, C.A.,
Early-stage Alzheimer’s disease prediction using machine learning mod-
els. Front. Public Health, 10, March, 1–13, 2022, https://doi.org/10.3389/
fpubh.2022.853294.
17. Wang, Y., Sun, Y., Luo, Y., Peng, D., Li, X., Biomarkers for the clinical diagno-
sis of Alzheimer’ s disease: Metabolomics analysis of brain tissue and blood.
Front. Pharmocol., 12, July, 1–14, 2021.
18. Kametani, F. and Hasegawa, M., Reconsideration of amyloid hypothesis and
tau hypothesis in Alzheimer’s disease. Front. Neurosci., 12, January, 2018,
https://doi.org/10.3389/fnins.2018.00025.
19. Ali, A., Hulipalled, V.R., Patil, S.S., Abdulkader, R., Alignment of protein
interaction networks and disease prediction: A survey. Int. J. Adv. Trends
Comput. Sci. Eng., 8, 4, 2019, https://doi.org/10.30534/ijatcse/2019/42842019.
20. Khan, P., Kader, M.F., Islam, S.M.R., Rahman, A.B., Kamal, M.S., Toha, M.U.,
Kwak, K.S., Machine learning and deep learning approaches for brain dis-
ease diagnosis: Principles and recent advances. IEEE Access, 9, 37622–37655,
2021, https://doi.org/10.1109/ACCESS.2021.3062484.
21. Petersen, R.C., Aisen, P.S., Beckett, L.A., Donohue, M.C., Gamst, A.C.,
Harvey, D.J., Jack, C.R., Jagust, W.J., Shaw, L.M., Toga, A.W., Trojanowski,
J.Q., Weiner, M.W., Alzheimer’s disease neuroimaging initiative (ADNI):
Clinical characterization. Neurology, 74, 3, 201–209, 2010.
22. Eke, C.S., Jammeh, E., Li, X., Carroll, C., Pearson, S., Ifeachor, E., Early detec-
tion of Alzheimer’ s disease with blood plasma proteins using support vector
machines. IEEE J. Biomed. Health Inform., 25, 1, 218–226, January 2021.
23. Deng, J., Dong, W., Socher, R., Li, L., Li, K., Fei-fei, L., ImageNet: A large-
scale hierarchical image database. IEEE Conference on Computer Vision and
Pattern Recognition, 2009, pp. 248–255, 2009, https://ieeexplore.ieee.org/
document/5206848.
24. LeCun, Y., Botou, L., Bengio, Y., Haffner, P., Gradient based learning applied
to document recognition. Proc. IEEE, 1998, http://yann.lecun.com/exdb/
publis/pdf/lecun-98.pdf.
25. Krizhevsky, A. and Hinton, G.E., ImageNet classification with deep convolu-
tional neural networks. Adv. Neural Inf. Process. Syst., 25, 1–9, 2012.
26. He, K. and Sun, J., Deep residual learning for image recognition. IEEE
Conference on Computer Vision and Pattern Recognition, 2016, https://doi.
org/10.1109/CVPR.2016.90.
27. Sarraf, S. and Tofighi, G., Classification of Alzheimer’ s disease struc-
tural MRI data by deep learning convolutional neural networks. CoRR,
Vol:abs/1607.06583. 1–14, 2016.
28. Guan, Z., Ritesh, K., Yi Ren, F., A comprehensive study of Alzheimer’ s dis-
ease. Proc. Mach. Learn. Res., 1–16, 2019.
29. Fong, J.X., Shapiai, M.I., Tiew, Y.Y., Batool, U., Fauzi, H., Bypassing MRI
pre-processing in Alzheimer’ s disease diagnosis using deep learning detec-
tion network. 16th IEEE International Colloquium on Signal Processing & Its
Applications (CSPA 2020), CSPA, pp. 28–29, 2020.
30. Marcus, D.S., Fotenos, A.F., Csernansky, J.G., Morris, J.C., Buckner, R.L.,
Open access series of imaging studies (OASIS): Longitudinal MRI data in
nondemented and demented older adults. J. Cogn. Neurosci., i, 12, 2677–
2684, 2010, https://doi.org/10.1162/jocn.2009.21407.Open.
31. Sarvamangala, D.R. and Kulkarni, R.V., Convolutional neural networks
in medical image understanding: A survey. Evol. Intell., 15, 1, 1–22, 2022,
https://doi.org/10.1007/s12065-020-00540-3.
32. Thirumagal, E. and Saruladha, K., Design of FCSE-GAN for dissection of
brain tumour in MRI. International Conference on Smart Technologies in
Computing, Electrical and Electronics (ICSTCEE), pp. 1–6, 2020.
9
Design of Fuzzy Logic-Based
Smart Cars Using Scilab
Josiga S.*, Maheswari R. and Subbulakshmi T.
School of Computer Science and Engineering, Vellore Institute of Technology,

Chennai, India
Abstract
Fuzzy logic could be implemented to build smart cars that have the ability to
reduce human intervention while driving which helps in taking precise and
immediate decisions. A Fuzzy Inference System (FIS) is programmed using the
software ‘Scilab’ to work in varying geographic regions, road and atmospheric
conditions. To ensure unhinged functioning of the smart car, the inputs to the FIS
such as atmospheric conditions, external factors, traffic signal, obstacle distance,
and obstacle position must be calculated in a proper way. The fuzzy inputs are
obtained, and a rule base is created using Scilab to create the FIS that can provide
the speed and direction of the vehicle as the outputs of FIS. The objective of this
chapter is to advocate a novel approach to design a FIS to control the speed and
direction of a vehicle. The implementation of FIS based smart cars result in lesser
accidents and faster response system for the world.
Keywords: Fuzzy logic, smart car, scilab, fuzzy inference system, fuzzy rules
9.1 Introduction
Smart cars are vehicles that are capable of secure driving without the
intervention of the driver. Automation of driving was intiated on different
levels covering manually operated vehicles to completely automated vehi-
cles. Society of automotive engineers have a classification system with six
levels ranging from complete driver supervision to no driver supervision.

143
Companies are surveying the best level of automation for the vehicle.
Automatic braking system is a quintessential part in a smart car that plays
an integral role in reducing accidents. The survey indicates that the smart
cars practice safe and economic methods of driving about 20% more than
humans. Autonomous cars [1] are developed differently for each region
based on the data collected in that region. Smart cars are in the developing
and testing stage in all the countries. A major proportion of smart cars are
electricity operated rather than being fuel operated. As emission of gases
from vehicles play a large role in air pollution, the proposed change can
bring about a noticeable decrease in air pollution.
Fuzzy logic was introduced in 1965 to deal with the problem of vague-
ness and ambiguity in data that requires accurate results which can be
provided only through numerical data. Fuzzy logic focusses on the inter-
mediate values which are partially true or partially false values rather than
the usual absolute values [18]. Fuzzy logic converts the uncertainty in a
problem into a precise quantity. Fuzzy logic balances the impreciseness in
the system by designing to operate within a window of uncertain condi-
tions. Fuzzy logic is primarily used in two situations [13]. First is when the
problem is of high complexity and is difficult to understand and second is
when the problem requires a tentative but fast solution [17]. We can infer
a general solution from a precise input by ignoring accuracy and uncer-
tainty. Fuzzy input can be obtained from real life experiments or through
surveys and research. The fuzzy rule base is a set of rules framed based on
the IF…THEN conditions [6]. The other rules are inferred from the given
set of rules [10]. Both the framed rules and the inferred rules are integrated
to provide the fuzzy output. The fuzzy input is fuzzified and classified into
an appropriate range of membership functions [11].
In this book chapter we use fuzzy logic application to manage the uncer-
tainty in operating smart cars. The evolution of the automotive technology
proves beneficial to people in various ways. Smart cars are safer, comfort-
able and in overall provides an enjoyable experience. Smart cars have a
variety of cameras and sensors to track and gain knowledge of the external
environment and drive the vehicle accordingly [15]. The sensors could be
used to gain input from the environment [2]. Accidents due to the driver’s
fault [16] is the most common cause of accidents in countries all over the
world [5, 7]. Automatic forward and reverse parking of vehicles were also
made automatic with the help of fuzzy logic [3, 8]. A fuzzy control system
is proven to be sure to reduce the chance of accidents to a greater extent
as the number of blunders are reduced to a great extent [10, 12]. Reduced
number of accidents also improves the economy as a huge sum of money is
spent on accidents. Smart cars also helps people save a lot of precious time
Design of Fuzzy Logic-Based Smart Cars Using Scilab 145
Trend Analysis of fuzzy logic paper in each decade

70000
60000
50000
40000
30000
20000
10000
0
1965–1975 1975–1985 1985–1995 1995–2005 2005–2015 2015–Present
Figure 9.1 Recent trend analysis of fuzzy logic.
instead of wasting it on driving. Smart cars help the physically and visually
challenged to use cars in ease without the assistance of other people. Smart
cars also will reduce the traffic congestion [4] as the vehicles coordinate
with one another and reduce the crowding of vehicles [14]. We design a
FIS which provides the required output with the inputs which could be
fetched from the sensors. Speed and direction are the fuzzy outputs gen-
erated based on the given inputs. Fuzzy rule base is designed based on
the gathered data and real time data which provides a working solution to
operating smart cars.
The development of technology faces many barriers like social, eco-
nomic and environmental ones. A new paradigm is to be created to intro-
duce the concept of smart cars. The field of work is to be expanded to meet
the growing demand of the population. The interest of many researchers
has increased in the automation of the automobile industry. Due to their
advantages, smart cars have been researched in different parts of the world.
In the recent years, the use of fuzzy logic has been increasing. Fuzzy logic
is incorporated in several new technologies to design and enhance their
ability. The graph in Figure 9.1 shows the increase in use of fuzzy logic over
the years.
9.2 Literature Survey

The development of technology faces many barriers like social, economic
and environmental ones. A new paradigm is to be created to introduce
the concept of smart cars. The field of work is to be expanded to meet

the growing demand of the population. The interest of many research-
ers has increased in the automation of the automobile industry. Due to
their advantages, smart cars have been researched in different parts of the
world. This paper focusses on the development of smart cars and the fac-
tors involving in the functioning of smart cars in real time setting.
9.2.1 Fuzzy Logic for Automobile Industry

A car-like mobile robot (CLMR) method has been described to reverse an
empty parking bay automatically using a fuzzy logic control approach. The
Fuzzy logic approach was expressed in the form of fuzzy rules based on the
data given by experienced drivers. Readings from ultrasonic sensors and
fuzzy rear corner and the rear back mirror are used to have the input in
different angles of the car. Based on the rules obtained from experienced
drivers, the steering angle is obtained as the fuzzy output [3].
Fuzzy input parameters that are included while making the fuzzy rule
base are acceleration, braking, gear and terrain. A controller is designed
based on the fuzzy inputs and is used to control the speed of the vehi-
cle. Unlike conventional methods which are time consuming, this fuzzy
controller contains a self-tuning algorithm to obtain the optimum control
parameters [2].
An automatic cruise control system in self-driving cars is designed
based on Fuzzy logic and Deep Steering Neural Networks using MATLAB
with GPU coder. The steering control and velocity signal in constraints.
The front-end stage uses CNNs with the input from raw sensory data for
predicting steering control. The output works as natural inferences for rec-
ommending velocity and adapting new steering control. All the vehicle
dynamics are considered while developing a rule base for the fuzzy control
system [10].
A fuzzy neural network is described that is able to change the duration
of the green signal according to the traffic situation. This system can con-
trol the given junction as well as the adjacent intersections on fuzzy logic
using neural networks based on the type of vehicle. The fuzzy model con-
trols traffic-light systems to reduce traffic congestion [14].
9.2.2 Fuzzy Logic for Smart Cars

A method is provided using fuzzy logic for traffic control. Fuzzy logic is
used to analyze environmental factors like road carpet, visual range and
weather conditions detected by sensors. Traffic control system is the
fuzzy output. Evaluation of the road condition and providing slippery or

fog warning. The fuzzy rule base also includes some uncertain but essen-
tial factors like the quality of the sensors, road condition, fog or slippery
warning [4].
An intelligent research method has been described to detect the posi-
tion of the nearby vehicles using sensors and to apply fuzzy logic to park
the cars in the right position. Four sensors in the front and rear side are
used to have the input of four directions for the car. Based on the distance
and decision, the position of the steering wheel and direction would be
determined. The implementation is done on fuzzy logic and a prototype is
also built in continuation with the implementation. The system is imple-
mented by using infrared sensors and micro controllers which are con-
trolled by fuzzy inference system for parking in the available space using
forward and reverse direction moves [1].
A fuzzy logic based behavioral system is demonstrated to provide accu-
rate and effective suggestions. An appropriate fuzzy logic subset is chosen
and the a set of selected parameters are evaluated using frequent pattern
information and will be optimized to create the behavioral weight which is
essential for building the fuzzy inference system. This paper provides sug-
gestions to the smart vehicle using only fuzzy logic and without modelling
or machine learning techniques [16].
9.2.3 Fuzzy Logic for Driver Behavior Detection

A fuzzy logic controller for an automated car braking system using
MATLAB’s fuzzy logic toolbox is developed for research. This aim of this
controller is to brake a car when the car approaches an obstacle at a specific
range. PI and PD type fuzzy logic controllers are compared and analyzed.
The fuzzy rule base consists of position and velocity as inputs and the brake
system is output. The rule base consists of four rules and three membership
function. The main objective of this controller is to apply brakes at the
right time to avoid injuries [5].
The main objective of the fuzzy system is to prevent accidents and inju-
ries in vehicles. The control system positions the car in a loading dock sys-
tem without any accidents. Experiments on different conditions proved the
efficiency of this fuzzy logic control system [7].
The active suspension control of a vehicle model using fuzzy logic with
five classifications has been demonstrated. Three types of control systems
are designed using fuzzy logic. The first type is a vehicle model with a
combination of passive suspensions and an active passenger seat is con-
trolled. The second type is a combination of active suspensions and passive
passenger seat. The third type is a combination of the passenger seat and
active suspensions. Vibrations in the passenger seat due to road bump
input are recorded for the three types of control systems [12].
9.2.4 Fuzzy Logic Applications for Common Industry

A method to allow its consumers to participate in a smart grid, coordinated
charging and discharging of vehicle’s batteries with minimum system volt-
age in proper limits is developed. The inputs to the Fuzzy inference system
include the states of charge of the electric vehicles, the grid parameters
represented in the system minimum voltage, and the hourly energy price
and the output is the charging levels of the electric vehicles’ batteries. The
effectiveness of this fuzzy control system is proved when being compared
to other uncoordinated charging schemes [6].
A method for describing the control problem of autonomous bay park-
ing system. Inputs to this fuzzy control system are initial position, forward
and backward speed and steering angle. A fuzzy rule base is designed and
a fuzzy speed and steering controller is designed for the parking of the
vehicle. Accuracy of this fuzzy control system can be measured from the
simulation results [8].
Chargeable vehicles use electricity from renewable sources, so they are
fuel and energy effective than fuel powered vehicles. A fuzzy rule base
is developed based for a system derived based on the relation between
Distributed, Fast charging and Battery swapping infrastructures. This
fuzzy logic controller provides the total consumption and cost scenario as
the fuzzy output [9].
A research of fuzzy membership functions based for car-following mod-
els is under development. In this paper the least squares to fit discrete data,
and their membership functions are found out. The method is backed up
with satisfying results from the testing experiment [11].
The analysis of the methods in which teachers of the 21st century use of
digital tools are demonstrated. Digital competencies consist of digital com-
munication, management, preservation, analysis and presentation of data.
The pentagonal fuzzy number was used in the analysis and other arithme-
tic operations were also included in the analysis of the teaching methods
in mathematics edition [17].
Recent works in fuzzy logic control applications is sufficient to survive
only moderate level problems. Other difficult problems need more com-
plex technologies [15].
A video based detection method is described to make a fused video by
using optical flow technique. The detection and estimation of motion of
an object is observed in the fused video and an algorithm is developed

based on fuzzy logic using Matlab. The fuzzy logic system uses fuzzy mea-
sure theory and aggregation for suitability and image and video fusion are
checked. Fuzzy logic using fuzzy measure theory and aggregation has been
used to detect and estimate the speed of the motion by comparing with the
threshold and mean values [18].
The fuzzy control system is can generate summarized models from a
given number of rules. This helps in understanding the generated fuzzy
model and predicting the financial applications. Experiments have been
performed to predict the good/bad customers and to predict the arbitrage
opportunities in the stock markets [13].
9.3 Proposed Fuzzy Inference System for Smart Cars

The proposed FIS contains seven modules namely input, fuzzification,
membership functions, rule base, rule aggregation, defuzzification and
output. The seven modules are implemented in scilab–fuzzy logic toolbox.
The functions and GUI available in Scilab makes our work easier for imple-
mentation of the smart cars. The GUI interface of scilab is given in Figure
9.2.
9.3.1 Fuzzification
Fuzzification is the process of converting organized numerical quan-
tity into a vague and ambiguous quantity based on a given set of rules.
Figure 9.2 GUI Interface of the FIS in Scilab.

Figure 9.3 Fuzzy inputs in Scilab.
The input block consists of five inputs like atmospheric conditions, exter-
nal factors, traffic conditions, obstacle distance and the obstacle position.
The input data is collected from various sources that record these type of
data in real life. Atmospheric conditions include wind speed, rain, sun,
and other factors. External factors include lighting of the place, road con-
ditions, and other factors. Traffic conditions include the amount of vehicles
in a specific place at a specific time, obstacle distance include the distance
between the obstacle and our vehicle, obstacle position specifies the direc-
tion of the obstacle with respect to our vehicle. The input data are collected
and preprocessed using appropriate methods and are classified into vari-
ous membership functions as shown in Figure 9.3.
9.3.2 Membership Functions

Membership functions are used to represent the input values in graphical
representation. The number of membership function may differ for each
input value but the interval between the membership functions are equal.
For external factors, there are five membership functions named Very
Good, Good, Partly good, Bad, Very bad as shown in Table 9.1. The second
input atmospheric conditions also has five membership functions namely
very good, good, partly good, bad and very bad as shown in Table 9.2. The
GUI interface of Scilab for the input variable atmospheric conditions are
shown in Figure 9.4. The third input Traffic has three membership func-
tions Red, Yellow and Blue as shown in Table 9.3. The fourth input obstacle
distance that has five membership functions very low, low, medium, high,
very high as shown in Table 9.4. The fifth input obstacle position has eight
Table 9.1 External factors.

External factors Factors (qualitative)
100 to 80 Very good
60 to 80 Good
40 to 60 Partly good
20 to 40 Bad
0 to 20 Very bad
Table 9.2 Atmospheric conditions.

Atmospheric conditions Factors (qualitative)
100 to 80 Very good
60 to 80 Good
40 to 60 Partly good
20 to 40 Bad
0 to 20 Very bad
Figure 9.4 Membership functions for the input variable atmospheric conditions.
Table 9.3 Traffic.

Traffic Signalcolor (qualitative)
0 to 5 Red
5 to 10 Yellow
10 to 15 Green
Table 9.4 Obstacle distance.

Obstacle distance (in m) Distance (qualitative)
0 to 10 Very Low
10 to 20 Low
20 to 30 Medium
30 to 40 High
40 to 50 Very high
Table 9.5 Obstacle position.

Obstacle Position (in degrees) Position (qualitative)
-22.5 to 22.5 North
22.5 to 67.5 North East
67.5 to 112.5 East
112.5 to 157.5 South East
157.5 to 202.5 South
202.5 to 247.5 South West
247.5 to 292.5 West
292.5 to 337.5 North West
membership functions North, North East, South, South East, West, North
West, East and South West as shown in Table 9.5.
9.3.3 Rule Base

Rule base is a collection of a set of rules which are used to convert an organized
quantity into a vague quantity in the process of fuzzification. The rule base
used for this implementation is If…Then. The rules in the rule base are stud-
ied and enhanced and inferences are observed from the given set of rules and
the aggregation of all the rules take place. There are five inputs are aggregated
using AND condition. A fuzzy-associated memory table is created using all
possible combinations of input values and the outputs speed and direction are
determined based on the given set of rules as shown in Table 9.6.
9.3.4 Example Rules

If atmospheric condition is Very Good AND External factor is Very Good
AND Traffic is Green AND Obstacle distance is very Low AND Obstacle
position is South East, then Speed is Very High AND Direction is East.
If atmospheric condition is Very Bad AND External factor is Very Bad
AND Traffic is Red AND Obstacle distance is Very High AND Obstacle posi-
tion is North, then Speed is Very Low AND Direction is North East.
9.3.5 Defuzzification
The data are then defuzzified using, the most common method for defuzzi-
fication, the centroid method. Defuzzification is the conversion of the
Table 9.6 FAM Table.

Atmospheric External Obstacle Obstacle
conditions factors Traffic distance position Direction Speed
VB VB R VL N NE VL
G B R L NE E L
PG PG Y M S SE M
VG VG G VH SE E VH
B VB G H SW W H
vague, ambiguous quantity back into an organized numerical quantity

after the operations are done. So the input and output are numerical quan-
tities, but the intermediate processes are done using ambiguous data. The
defuzzified output is then displayed as the speed and direction. The first
output speed is the measure of the speed which is advisable in a particular
situation. Speed has five membership functions Very low, Low, Medium,
High and Very high as shown in Table 9.7. The other output direction tells
the direction that is to be followed so as to have a safe and smooth drive.
Direction has eight membership functions North, North East, South,
South East, West, North West, East and South West as shown in Table 9.8.
Table 9.7 Speed.

Speed Speed (qualitative)
100 – 125 Very high
75 – 100 High
50 – 75 Moderate
25 – 50 Low
0 – 25 Very low
Table 9.8 Direction.

Direction (in degrees) Position (qualitative)
337.5 to 22.5 North
22.5 to 67.5 North East
67.5 to 112.5 East
112.5 to 157.5 South East
157.5 to 202.5 South
202.5 to 247.5 South West
247.5 to 292.5 West
292.5 to 337.5 North West
9.4 Implementation Details and Results

The five state variables for this Scilab implementation will be the atmo-
spheric conditions, external factors, traffic, obstacle position, and obstacle
distance as shown in Figure 9.5. The fuzzy output will be the direction and
speed. The control system when given some speed and direction will alter
the input state variables. Unless the input conditions are changed, the out-
put will not differ. Next we design membership functions for the five sinput
state variables. After that, we design membership functions for the output
variables speed and direction as shown in Tables 9.7 and 9.8. Following
that, define the rules and summarize them in a FAM table. The values
in the FAM table are the fuzzy outputs. After designing the membership
functions and FAM table, the information is coded in Scilab.
The membership functions of the input and output parameters are
stored in the fuzzy inference system. Then Scilab fuzzy logic toolbox func-
tion plotvar () is used to plot the membership functions. The plot for the
input and output membership functions are obtained from Scilab as shown
in Figures 9.6 and 9.7. Then the plotsurf () function is used to plot the
surface of the output plot. In the fuzzy logic toolbox in Scilab, evalfls ()
function is used to calculate the output for any given input value. Hence
Fuzzy Inference System

Fuzzy Inputs
1. Atmospheric Defuzzifi-
conditions
2. External Fuzzification Rule cation Fuzzy
Triangular
factors Vague to Rule Base Aggregation Numerical Outputs
Membership
3. Traffic Numerical IF..THEN (AND to 1. Speed
4. Obstacle functions
position
values method) Vague 2. Direction
5. Obstacle values
distance
Figure 9.5 Design of smart car using fuzzy logic framework.

Member functions for input number 1 named factors

mu(factors)
1
VG
0 G
0 10 20 30 40 50 60 70 80 90 100 PG
B
factors
VB
Member functions for input number 2 named traffic

mu(traffic)
1
R
Y
0
G
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
traffic
mu(atm_conditions)
Member functions for input number 3 named atm_conditions

1 VG
G
0
PG
0 10 20 30 40 50 60 70 80 90 100
B
atm_conditions VB
Member functions for input number 4 named obs_dist
mu(obs_dist)
1
VH
obs_dist H
0
M
0 5 10 15 20 25 30 35 40 45 50
L
obs_dist
VL
mu(obs_position)
Member functions for input number 5 named obs_position

1
obs_position N
NE
0
E
0 50 100 150 200 250 300 350
SE
obs_position
S
Figure 9.6 Plot of the five input membership functions using Scilab.
when the input values are provided in the code, the desired fuzzy output
is obtained in the Scilab console. The fuzzy output for n input values can
be calculated using the evalfls () function by passing it inside a for loop for
n times. By using this method, the output for any number of input values
can be obtained.
9.5 Conclusion and Future Work

This work proposed the design of a fuzzy inference system in Scilab.
Direction and speed of a vehicle were obtained with varying factors like
the atmospheric conditions, traffic light signal, external factors, obstacle
position and obstacle distance. The relatively accurate output is obtained
from Scilab is close to the experimental output obtained which proves the
accuracy of the results. In future, a real time smart car can be designed
with the given data and checked for the accuracy with respect to the
Member functions for output number 1 named speed
1 VH
H
0.8 M
L
mu(speed)
0.6
VL
0.4
0.2
0 10 20 30 40 50 60 70 80 90 100 120
speed
Member functions for output number 2 named direction
1 N
NE
0.8 E
mu(direction)
SE
0.6
S
0.4 SW
W
0.2 NW
0
0 50 100 150 200 250 300 350

direction
Figure 9.7 Plot of the two output membership functions using Scilab.
simulated results. Also other input and output factors required for the
functioning of a smart car can be added in this implementation.
References
1. Sokri, M.N., Mahamad, A.K., Saon, S., Yamaguchi, S., Ahmadon, M.A.,
Autonomous car parking system using fuzzy logic. IEEE International
Conference on Consumer Electronics (ICCE), 2021.
2. Mahmud, K. and Tao, L., Vehicle speed control through fuzzy logic. IEEE
Global High Tech Congress on Electronics, 2013.
3. Liu, Z., Wang, Y., Lu, T.-F., Car-like mobile robot reverse parking using fuzzy
logic control approach. First International Conference on Robot, Vision and
Signal Processing, 2011.
4. Krause, B. and Pozybill, M., Fuzzy logic data analysis of environmental data
for traffic control. Proceedings of 6th International Fuzzy Systems Conference,
2002.
5. Mamat, M. and Ghani, N.M., Fuzzy logic controller on automated car brak-
ing system. IEEE International Conference on Control and Automation, 2009.
6. Viegas, M.A.A. and da Costa Jr., C.T., Fuzzy logic controllers for charging/
discharging management of battery electric vehicles in a smart grid. J. Control
Autom. Electr. Syst., 2021.
7. Riid, A., Pahhomov, D., Rustern, E., Car navigation and collision avoidance
system with fuzzy logic. IEEE International Conference on Fuzzy Systems,
2004.
8. Wang, Z.-J., Zhang, J.-W., Huang, Y.-L., Zhang, H., Mehr, A.S., Application
of fuzzy logic for autonomous bay parking of automobiles. Int. J. Autom.
Comput., 2011.
9. Sachan, S. and Mortka, M.Z., Fuzzy logic triggered charging infrastruc-
tures for electric vehicles. IEEE First International Conference on Smart
Technologies for Power, Energy and Control (STPEC), 2020.
10. Van, N.D. and Kim, G.-W., Fuzzy logic and deep steering control based rec-
ommendation system for self-driving car. 18th International Conference on
Control, Automation and Systems, 2018.
11. Xiong, Q., Chen, Z., Zeng, X., Guo, J., Development of membership
degree functions of the car-following models based on fuzzy logic. Second
International Conference on Intelligent Computation Technology and
Automation, 2009.
12. Yagiz, N., Sakman, L.E., Guclu, R., Different control applications on a vehicle
using fuzzy logic control. Sadhana, 2008.
13. Bernardo, D., Hagras, H., Tsang, E., A genetic type-2 fuzzy logic based sys-
tem for the generation of summarised linguistic predictive models for financial
applications, Springer-Verlag Berlin Heidelberg, Germany, 2013.
14. Cheng, S.-T., Li, J.-P., Horng, G.-J., Wang, K.-C., The adaptive road routing
recommendation for traffic congestion avoidance in smart city. Wirel. Pers.
Commun., 2014.
15. von Altrock, C., Krause, B., Zimmennann, H.-J., Advanced fuzzy logic con-
trol technologies in automotive applications. IEEE International Conference
on Fuzzy Systems, 1992.
16. Le, D.-N., An efficient driver behavioral pattern analysis based on fuzzy log-
ical feature selection and classification in big data analysis. J. Intell. Fuzzy
Syst., 2021.
17. Kar, R. and Shaw, A.K., Analysis of digital competencies of 21st century
teachers of mathematics education by pentagonal fuzzy number and some of
its arithmetic operations. J. Educ. Learn. Math. Res., 2021.
18. Mukherjee, G., Motion detection and estimation in fused video by using
optical flow technique with fuzzy application. Int. J. Adv. Res. Comput. Sci.,
2010.
10
Financial Planning and Decision Making
for Students Using Fuzzy Logic
G. Surya Deepan1* and T. Subbulakshmi2
1
Electrical and Electronics Engineering Student, Vellore Institute of Technology,
Chennai, Tamil Nadu, India
2
School of Computer Science and Engineering, Vellore Institute of Technology,
Chennai, Tamil Nadu, India
Abstract
Fuzzy logic has lot of applications in daily life, which includes financial plan-
ning as well. Common applications like stock market, investment decision
making is implemented using fuzzy logic concepts. The focus of this research
is to take buying decisions for a product using fuzzy logic. The data collection
is done from students. The qualitative and quantitative data such as income
and necessity which are collected from the students are used as inputs. Fuzzy
inference system (FIS) takes the inputs and processes further to produce the
required decisions based on the intelligent rules. The output of FIS helps the
student to take a decision towards deciding the worthiness of the product.
The buying decisions are made effective with the use of FIS since the buying
options are completely analysed. The results enable the student to decide the
worth of the product toward buying based on intelligent output from fuzzy
logic.
Keywords: Fuzzy logic, planning, finance, students, decision making
10.1 Introduction
Planning is the most essential element in a person’s everyday life. In
finance, planning is everything. Planning makes people well prepared
for their future. One can describe planning as the process of reducing

159
uncertainties. Ironically planning itself is an imprecise and uncertain pro-

cess. This impreciseness is the one that opens the gate to the application
of fuzzy logic, which is created for a single purpose and that is solving
imprecise sets.
The term “fuzzy logic” was introduced by Zadeh in 1965. Its main pur-
pose is to deal with solving problems of impreciseness and uncertainty. In
crisp logic the values are limited to 0’s and 1’s but in fuzzy logic the value
can be any number between the range of 0 and 1. These values do not rep-
resent probability like many cases but it represents the degree of truth. It
basically gives a mathematical model to ambiguousness. The field of fuzzy
logic is vast and it has a lot of real-life applications. They are used in pattern
recognition, engineering, artificial intelligence, economics, quality control,
decision making, sensors and so on. Basically, fuzzy logic is a contrast to
Boolean logic which is very well known to many.
Fuzzy logic has numerous applications in real life and they can be
applied anywhere or to any field. This adaptive nature of fuzzy logic is
the reason why it is so popular in the recent days. The number of papers
released in this topic has been increasing over the decades. Over the years
the number of publications in the field of fuzzy logic have been increas-
ing exponentially. The popularity of fuzzy is more than ever in the current
trend. According to Figure 10.1, in the past decade nearly 65,000 articles
have been published under the field of fuzzy logic. This popularity shows
the importance of fuzzy logic in our current world.
The number of research conducted on fuzzy logic in the field of finance
is very low despite being so popular. Fuzzy logic has a lot of potential in
Trend Analysis of fuzzy logic paper in each decade.

70000
60000
50000
40000
30000
20000
10000
0
1965 - 1975 1975 - 1985 1885 - 1995 1995 - 2005 2005 - 2015 2015 - Present
Figure 10.1 Trend analysis of fuzzy logic papers in the past decades.
Student Financial Planning Using Fuzzy Logic 161
the field of finance. This potential is due to the unpredictable and imprecise
nature of the subject, i.e., “finance.” In this paper, we are going to see just
one application of fuzzy logic in the field of finance.
The main objective of this paper is to discuss the importance of financial
planning for a student. In this research, we are using fuzzy logic to solve the
impreciseness in the planning process. Financial stress has a large impact
on students’ daily lives. This stress also causes them to fail in academic
performance. We are introducing our methodology to reduce this stress.
Our methodology is designed to eliminate the impreciseness in the process
of planning. This research will make it possible for everyone to plan their
monthly expenditures.
In this research, we are using the help of an open-source software named
“Scilab”. Scilab is a famous open-source software which is mainly used to
test algorithms and carry out numerical computations. One of the greatest
advantages in Scilab is the feature called “sciFLT,” where FLT stands for
fuzzy logic toolbox. This feature allows the user to solve the fuzzy logic
problem with ease. This feature is the main reason for us to opt the Scilab
platform for the implementation of our paper. Since Scilab is an open-
source software, after the completion of program, it can be published as
a demo tool box, which can be later used by upcoming scholars for their
profit.

In the literature review, some work related to the topic of fuzzy logic in the
field of finance have been reviewed, along with those papers some of the
general papers which studies about the financial behavior of students were
also reviewed.
Credit scoring is an important activity in the field of finance. Some of
the advantages of credit scoring involve improvement in cash flow, reduc-
ing credit losses, purchasing behavior of existing customers. This domain
handles problems of prediction and classification and hence fuzzy logic
is used. Since the traditional fuzzy models have several disadvantages,
they have used hybrid models in this paper. The inputs are taken from
Australian Credit Data Set, German Credit Data Set, Japanese Credit
Data Set. The paper concludes by quoting that they have seen a 10.95%
to 14.91% improvement in their hybrid method compared to other tradi-
tional methods [1].
Generalization of fuzzy sets is known as IFSs and in this paper they
are using IFSs to stimulate decision making process and Multi Criteria
Decision Making (MCDM) of a human being. A methodology has been

proposed to estimate the weight of an alternative in a MCDM. A numeri-
cal has also been solved in order to conclude the paper. The results give us
some flexible ways for us to stimulate real MDCM [2].
A mobile application which is developed to help students to man-
age their time and money. The research method is comparative analysis.
The research is conducted by integrating both quantitative and qualita-
tive methods. The data were collected via Google Forms in the country
Malaysia. The research concludes saying that the higher academic success
is highly related with time management and financial management which
will help reducing stress and the application they developed will help them
in this process [3].
A financial management system based on fuzzy rules is developed. The
research introduces important evaluation methods and major decision
rules used in financial management. We are introduced to three major
steps. The inputs are based on judgements as “poor,” “fair,” “good.” The
defuzzified score used rank the inputs we got. The research is concluded
with the ranking of cities with good financial management [4].
The research discusses about the importance of personal financial
planning especially for college students and also discuss about retirement
financing that are available to them. It talks about the history and other
alternative retainment plans. The input is taken form student survey. The
research concludes as a wakeup call for youngsters and also for workers
and employers [5].
The research of introduces a new method for group decision making
using fuzzy rules. They are making use of fuzzy Delphi method. The result
is a rank of fuzzy number problem, ranking from best to worst. They have
proposed an algorithm to obtain the best selection for evaluating systems
[6].
Students have lot of option to strategize their higher studies and this
paper explores all the option a student has in their hand. The work starts
with academic planning and it goes on with all the problems a student may
face. It follows up with several places where students can get some help
from [7].
This paper mainly discusses about the expenditure and savings pattern
of the students and the sources of income of the student. The inputs were
collected via google forms, i.e., by link sharing method. The data analysis
is done by quantitative analysis methodology. This methodology mainly
includes minimum, maximum, frequency, and mean calculations [8].
In this work, an investment decision-making software is developed.
To verify this software, we are implementing it in the case of a solar
investment project. The branch is fuzzy logic implemented here is spe-

cifically designed to imitate human thinking and reasoning. The result
obtained verified that the software developed is appropriate for assisting in
investment decision making [9].
This work implements artificial intelligence (AI), to control risk-based
money management decisions. The experiment proposes an innovative
fuzzy logic approach which categorizes technical rules performance across
different regions in the trend and space. This work conveys a lot of contri-
butions to the field of finance [10].
This work analyzes a lot of financial factors which will essentially affect
a student’s decision in the change of paths in education. A fuzzy logic using
Mamdani style was created to according to the following parameters. Three
input rules are used to establish an output variable of probability of change.
The results give the probability in the change of paths of students [11].
The planning model established in this paper is adapted from the hybrid
of fuzzy logic. This analysis shows the pattern of financial planning in cam-
pus level. Different sources of income were taken as input, and central
average defuzzification method is used to gain the output. The results con-
clude that students have different spending behaviors [12].
Mobile applications are very useful with our day today life and as such
application is used in this upcoming paper, The paper mainly shows how
fuzzy logic, mobile application and GPS (Global Positioning System) can
analyse a student’s life style. The input is obtained from the GPS (Global
Positioning System) and with the mobile application. The mobile applica-
tion was designed in such a way that the fuzzy logic in the paper can be
implemented in the real life. The mobile application was programmed to
give recommendations and suggestions to the users after analysing their
lifestyle [13].
The article basically gives us an overview about uses of fuzzy logic in
the field of finance. The main reason behind the article was to address a
single problem, i.e., The lack of usage of Fuzzy logic in the field of finance.
This article was written with one motive in mind that is to provide a rele-
vancy for supervisory, banks, academic researchers and regulatory bodies
as well [14].
10.3 System Architecture

The blueprint of the research is given below in Figure 10.2. The system
contains seven steps in total. A brief explanation about each step is given
below.
Membership Function:
Input:
Fuzzification. 1. Product Cost
1. Cost of the product
Fuzzification is done with 2. Necessity
2. Necessity
fuzzy operators. 3. Cost Percentage
3. Quality
4. Quality
Deffuzzification:
Rule Base: Rule Aggregation:
Deffuzzification is done
IF…THEN. Fuzzy Output.
by centriod method.
Output:
Essentiality.
Figure 10.2 Proposed architecture of fuzzy finance planning system.
In this paper, the data are collected from first year college students.
Around 120 of them participated. The inputs will be the data collected
from these students. The students have to answer these questions in both
qualitative and quantitative order. The input will be classified as the mem-
bership functions. These membership functions are processed by the help
of the fuzzy rules. Finally, defuzzification takes place. The defuzzified value
is seen as the output that will give the essentiality of the item of purchase.
10.3.1 Input
As mentioned, the input will be taken from students of first years. The
questions will be regarding several things, such as their monthly allowance,
their spending spree, the essentiality of various items in their perspective
and so on. The data will be gathered both in qualitative and quantitative
means. These data will be our input. These inputs are named as Quality,
Necessity, Income.
10.3.2 Fuzzification
Fuzzification can be defined as the process of conversion of a crisp set to a
fuzzy set with the help of the information we know. Now, in this research,
we are converting the data collected from the students (crisp set) into a
fuzzified sets with the help of fuzzy operators. The fuzzified sets are then
categorized into membership functions.
10.3.3 Membership Function

In fuzzy logic, membership function actually denotes the degree of truth of
an instance. It was introduced by Zadeh in 1965 in his first paper on fuzzy
logic. The value of a membership function ranges from 0 to 1. It can be
operated in any domain. In this particular paper there are three member-
ship functions being employed. They are
• Necessity,
• Cost percentage,
• Quality.
These membership functions represent unique set of values and those

values are categorized into different sets based on the linguistic variables.
Figure 10.3 represents the waveform of the input membership functions.
This research makes use of Triangular member ship function.
Member functions for input number 1 named Percentage

1
mu(Percentage)
VL
L
M
0.5 H
VH
0
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100
Percentage
Member functions for input number 2 named Quality

1 VL
mu(Quality)
L
M
0.5 H
VH
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 10.5 11
Quality
Member functions for input number 3 named Necessity_Level

mu(Necessity_Level)
1 VL
L
M
0.5 H
VH
0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4 4.2 4.4 4.6 4.8 5
Necessity_Level
Figure 10.3 Input membership functions.

10.3.3.1 Necessity
This membership function actually represents the necessity of the items
which are being purchased. In other words, this function answers the ques-
tion “whether the item purchased is worth the cost?” The money spend
by a student goes into many things, such as food, clothing, rent, Internet,
entertainment. This section is also consisting five sets of levels ranging
from Most Necessary to Least Necessary. The Items purchased are catego-
rized into these five levels based on the data collected from the first years.
The five levels are stated in Figure 10.4, which is given below.
10.3.3.2 Cost Percentage

This membership function gives us a value which represents how much
percentage does the cost of the purchased item holds with respect to the
total resource. In other words, if the cost percentage of the purchased item
Level • Very High (VH) which ranges between 0 – 1

1
Level • High (H) which ranges between 1 – 2
2
Level • Medium (M) which ranges between 2 – 3
3
Level • Low (L) which ranges between 3 – 4
4
Level • Very Low (VL) which ranges between 4 – 5
5
Figure 10.4 Membership variables of necessity.
• 0% to 20% - Very High (VH)

1
• 20% to 40% - High (H)
2
• 40% to 60% - Medium(M)
3
• 60% to 80% - Low (L)
4
• 80% to 100% - Very Low (VL)
5
Figure 10.5 Membership variables of the membership function cost percentage.

is 50% then the cost of the purchased item is half of the total resource of the
month. This function has five categories which denotes the ranges of the
percentage. The levels of cost percentage are given in Figure 10.5.
10.3.3.3 Quality
Every product has a different quality. The quality of a product has a great
impact on its worthiness. Nowadays, students tend to care less about the
quality of a product. This is mainly because of lack of awareness. Students
do not know how low-quality products destroy their finances. And in this
paper, we are not going to leave out such an important part of finance. We
are going to segregate this membership function into five different catego-
ries, and they are explained in Figure 10.6.
10.3.4 Fuzzy Rule Base

Fuzzy rules are rules which are used to operate fuzzy logic. These are the
base structure of problems of fuzzy logic. The base rule used in this paper
is IF…THEN.
In total, there are 125 rules applied to this problem. Table 10.1 above
describes some of the rules which are applied. The rows denote the level
of quality (Q) of the product. The columns denote the necessity (N) of the
product. The cost–percentage membership function is set to be constant
value and the value is set to be very high (VH).
Rule 1:
If the cost–percentage is very high (VH), quality is high (H), necessity level
is medium (M), then the essentiality is medium (M).
• Very Low (VL) which ranges from 0 – 2

1
• Low (L) which ranges from 2 – 4
2
• Medium (M) which ranges from 4 – 6
3
• High (H) which ranges from 6 – 8
4
• Very High (VH) which ranges from 8 – 10
5
Figure 10.6 Membership variables of the membership function quality.

Table 10.1 Fuzzy rules.

N
Q VH H M L VL
VH VH H M M L
H H H M L L
M H M L L VL
L M M L VL VL
VL M L VL VL VL
Rule 2:
If the cost–percentage is high (H), quality is very high (VH), necessity level
is low (L), then the essentiality is low (L).
10.3.5 Fuzzy Output

The fuzzy output or the fuzzy output sets are the results we get after imple-
menting the fuzzy rules on the membership functions. This result will be
a value from 0 to 1. These values will later on be converted into desired
outputs by the defuzzification process. Figure 10.7 represents the output
waveform of the fuzzy output.
Member functions for output number 1 named Plan

1.1
DB
1 EL
CB
B
0.9 SB
0.8
0.7
0.6
mu(Plan)
0.5
0.4
0.3
0.2
0.1
–0.1
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 3.4 3.6 3.8 4 4.2 4.4 4.6 4.8 5
Plan
Figure 10.7 Fuzzy output of the membership functions.

10.3.6 Defuzzification
Defuzzification can be defined as the process of conversion of fuzzy sets to
crisp set. After the initial crisp set has been implemented using fuzzy rules,
quantifiable values will be gained. There are many methods of defuzzifica-
tion. The method that this research will be making use of is the “centroid”
method or also known as “centre gravity” method for defuzzification.
This method is the most popular method used for defuzzification. In this
method, we will be making use of center of gravity of fuzzy set and returns
the value in crisp set.
10.4 Conclusion and Future Scope

The fuzzy output is a set of results which is obtained from Fuzzy rules
aggregated. The output will give you the worthiness or the necessity of the
product. One can plan their expenditure from the output obtained from
the defuzzification process. The output will be a set of commands which
is programmed to give the output in the form of a suggestion. The field of
finance is vast and students have no time to submerge into it. Right now,
the process is unpolished. All the inputs are given directly. For example,
the necessity of a particular product is needed to be specified in order to
gain the result. With correct amount of data, we can refine the process. We
can make the algorithm give the output with just name of the product. We
believe that this algorithm can help a lot of students by automating and
optimizing this process with further use of fuzzy logic. This algorithm can
also be refined in a way that it can help a general audience. With proper time
and data, this algorithm can also be used by housewives, office employees,
multi-national companies (MNCs) and many more organizations. They
can use our algorithm to plan their expenditure too. We believe that this
idea of ours have an infinite number of applications in the near future. As
mentioned in the first line of the Introduction, planning is a very essential
element in the field of finance. Planning is an uncertain process, and it can
be very tiring. We are implementing fuzzy logic in order to remove the
uncertainty. Fuzzy logic which has been applied in this case is an effective
way to plan the monthly expenditure of a student. The result will be in the
form of a command which tells a person about the essentiality or worthi-
ness of the product. From the results, students can plan their expenditure
accordingly. We hope that many students will use our algorithm to plan
their monthly expenditure, and we hope to make the planning process eas-
ier for everyone and not only students.
References
1. Khashei, M., Rezvan, M.T., Hamadani, A.Z., Bijari, M., A bi-level neural-based
fuzzy classification approach for credit scoring problems. Complexity, 18, 6,
46–57, 2013.
2. Jana, B. and Mohanty, S.N., An intuitionistic fuzzy logic models for multi-
criteria decision making under uncertainty. J. Inst. Eng. (India): Ser. C, 98, 2,
197–201, 2017.
3. Yeo, J.L., JosephNg, P.S., Alezabi, K.A., Eaw, H.C., Phan, K.Y., Time sched-
uling and finance management. IEEE 2020 IEEE Student Conference on
Research and Development (SCOReD), Batu Pahat, Johor, Malaysia, 2020.9.27-
2020.9.29, University Student Survival Kit., pp. 1–6, 2020.
4. Ammar, S., Duncombe, W., Hou, Y., Wright, R., Evaluating city financial
management using fuzzy rule—Based systems. Public Budg. Finance, 21, 4,
70–90, 2001.
5. James, J., Hadley Leavell, W., Maniam, B., Financial planning, managers, and
college students. Manage. Finance, 28, 7, 35–42, 2002.
6. Sekiguchi, T. and Yu, L., Fuzzy inference-based manager evaluation of ven-
ture enterprises in investment decision making. 1999 IEEE International
Fuzzy Systems Conference Proceedings, Seoul Korea, August 22-25, 1999,
1999.
7. Grothe, W.L. and Cotellessa, R.F., Strategies to finance your higher educa-
tion: Strategies include cost reduction and academic planning, but the most
important involves identifying, evaluating, and selecting financial aid. IEEE
Potentials, 1, Spring, 12–14, 1982.
8. Samartkit, P. and Pullteap, S., A design of decision making-assisted software
using fuzzy logic technique: A case study of solar cell investment project.
Electr. Eng., 231–223, 2019.
9. Vella, V. and Ng, W.L., A dynamic fuzzy money management approach for
controlling the intraday risk-adjusted performance of AI trading algorithms.
Intell. Syst. Account. Finance Manage., 22, 2, 153–178, 2015.
10. Norasibah, A., Ramli, N., Hashim, A., Hashim, E., Zulkifli, N., The
income-expenditure-saving analyses of the university students. Int. J.
Psychosoc. Rehabilitation., 24, 7375–7383, 2020.
11. Ghosh, S., Boob, A.S., Nikhil, N., Vysyaraju, N.R., Kumar, A., A fuzzy
logic system to analyze a student’s lifestyle. IEEE 2017 Ninth International
Conference on Advanced Computational Intelligence (ICACI), Doha, Qatar
(2017.2.4-2017.2.6), pp. 231–236, 2017.
12. Madi, E.N. and Tap, A.O. Md, On fuzzy financial planning model. IEEE
2010 2nd IEEE International Conference on Information Management and
Engineering, Chengdu, China (2010.04.16-2010.04.18), pp. 254–258, 2010.
13. Tarasyev, A.A., Agarkov, G.A., Ospina Acosta, C.A., Koksharov, V.A., Fuzzy
logic and optimization of educational paths. IFAC-PapersOnLine, 51, 2, 511–
516, 2018.
14. Sanchez-Roger, Oliver-Alfonso, Sanchís-Pedregosa, Fuzzy logic and its uses
in finance: A systematic review exploring its potential to deal with banking
crises. Mathematics, 7, 11, 1091, 2019.
11
A Novel Fuzzy Logic (FL) Algorithm for
Automatic Detection of Oral Cancer
M. Praveena Kiruba bai1* and G. Arumugam2†
Department of Computer Science, Lady Doak College, Madurai,

1
Tamil Nadu, India

2
Department of Computer Science, Madurai Kamaraj University, Madurai,
Tamil Nadu, India
Abstract
A diagnosis system for biomedical applications is presented. Oral cancer is a com-
mon cancer that affects the people worldwide. A mathematical framework named
Adaptive Neuro Fuzzy Inference System (ANFIS) algorithm implemented focuses
primarily on the classification of oral cancer. The source oral images acquainted
are denoised for noise removal and enhanced for processing the images. The
enhanced images are transformed and classified using ANFIS classifier. The pro-
posed ANFIS model demonstrated enhanced performance metrics with 93% clas-
sification accuracy.
Keywords: ANFIS, oral cancer, classifier
11.1 Introduction
System modeling implemented using traditional mathematical techniques
is not well adapted for dealing with uncertain systems. A fuzzy inference
system, on the other hand, may mimic the qualitative features of reason-
ing processes and human knowledge without using exact quantitative
assessments. There are no established techniques for incorporating human
expertise into the rule set and database of a fuzzy inference system. Image

†
Corresponding author: [email protected]
173
processing is a thrust area that supports various fields like medicine,

astronomy, satellite imaging, and industrial applications. The most chal-
lenging work is the simulation of human visual system algorithms. These
algorithms involve modeling complex system with high degree of uncer-
tainty and subjectivity. A mathematical framework that incorporates these
tools is fuzzy logic. A novel architecture called Adaptive-Network-based
Fuzzy Inference System is proposed that serves as a model for classifying
the given inputs.
Image classification is a technique implemented for the classification of
images, performing pattern recognition. This paper analyzes the applica-
tion of the classifier, namely the Adaptive Neuro Fuzzy Inference System
(ANFIS). The classifier is used in the classification of medical images. The
classifier fuzzy inference system (FIS) framework is implemented on an
adaptive fuzzy neural network. It combines the knowledge of FIS and
an artificial neural network. The detection of oral cancer is the at most
demanding scenario that makes use of ANFIS classifier.
The essential steps in image processing are preprocessing. It removes the
noises and artifacts. The next step is enhancement used for improvising
the image. Feature extraction is implemented to extract the features in the
image. Classification is accomplished to detect the presence or absence of
oral cancer in MRI images based on the algorithm. Further segmentation
is carried out to segregate the region of interest from the classified image.
11.1.1 Significance of Pre-Processing

Pre-processing the images is mainly carried out using the following reasons:
• Images are to be appropriate on processing in the CAD

• The artifacts and the noises are to be removed
• The image quality needs to be improvised
As a pre-processing step, an adaptive median filter is used to the source

oral image in order to identify and remove any commotion-related material.
11.2 Image Enhancement

Low-resolution photos are recorded utilizing more expensive cameras
while taking oral images. The process of automatic classification is ham-
pered by many aberrant patterns in oral pictures. The adaptive local histo-
gram equalization method is applied to the filtered oral pictures to enhance
Novel FL Algorithm for Automatic Detection of Oral Cancer 175
the grey level intensity of each pixel. The primary goal of picture enhance-
ment is to raise the caliber of images.
11.3 Gabor Transform

The regular Fourier transform (FT) converts a signal from a spatial to a
frequency domain, when there is no link between any given pixel and the
region it represents in space. The Gabor change, which converts the spa-
tial space image into a multiple-pixel image that relates every pixel in spa-
tial, recurrence, and adequacy, overwhelms the confinement. The spatial
domain mode of the pixels in the improved oral images prevents their use
for direct feature extraction. As a result, the spatial domain oral images are
converted into a multiresolution mode utilizing the Gabor transform with
regard to its frequency, direction, and amplitude components.
Table 11.1 demonstrates that the PSNR value is superior to the median
and mean filters and that the adaptive median filter performs at 45.7 dB.
The adaptive median filter’s Mean Square Error value is 125.0, shows lesser
value than that the other filters. On source oral images, the proposed adap-
tive median filter delivers greater PSNR than the other filters. The edges are
preserved in adaptive median filter.
11.4 Image Transformation

Features are the traits that distinguish one object from another in an
image. Based on the attributes, the features are divided into general and
specific features. Generic features are common traits that can’t be used to
identify or categorize minute items in an image. Every object in an image
has unique features, and the unique features are the unique characteris-
tics of those objects. Therefore, this approach employs distinctive traits as
Table 11.1 Shows the performance of the adaptive

median filter.
Filter types PSNR (dB) MSE
Adaptive Median Filter 45.7 125.0
Median Filter 30.2 157.8
Mean Filter 30.0 163.8
opposed to generic qualities. Utilizing a combination of the Local Binary

Pattern (LBP) and the Grey Level Co-occurrence Matrix, the distinct fea-
ture attributes are retrieved from the Gabor-transformed magnitude oral
picture (GLCM). The feature combinations namely LBP and GLCM grants
better results with an accuracy of 94.5%, specificity as 91.9% and sensitivity
as 91.2% when compared with other feature combinations.
11.5 Adaptive Networks: Architecture

11.5.1 Classification of Images
The traditional method’s classification rate needs to be improved because
it is not optimal. Therefore, the phase uses ANFIS classifier approach for
segregating the oral images into either typical or cancer influenced oral
images. The NN classifier only generates output when the input patterns
are error-free. Adopting fuzzy rules with NN helps to get over the pro-
cess’ time constraint. In order to categorize the oral image into either typ-
ical or cancer-influenced oral picture, this phase uses the ANFIS classifier
technique.
The ANFIS classifier combines fuzzy logic and artificial neural net-
works. The ANFIS classifier receives input from the highlights of both
typical and cancer-influenced oral images that have been cropped out in
preparation mode. The precision of the arrangement is increased by cre-
ating an increasing number of regular and cancer-affected mouth images.
This ANFIS classifier (shown in Figure 11.1) has five internal layers, each
of which has 12 neurons and can be started after a number of training tri-
als. The ANFIS classifier is provided the extracted LBP and GLCM features
Layer:
0 1 2 3 4 5
A1
X0
A2
B1
X0
B2
Figure 11.1 ANFIS structure.

Novel FL Algorithm for Automatic Detection of Oral Cancer 177
Table 11.2 Performance metrics, ANFIS classifier with

different classification methods.
Classification methods Se (%) Sp (%) Acc (%)
ANFIS classifier 92.1 93.8 94.2
NN 84.6 89.2 89.6
SVM 87.7 87.2 90.2
Adaboost classifier 85.6 85.7 88.6
from the source oral picture as input in the classification mode. The output
is classified as cancerous and non-cancerous images.
11.6 Results and Discussions

The oral cancer detection system with respect to various classification
methods is analyzed using the metrics, namely sensitivity, specificity, and
accuracy.
From Table 11.2, it is understood that the proposed detection system for
oral images using ANFIS classifier delivers better cancer detection results
while compared with various state of the art approaches.
11.7 Conclusion
A mathematical model, ANFIS classifier is implemented to diagnose the
oral cancer MRI images. The images are taken from open access data set.
The images are preprocessed and enhanced for image clarity. The proposed
method is analyzed using the performance metrics namely sensitivity,
specificity and accuracy. The classifier achieved 92.1% of sensitivity, 93.8%
of specificity, and 94.2% accuracy than other classifiers.
Bibliography
Sharma, M. and Mukharjee, S., Artificial neural network fuzzy inference system
(ANFIS) for brain tumor detection. Computer Vision and Pattern Recognition,
arXiv:1212.0059, December 2012.
Odeh, S.M., Using an adaptive neuro-fuzzy inference system (ANFIS) algorithm

for automatic diagnosis of skin cancer, in European, Mediterranean and
Middle Eastern Conference on Information Systems. J. Commun. Comput.,
8, 9, 1–7, Januray 2011.
Lyakhov, P.A., Orazaev, A.R., Chervyakov, N.I., Kaplun, D.I., A new method for
adaptive median filtering of images, IEEE, 2019, 978-1-7281-0339-6/19/.
Pitas, I., Digital image processing algorithms and applications, Hoboken: A Wiley-
Inter Science Publication, 2002, ISBN: 9780471374077.
Sthevanie, F. and Ramadhani, K.N., Spoofing detection on facial images recogni-
tion using LBP and GLCM combination. J. Phys.: Conf. Ser., 971, 01 2014,
2018.
Malviya, A.M. and Joshi, A.S., Review on automatic brain tumor detection based
on gabor wavelet. Int. J. Eng. Res. Technol. (IJERT), 3, 1, 53–56, January – 2014.
Anuradha., K., Efficient oral cancer classification using GLCM feature extraction
and fuzzy cognitive map from dental radiograph. Int. J. Pure Appl. Math.,
118, 20, 651–656, 2018.
Fauzi, A.A., Utaminingrum, F. et al., Road surface classification based on LBP and
GLCM features using kNN classifier. Bull. Electr. Eng. Inform., 9, 4, August,
1446–1453, 2020.
Bourne, R., Image filters, in: Fundamentals of Digital Imaging in Medicine, Sydney,
Springer London, 2010.
Bovik, A., Handbook of image and video processing, Academic, New York, 2000.
12
A Study on Decision Making of
Difficulties Faced by Indian Workers
Abroad by Using Rough Topology
Nagadevi Bala Nagaram*, R. Narmada Devi , Kala Raja Mohan,
Regan Murugesan and Sathish Kumar Kumaravel
Department of Mathematics,Vel Tech Rangarajan Dr. Sagunthala R&D Institute of

Science and Technology, Feet Outer Ring Road, Avadi, Tamil Nadu, India
Abstract
Abroad jobs are the dream of all kind of people over a century. After getting their
jobs they may face practical problems. Few people give priority to work at abroad
due to their financial crisis or lead a luxury life. This article is a survey about the
merits and demerits faced by Indian workers abroad to decide to survey. The deci-
sion is concluded through rough topology.
Keywords: Rough set, rough topology, lower approximation, upper

approximation, basis
12.1 Introduction
Indian workers have been migrating to abroad over 100 years. For certain
reasons they may be migrated to abroad like to improve their financial sta-
tus, thirsty to work at abroad, to get exposure in the world, etc., In need of
this, their priority to work at abroad is in high frequency. Indians choose to
work abroad for four main reasons. They have the impression that they can
increase their savings. They think the western world has a decent lifestyle.
They think that working abroad will make them famous in front of their
kin.

179
However, there are other issues that Indian migrants confront at various
phases, which are made worse and more difficult by corruption, middle-
men, and scammers. Also they can feel like an outsider and face difficulties
while handling emergency situation at abroad.
Currently, 32 million NRIS and OCIS live outside of India. The larg-
est annual migration rate in the world is 2.5 million Indians who move
abroad each year. Here are a few succinct highlights of the main issues
Indian employees face:
12.1.1 Problems Faced by the Indian Workers

• Feeling like an outsider: Indian workers abroad can feel
like an outsider and feel loneliness abroad. Supporting their
parents, according to the findings, few NRIs are interested
to come back to India. Next on the list for NRIs is doing the
opposite, which is asking for help from family and friends.
It is true what they say—India has a strong sense of com-
munity and easy access to both material and emotional sup-
port. Most of them are single, so they missed their family
members for a long time. Among them, 24% get confused in
deciding to come back to India.
• Difficulties to get visa: Obtaining passports is not always
simple for migrants. For the local police to provide accurate
investigation reports, they must offer bribes. By overcharg-
ing for visas, providing inaccurate information about the
duration of the contract, the pay rate, overtime, and other
associated factors, the recruitment agencies of the sending
and receiving countries defraud them. Another source of
misery is the Indian diplomatic missions abroad. They do
not thoroughly investigate the legitimacy of the organiza-
tions offering visas. The majority of migrants from India are
uneducated blue-collar workers who move abroad in search
of work. Therefore, their first priority while working abroad
is to increase their income and send it home. So they may
fall into wrong hands when getting a visa.
• Handling Emergency Situation: Indian workers abroad
have difficulties back in their home country if they are uned-
ucated blue-collar workers because their passport is kept by
their agency. Due to emergency they are not permitted to
come back India. The Indian government has not yet estab-
lished a long-term plan with host nations to relocate its
Decision on Difficulties to NRI via Rough Top 181
employees. The quick evacuation of 1 to 2 million person-

nel poses a security risk in addition to being difficult. Apart
from that, other educated Indian workers can handle the
emergency situation without any difficulties.
• Pandemic Crisis: Indians who returned after losing their
jobs abroad during COVID-19 have no idea what to do
next because India lacks rehabilitation policy that can assist
them. The pandemic has made it harder for Indian migrant
workers abroad to survive. Gulf migrant laborers have not
received pay in months and are subject to fines for over-
staying their permits. Many people cannot afford the return
flights, and some are compelled to contact home and beg for
money.
Researchers have recently developed numerous models to include prob-

abilistic techniques into rough set theory, which was first established by
Pawlak, driven by a desire to describe information qualitatively.
Yiyu Yao and Yan Zhao discussed attribute reduction in decision-theoretic
rough set models [1]. Nasiri and Mashinchi expressed the accuracy of
decision rules on decision tables [2]. Thivagar et al. demonstrated the
usage of rough topology in the analysis of numerous practical/real-world
issues. Using this theory, they identify the determining elements for the
most widespread illnesses, diabetes and chikungunya [3]. Tao Yan and
Chongzhao Han proposed a rough topology method of attribute selection
using conditional entropy [4]. Zhang et al. discussed the expansions of the
rough set model and the context of their applications [5].
Nirmala Rebecca Paul made a medical diagnosis analysis using rough
topology [6]. Pooja Chaturveri discussed about the various concepts
involved in rough topology [8]. Kanchana and Rekha discussed about
corona virus diagnosis [9]. Gomathi et al. have made a study fuzzy using
its properties with local closed sets and local compactness [7, 10].
The survey about decision making on difficulties faced by INDIAN
workers abroad helps to take decision-related migration. Those who are
migrated from one place to another place will give their visa difficulties,
practical issues, surveying difficulties, food, savings, education for their
children, missing their families, etc., are taken into consideration. These
aspects are required to make decision to go for abroad are discussed. From
that, the important factors, which play a major role in making the decision
to go for abroad, are identified.
In this paper, section 12.2 describes the standard definitions applied in
rough topology. Section 12.3 presents the algorithm involved in the process.
Section 12.4 demonstrates the information system applied in this process.

Section 12.5 depicts the working rule applied with an example. Section
12.6 is about the conclusion followed by references.
12.2 Fundamental Idea of Rough Topology

The following standard definitions are applied in the analysis:
12.2.1 Conditional Attribute

The set of condition required to bring out the output are defined by condi-
tional attribute.
12.2.2 Decision Attribute

The output which is aimed for the study is named as decision attribute.
12.2.3 Rough Topology

The universal set and null set together with the lower approximation,
upper approximation, boundary region, and the outer region is defined as
the rough topology.
12.2.4 Lower Approximation

The nonempty set chosen from the target set, which is the union of equiv-
alent classes is lower approximation.
12.2.5 Upper Approximation

The collection of equivalent classes which has nonempty intersection with
the target set is upper approximation.
12.2.6 Boundary Region

The difference set between lower and upper approximation is referred as
the boundary region.
12.2.7 Basis
The collection of the universal set, null set together with the lower approx-
imation and the boundary region is the basis.
12.2.8 Information System

The data collected with respect to the attributes, from various persons
referred as objects, are arranged in the form of a matrix. The matrix is
called as information system.
12.2.9 Core
The rough topology and the basis set removing each attribute are framed.
The attributes which correspond to different basis corresponding to the
basis of the target set, are collectively known as the core.
12.3 Algorithm
The algorithm for this analysis consists of eight stages is as follows.
The set of condition and judgment attributes are determined in stage
1. At stage 2, to perform the analysis a table of information is created by
gathering the data pertaining to the qualities. The items that correspond to
the characteristics are represented by rows, whereas the attributes them-
selves are represented by columns. Followed by stage 2, the universal set U
and the subset X of U are framed in stage 3. Hence, lower approximation,
upper approximation, and boundary region are found at stage 4. In stage 5,
rough topology and basis are identified. Stages 4 and 5 are repeated remov-
ing each attribute one by one in stage 6. At stage 7, which of the basis are
different is identified. Finally, at stage 8, the core set is framed using the
results of stage 6.
12.4 Information System

To do this analysis, the number of condition attributes is framed as part of
the information system. This leads to formation process in order to obtain
the decision attribute. The decision property refers to whether they wish or
not the Indian to go for abroad. The next 10 traits are recognized as condi-
tion attributes in the process of decision making reaching the goal.
Q1) Working for financial crisis?

Q2) Food at abroad?
Q3) Stay with family or individual?
Q4) Missing their family?
Q5) If emergency are they allowed to home town?
Q6) Whether to lead luxury life?
Q7) Satisfied working there?
Q8) COVID-19 beginning of pandemic suffer a lot?
Q9) Friends or relatives to help there?
Q10) Face difficulties to get Visa?
Q11) Willing to come back to India?
Q12) Is foreign exchange of money easy?
The questionnaire raised with respect to the decision attribute is

“Working for Financial crisis?” the attribute D in the information system.
All of these characteristics are set up to have a yes/no response option.
Academicians have been given the questions about the aforementioned
Table 12.1 Statistical data received from applicants who stayed abroad from
2010 onward.
Object S1 S2 S3 S4 S5 S6 S7 S8 S9 S10 S11 S12 D
N1 0 0 1 1 1 0 1 1 1 1 1 0 1
N2 0 1 1 1 1 0 1 1 1 1 1 0 1
N3 0 0 1 0 1 0 1 1 1 1 1 0 1
N4 0 0 1 1 1 1 1 1 1 1 1 0 1
N5 0 0 1 1 1 0 1 1 1 1 1 0 1
N6 0 0 1 1 1 0 1 1 1 1 1 0 1
N7 0 0 1 1 1 0 0 1 1 1 1 0 1
N8 0 0 1 1 1 0 1 0 1 1 1 0 1
N9 0 0 1 1 1 0 1 1 1 0 1 0 1
N10 0 0 1 1 1 0 1 1 1 1 1 1 1
characteristics and a Google form in order to receive their responses. For

this survey, Indian workers abroad from Saudi Arabia, Ireland, Singapore,
Dubai, Malaysia, United States of America were taken into consideration.
A total of 50 data were collected. These statistics were gathered based
on applicants who stayed abroad from 2010 onwards. The outcomes of this
work will aid to make a better decision to go abroad and work there in
upcoming years. In order to achieve this, ten randomly selected responses
from the respondents are taken into account for this study and the relevant
information system is created. The 10 respondents are identified as N1, N2,
N3, N4, N5, N6, N7, N8, N9, and N10. In Table 12.1, the options that were
selected as yes are denoted as “1” and those that were selected as no are
denoted as “0.”
12.5 Working Procedure

The universal set is U = { N1, N2, N3, N4, N5, N6, N7, N8, N9, and N10}
The target set X = {S1, S4, S5, S6, S9, S10}
Equivalence Class is {S1, S5, S6} {S2} {S3} {S4} {S7} {S8} {S9} {S10}
Lower Approximation is RL = {S1, S4, S5, S6, S9, S10}
Upper Approximation is RU = {S1, S4, S5, S6, S9, S10}
Boundary Region RNR = {Ø}
Outside Region RNO = {S2, S3, S7, S8}
Rough Topology is {U, Ø, {S1, S4, S5, S6, S9, S10}, {S1, S4, S5, S6, S9, S10}, {Ø},
{S2, S3, S7, S8}}
Basis is {U, Ø, {S1, S4, S5, S6, S9, S10}, {Ø}}
By omitting each attribute, the aforementioned working process is

repeated. The basis that results from deleting the qualities S6, S10 and S12 is
discovered to be identical to the previously described basis. The basis that
was created after the qualities S2, S4, S7 and S8, are, respectively {U, Ø, {S4,
S9, S10}, {S1, S5, S6}}, {U, Ø, {S4, S9, S10}, {S1, S3, S5, S6}}, {U, Ø, {S4, S9, S10}, {S1,
S5, S6, S7}} and {U, Ø, {S4, S9, S10}, {S1, S5, S6, S8}} which are different from
the above mentioned basis. These attributes causes Indian are interested to
come back India and felt uncomfort.
12.6 Conclusion
Rough topology has attracted researchers in bringing out solution to deci-
sion making problems. In this paper, various aspects involved to work at
abroad is comfort or not are discussed. Using the method of rough topol-
ogy, the predominant attributes required for comfort at Indian workers at
abroad is identified.
References
1. Yao, Y. and Zhao, Y., Attribute reduction in decision-theoretic rough set
models. Inf. Sci., 178, 3356–3373, 2008.
2. Nasiri, J.N. and Mashinchi, M., Rough set and data analysis in decision
tables. J. Uncertain Syst., 3, 232–240, 2009.
3. Lellis Thivagar, M., Richard, C., Paul, N.R., Mathemcatical innovations of a
modern topology in medical events. J. Inf. Sci., 2, 33–36, 2012.
4. Yan, T. and Han, C., A normal approach of rough conditional entropy-based
attribute selection for incomplete decision system. Math. Probl. Eng., 2014,
1–15, 2014.
5. Zhang, Q., Xie, Q., Wang, G., A survey on rough set theory and its applica-
tions. CAAI Trans. Intell. Technol., 1, 323–333, 2016.
6. Paul, N.R., Rough topology based decision making in medical diagnosis. Int.
J. Math. Trends Technol., 18, 40–43, 2016.
7. Gomathi, G., Narmada Devi, R., Praba, B., “Nx locally compactness in fuzzy
rough topological spacession making in medical diagnosis. D Appl. Math.,
109, 302–310, 2016.
8. Chaturveri, P., Daniel, A.K., Khusboo, K., Concept of rough set theory
and its applications in decision making processes. Int. J. Adv. Res. Comput.
Commun. Eng., 6, 43–46, 2017.
9. Kanchana, M. and Rekha, S., Decision making using rough topology and
indiscernibility for corna virus diagnosis. Int. J. Sci. Res. Sci. Eng. Technol., 7,
31–33, 2020.
10. Gomathi, G., Narmada Devi, R., Sophia Ponmalar, D., A study on properties
of fuzzy rough topological spaces. Int. J. Adv. Res. Manage. Architecture Tech.
Eng., 7, 13–17, 2021.
13
Case Study on Fuzzy Logic: Fuzzy
Logic-Based PID Controller to
Tune the DC Motor Speed
Devendra Kumar Somwanshi *
Department of ECE, Poornima College of Engineering, Jaipur, Rajasthan, India
Abstract
The cost of DC motor is much higher than induction motor, but DC motors has
fantastic speed control attributes, and due to this attributes, it is highly utilized in
the industry. Many researches have done to control or to tune the DC motor speed.
To better control or tune the DC motor speed researches were started from the
basic PID controller. Then, many combinations were used with the PID control-
lers to tune better. In combinations, fuzzy logic controllers, fractional order fuzzy
tuning, gravitational search algorithms, genetic algorithm, micro controller-based
PID controllers, and many other methods were developed.
More than 50 papers are reviewed in several areas, such as to tune the DC
motor speed, controller design, and fuzzy-based tuning of boundaries to examine
and figure out current difficulties and extent of work.
In this chapter, the fuzzy PID controller is designed and analyzed to tune the
DC motor speed on the LabVIEW. The whole designing process is elaborated in
which all the steps are included that how fuzzy PID controller design, how the
fuzzificaion is done for the parameter values on the basis of membership func-
tions. The fuzzy inference system is developed on the basis of fuzzy-associated
memory table and rule base system. The values of FIS are defuzzified with the help
of defuzzification methods to find the values of gain parameters of PID controller.
After design, the analysis is done with respect to the output parameters, i.e., set-
tling time, rise time, and maximum overshoot. The results of the proposed system
is compared to the previous researcher’s outputs. After the comparison, it is found
that damping ratio, rise time, settling time, and peak time are improved from 10%
to 50%. From the results, it can undoubtedly be concluded that fuzzy PID can give
Email: [email protected]
187
the significantly better boundaries then different controllers, like PID controllers
and fuzzy controller.
Keywords: FLC, fuzzy PID controller, DC motor, speed control
13.1 Introduction
Because of magnificent speed control attributes, DC Motor has been
broadly utilized in the available motors in the business. Several techniques
are used to control the speed of DC Motor in which PID controller is
mostly used with combination of different AI techniques such as Fuzzy
Logic, Neural Network, Genetic Algorithm, Micro controller based tuning,
etc. [2, 5–7].
The most generally utilized strategy of speed control of DC Motor is
PID Controller in view of the Fuzzy Method. The show of PID Controller
was connected with the setting of boundaries, for example the Proportion
(P) in addition to the Integration (I) in addition to the Derivation (D). The
reason for Tuning to demonstrate the way that settings for Controllers can
be gotten from data of the cycle to be controlled as displayed in Figure 13.1.
The best part of tuning of PID controller based on Fuzzy logic is that
the Fuzzy Controller was nonlinear and versatile in nature giving power-
ful execution under boundary varieties and burden aggravation impact.
Tuning of the PID Controller boundary is finished with the different tuning
strategies like ANT Algorithm, Swarm Optimization, Genetic Algorithm,
Gravitational Search Algorithm and so forth. Plan of Fuzzy PID Controller
Proportional
Gain
+
1 I 1 + 1
s +
e Integral Integrator u
Gain
D du/dt
Derivative Derivative
Gain
Figure 13.1 Block diagram PID controller.

DC Motor Performance Optimization 189
experiences troubles in the determination of streamlined Membership

Functions and the Fuzzy rule base, which was customarily accomplished
by a drawn-out experimentation process.
13.1.1 DC Motor
An electric Motor is a device which works on electrical energy and cre-
ate mechanical energy on the communication of attractive fields and flow
conveying transmitters. By a generator the converse course of switching
mechanical energy over completely to deliver electrical energy is achieved.
Figure 13.2 shows the opposite course of changing mechanical energy over
completely to create electrical energy is achieved by a generator.
13.1.2 DC Motor Speed Control Methods

13.1.2.1 PID Controller
A PID controller is based on closed loop controlling mechanism in which
feedback is used to calculate the error between desired output and mea-
sured output. Based on error, correction is applied to the P, I, PI, PD PID
part.
13.1.2.2 Fuzzy-Based PID Controller

In this controller the tuning is based upon the rule base and knowledge
base of fuzzy controller [17, 22, 29]. The rule base is created on the basis of
knowledge, desired output and error to make the controller more sensitive
to give the output. On the basis of tuning the kp ki, and kd are changed so
that self-tuning mechanism is created [19, 28].
Ra La Ia
θ Jm
Va Eb Tm
Bm
Figure 13.2 DC motor model.

13.1.2.3 Micro Controller-Based PID Controller

The Micro Controller is arranged with 32 pieces drifting point Algorithm.
The execution of the Algorithm was finished in CPU MC9S12, from free
scale and was tried on two assessment sheets, the first with the Micro
Controller mc9s12e64, and the second with the mc9s12gc32. The heartbeat
wide balance yield (PWM) is utilized to create the control signal while uti-
lizing Micro Controller [14]. For the instance of Fuzzy control this worth
is 10ms without really any need of an inner counter. Consequently, the
inspecting happens 100 times each second [10].
13.1.2.4 Genetic Algorithm-Based PID Controller

In genetic algorithm-based tuning PID controller factor, the PID con-
trolling factors kp, ki and kd are coded in string structure [13, 27, 30]. To
code the string 1s and 0s are used. Once all the deciding factors are coded
in the string, the mating pool is created by which mutation on strings are
performed to create the new string on the basis on mutation. The final
strings denoted to the PID controlling factors are decoded so that PID con-
troller is tuned on the basis of the factors.

13.2.1 Common Findings
Common findings are extracted from the literature review of the topic
“Speed Control of DC Motor using PID Controller” and these are as
follows:
• To tune the speed of motor, most of the researchers worked

on PID Controller and Mostly in MATLAB Software.
• Armature current and the terminal voltage were mostly used
control parameters to tune the motor speed.
• Error and change in error were taken as input parameters to
control or tune the motor speed.
• Higher rating up to 4000rpm researchers used Buck con-
verters with PID Controllers [15].
• Authors used Algorithm to tune the PID Controller as
Genetic Algorithm, PSO Algorithm, Ant colony optimization
Algorithms, Gravitational Search Algorithm. In which

mostly GA (Genetic Algorithm) used widely for Tuning.
• The duty cycle can be varied from 0-255 for speed control
of Motor when using AT mega 8A Micro Controller in
LABVIEW [23].
• Steady state error obtained ±6 rpm in fuzzy self-tuning
methodology [3].
• PID-OP amp technique was allowed the Motor to be driven
at the maximum current load of 50amp for 5sec before shut-
ting down the Motor [20].
13.2.2 Comparative Analysis of Research Works Reviewed

In this section comparative analysis of literature has been done on the
basis of research works carried out in the area of DC motor speed con-
trol or tune. In this area researchers used various Controllers, approaches
& Algorithm for this issue. They are PID, fuzzy, neuro-fuzzy, fuzzy PID,
genetic algorithm, dynamic SMC controller, micro controller, gravitational
search algorithm, ANFIS technique, etc., in which fuzzy-based PID con-
troller able to control speed in very less time with zero steady-state error
and low settling and rise time for every kind of input. Table 13.1 shows
comparison of speed control approach for DC motor.
13.2.3 Strengths in the Literature Reviewed

• Reduction technique was used in FPID Controller to reduce
the higher order Transfer Function in 2nd-order transfer
function [21].
• Digital Controller had maximum operating frequency about
51.509 MHz and that gave the advantage of the high speed
achievable using hardware for the speed of 700–1,400 rpm
and also reduced overshoot to 1.77% [8].
• From the PID controller overshoot was decreased from
5%.for Motor speed of 1750 rpm in the regulatory control
system [4].
• Hybrid fuzzy PID based on genetic algorithm gave zero
overshoot and 1.5% reduction in settling time for speed
varying from −600 to 600 rpm for a DC motor [1].
• From fuzzy PID controller based on 16-bit micro controller
without any internal counter calculated speed in 10 ms [10].
Table 13.1 Comparison table.

Tuning parameter
for controller/
technique/ Technique /
Type of algorithm algorithm/ I/p for DC Motor
Ref nu. Parameter controller specification method motor specification Result
[3] Speed/ PID Controller kp=3, ki=5, kd=1 Soft computing Voltage signal Voltage rating=12v Steady state error
terminal –Fuzzy from DAQ Current obtained minimum
voltage Logic based cards range rating=1.5amp ±6rmp & overshoot
2.4v-12v Rated RPM=1500 obtained 0% due to
Ra=8.8Ω integral action
[26] Speed/ Adaptive PID kP = 1, ki=5 , kd =1 Dynamic SMC Load torque of Not specified Chattering problem is
armature Controller Performance Controller 0.3nm reduced with the
current parameter = 0.034 help of this technique
Learning parameter
=1
[18] Speed/field PID kP = 136.4, ki = 1287 Genetic Output of TF= 0.01/0.05s2 + Transfer Function is
current Controller/ kd = 3.62 Algorithm/ FFPID 0.06s + 0.1001 reduced in 2nd order,
Fuzzy GA parameters:- Ziegler steady state error
Controller Population size=20 Nichols reduced up to 0.014%
Generation= 62 method with disturbances.
Cross over
fraction=0.8
(Continued)
Table 13.1 Comparison table. (Continued)

Tuning parameter
for controller/
[23] Power cycle PID Controller Flash memory=8kb AT mega 8A O/p of the driver Voltage rating=12v The technique helps to
on-off ratio/ ISP Micro circuit Current maintain stability of
speed EEPROM=512B Controller rating=1.5amp the system
23GPI/O based
[15] Speed/ PI-like Fuzzy Error and change of Buck convertor/ O/p voltage Voltage rating=24v Overshoot=2.05% & rise
Counting Controller error Genetic from buck Current = 2.1amp time 20ms reduced.
pulse from Algorithm convertor Rated RPM=4000 Also worked for higher
encoder ratings Motor
[25] Speed/ PID Controller Initial value of GSA Gravitational 3rd order transfer Reduced settling
armature Go=100 Search function time0.00000458sec,
current/ Total iteration=200 Algorithm R= 5ohm rise time
field current N= 100 L= 0.025 Hennery 0.000000257sed and
overshoot 0.0002%
than ZN method.
[12] Speed/ Fuzzy Input parameter of Neuro-Fuzzy Error and error Power= 200w Reduced the settling
terminal Controller error and change method change ratio Current=3.5amp time about 10% with
voltage/ in error= +100 to calculated by Rated RPM=3000 disturbances of load.
varying -100 comparator
load
(Continued)
Table 13.1 Comparison table. (Continued)

Tuning parameter
for controller/
[10] Speed/ PID Floating point input 16 bit Micro From Micro Second order Sampling time obtained
armature Controller/ parameters of Controller Controller transfer in 10ms for Fuzzy
current Fuzzy Algorithm for based/ bits function control
Controller Micro Controller floating
point
format
[11] Speed/ PID Controller kP = 19.88 ,ki = 0.1376 ANFIS From shunt field TF= Settling time reduced
armature kd = 0.5578 technique Voltage=230v up to 0.5 sec and
resistance Rated RPM=1500 maximum overshoot
and Current ratings= 8a is zero for third order
inductance Transfer Function
• PWM technique was used in neuro-fuzzy controller for volt-

age control of DC motor and obtained settling time in 0.19
sec with zero overshoot [12].
• Motor speed up to 4000 RPM fuzzy-PI controller based on
buck converter showed considerable improvement in startup
response and changes in speed with the very small overshoot
of about 20 ms [15].
13.2.4 Weaknesses in the Literature Reviewed

• The working voltages of Buck converters were between 10 v
dc and 20 v, which was a major drawback. It was introduced
costly technique due to real implementation of the control-
ler on DSP processor [15].
• The disadvantage of the neuro-fuzzy controller was that the
controller is only about 60% of the value of the current peak
of PID controller, which was less efficient to tune motor
speed [12].
• Proposed GSA-based PID controller used root locus analy-
sis for position control to tune the motor speed and was very
complex to design [25].
• In fractional fuzzy PID controller using two inputs together
as square and step input for the purpose to tune the motor
speed then the controller gave higher overshoot of 13.76%
with disturbances [18].
13.2.5 Findings in the Literature Reviewed

• Less research work in the fuzzy self-tuned PID control
method to tune the motor speed.
• Most of the researchers used MATLAB Software.
• Widely the algorithm like Genetic, PSO, and Ant colony
optimization were used [13, 27].
• Mostly techniques provided higher overshoot in the pres-
ence of disturbances and also providing the zero overshoot
but higher settling time in no disturbances condition.
• Commonly application based PID Controllers were
designed.
13.3 Design of Fuzzy-Based PID Controller

13.3.1 Fuzzy Controller
In fuzzy logic (FL) straightforward rule-based utilizing IF X AND Y
THEN Z approach is utilized to control issue as opposed to endeavoring
to numerically display a framework. At the point when linguistic and sub-
jective attributes address of this present reality in registering then fuzzy
logic gives another option to deal with this. By dealing with a mathemat-
ical boundaries to work, for example, what is viewed as huge mistake and
critical pace of-progress of-blunder, yet careful upsides of these numbers
are typically not basic except if extremely responsive execution is expected
in which case experimental Tuning would decide them. Figure 13.3 shows
the process of Fuzzy logic.
The Fuzzy Controller tunes the PID controller on two inputs i.e. Error
and change in Error.
Error êt = Rt – Ut (13.1)
Change in Error đêt = êt – ê (t-1) (13.2)
13.3.2 Flowchart for Fuzzy Controller

Figure 13.4 shows the flow chart of the working of fuzzy controller.
Knowledge Base
Rule Base Data Base
e(t)
Output
Fuzzification Decision-making Defuzzification
Input
Interface Unit Interface
de(t)
Figure 13.3 Block diagram.

Start
Select Parameters
Design DC Motor Transfer

Function
Select Fuzzy Controller
Design Membership Function

and Create Rules for Fuzzy
Controller
Simulate Using LABVIEW
Meet
specification
Desired Speed
Collect the Results
End
Figure 13.4 Flow graph of fuzzy controller.
13.3.3 Fuzzy Logic Controller Membership Function

and FAM Table
The FLC works on linguistic values, and real world works on crisp val-
ues, therefore all the parameters values of FLC must be converted from
crisp to linguistic values. For this conversion, fuzzy membership function
is defined. The membership values lies between 0 and 1. Membership func-
tion is used to map the crisp value of input space to the membership values
and draw a curve on the basis of mapping. For proposed method, trian-
gular function is used as membership function for all input and output
variables. For proposed approach, five membership functions are defined
for error between −10 to +10 values. Figure 13.5 shows the membership
function for error, and Table 13.2 shows the ranges of the membership
functions.
Membership functions graph

1 error
PA
0.8
PB
Membership (u)
NA
0.6
NB
CERO
0.4
0.2
0
–10 –8 –6 –4 –2 0 2 4 6 8 10
Range
Figure 13.5 Membership function for error.
Table 13.2 Defining membership functions range for error.

Sr. no. Membership function Range
1 Negative big (NA) 0–10
2 Negative large (NB) −6 to 4
3 Positive big (PA) −10 to 4
4 Positive large (PB) −2 to −4
5 Zero −6 to 10
In Table 13.2, all membership functions range defined for input variable,
error five membership functions designed in the range of 0–10 interval in
which negative big is 0 to 10, negative large is −6 to 4, positive big is −10 to
4, positive large −2 to 4, zero is −6 to 10.
Further, two membership functions are defined for the input variable
change in error between 0 and 100. Figure 13.6 shows the membership
function, and Table 13.3 shows the membership values.
In Table 13.3, all membership functions range defined for input variable
change in error two membership functions designed in the range of 0 to
100 interval in which negative big is 0 to 60, negative large is 30 to 100.
Seven Membership Functions with seven linguistic terms designed for
output variable for speed. Figure 13.7 shows the membership function. The
interval for membership functions used 0 to 30.

1 Change in error
ENA
0.8
EPA
Membership (u)
0.6
0.4
0.2
0
0 10 20 30 40 50 60 70 80 90 100
Range
Figure 13.6 Membership function for change in error.
Table 13.3 Defining membership functions range.

Sr. no. Membership function Range%
1 Negative big (ENA) 0–60
2 Negative large (EPA) 30–100

1 Desired Speed
DM
0.8
DB
Membership (u)
STR
0.6
AM
AP
0.4
MD
MP
0.2
0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
Range
Figure 13.7 Membership function for output value.
In Table 13.4, all membership functions range defined for input variable
error seven membership functions are designed in the range of 0 to 20
interval in which negative big is 0 to 10, negative large is 0 to 6, negative
medium is 10 to 20, positive big is 10 to 26, positive large 16 to 30, zero is
6 to 20.
Table 13.4 Defining membership functions range for desired speed.

Sr. no. Membership function Range
1 Negative Big (DM) 0–10
2 Negative Large (DB) 0–6
3 Negative Medium (AM) 10–20
4 Positive Big (AP) 10–26
5 Positive Large (MD) 16–30
6 Positive Medium (MP) 20–30
7 Zero (STR) 6–20
13.3.4 Rules for the Fuzzy Controller

To design the fuzzy logic controller, fuzzy-associated memory (FAM)
Table, FAM Table is required. The FAM Table is defined the each rules
between the input and output variables directly and in simple way. Using
this table, two variables can be mapped to a two-dimensional matrix.
Based on the FAM table, the rules are generated to find the output. Figure
13.8 shows the FAM table for fuzzy controller. The best thing of rules that
multiple rules may be used at a point and the rule contribute to gener-
ate the output more who has stronger effect. Once the output is generated
by the rule base method, it is converted to the crisp value by defuzzifica-
tion method. Many standard defuzzification methods are defined. As per
requirements, suitable method is used to defuzzify the rule base output.
đêt VERY
LOW HIGH VERYLOW MEDIUM
t HIGH
ê
LOW FAST SLOW VERY FAST SLOW FAST
VERY VERY
HIGH SLOW MEDIUM SLOW
SLOW SLOW
VERY VERY
MEDIUM VERY FAST MEDIUM FAST
LOW FAST
VERY VERY VERY VERY
SLOW SLOW
HIGH SLOW SLOW SLOW
MEDIUM FAST FAST FAST FAST FAST
Figure 13.8 FAM table for fuzzy controller.

To tune the variable kp, ki, kd, values are received by the output of rule base
defuzzification system.
Two input variables are taken for rules design as Error êt and Change
in Error đêt and one output variable which is actual speed ut. All rules
designed between three variables for consequent results.
1. IF ‘êt’ IS ‘LP’ AND ‘đêt’ IS ‘LP’ THEN ‘ut’ IS ‘Actual Speed

2. IF ‘êt’ IS ‘HP’ AND ‘đêt’ IS ‘LP’ THEN ‘ut’ IS ‘Actual Speed’
3. IF ‘êt’ IS ‘VLP’ AND ‘đêt’ IS ‘LP’ THEN ‘ut’ IS ‘Actual Speed’
4. IF ‘êt’ IS ‘VHP’ AND ‘đêt’ IS ‘LP’ THEN ‘ut’ IS ‘Actual Speed’
5. IF ‘êt’ IS ‘MD’ AND ‘đêt’ IS ‘LP’ THEN ‘ut’ IS ‘Actual Speed’
6. IF ‘êt’ IS ‘LP’ AND ‘đêt’ IS ‘HP’ THEN ‘ut’ IS ‘Actual Speed’
7. IF ‘êt’ IS ‘LP’ AND ‘đêt’ IS ‘HP’ THEN ‘ut’ IS ‘Actual Speed’
8. IF ‘êt’ IS ‘VLP’ AND ‘đêt’ IS ‘HP’ THEN ‘ut’ IS ‘Actual Speed’
9. IF ‘êt’ IS ‘VHP’ AND ‘đêt’ IS ‘HP’ THEN ‘ut’ IS ‘Actual Speed’
10. IF ‘êt’ IS ‘MD’ AND ‘ đêt’ IS ‘HP’ THEN ‘ut’ IS ‘Actual Speed’
11. IF ‘êt’ IS ‘LP’ AND ‘đêt’ IS ‘VLP’ THEN ‘ut’ IS ‘Actual Speed’
12. IF ‘êt’ IS ‘HP’ AND ‘đêt’ IS ‘VLP’ THEN ‘ut’ IS ‘Actual Speed’
13. IF ‘êt’ IS ‘VLP’ AND ‘đêt’ IS ‘VLP’ THEN ‘ut’ IS ‘Actual Speed’
14. IF ‘êt’ IS ‘VHP’ AND ‘đêt’ IS ‘VLP’ THEN ‘ut’ IS ‘Actual
Speed’
15. IF ‘êt’ IS ‘LP’ AND ‘đêt’ IS ‘VHP’ THEN ‘ut’ IS ‘Actual Speed’
16. IF ‘êt’ IS ‘HP’ AND ‘đêt’ IS ‘VHP’ THEN ‘ut’ IS ‘Actual Speed’
17. IF ‘êt’ IS ‘VLP’ AND ‘đêt’ IS ‘VHP’ THEN ‘ut’ IS ‘Actual
Speed’.
18. IF ‘êt’ IS ‘MD’ AND ‘đêt’ IS ‘VHP’ THEN ‘ut’ IS ‘Actual Speed’
19. IF ‘êt’ IS ‘VHP’ AND ‘đêt’ IS ‘VHP’ THEN ‘ut’ IS ‘Actual
Speed’
20. IF ‘êt’ IS ‘LP’ AND ‘đêt’ IS ‘MD’ THEN ‘ut’ IS ‘Actual Speed’
21. IF ‘êt’ IS ‘HP’ AND ‘đêt’ IS ‘MD’ THEN ‘ut’ IS ‘Actual Speed’
22. IF ‘êt’ IS ‘MD’ AND ‘đêt’ IS ‘MD’ THEN ‘ut’ IS ‘Actual Speed’
23. IF ‘êt’ IS ‘VLP’ AND ‘đêt’ IS ‘MD’ THEN ‘ut’ IS ‘Actual Speed’
24. IF ‘êt’ IS ‘VHP’ AND ‘đêt’ IS ‘MD’ THEN ‘ut’ IS ‘Actual Speed’
25. IF ‘êt’ IS ‘LP’ AND ‘đêt’ IS ‘LP’ THEN ‘rt’ IS ‘Desired Speed’
26. IF ‘êt’ IS ‘LP’ AND ‘đêt’ IS ‘VLP’ THEN ‘rt’ IS ‘Desired Speed’
27. IF ‘êt’ IS ‘VLP’ AND ‘đêt’ IS ‘LP’ THEN ‘rt’ IS ‘Desired Speed’
28. IF ‘êt’ IS ‘MD’ AND ‘đêt’ IS ‘LP’ THEN ‘rt’ IS ‘Desired Speed’
29. IF ‘êt’ IS ‘MD’ AND ‘đêt’ IS ‘VLP’ THEN ‘rt’ IS ‘Desired
Speed’ ALSO ‘u(t)’ IS ‘Actual Speed’
13.3.5 Simulation Diagram of FLC

The following steps are used to design the FLC:
• Use the fuzzy logic controller from the control & simulation
block,
• Selecting the signal input and output,
• Defining the membership function for all variables,
• Design rules for all individual membership functions,
• Give path to fuzzy controller,
• Simulate the program.
In simulation, transfer function is finalized first and on the basis of the

transfer function error calculated and processed to tune the PID Controller
and FLC. Simulation of FLC in LabVIEW is shown in Figure 13.9.
1 %Motor plant Model

0.028 K
2 num=[K];
1.8E-6 B 3 den=[J*L B*L+J*R B*R+K^2]; Equation
4 plant=tf(num,den); plant
9.64E-6 J
G
3.3 R
0.00464 L
Control & Simulation Loop
(
no Error
proptional 1
integral Integrator
Gain Summation 2
Summation 1 + Transfer Function Waveform Chart 2
Step Signal S + =
+ H(s)
k
derivative
G Derivative 2
k s
1
(
Summation 3 Transfer Function 2
+ = 1
– Summation 4 H(s)
MISO
+ =
–
( Derivative
Halt Simulation
1 Halt?
s
file path FL Load Fuzzy System.vi
Path
Figure 13.9 Simulation of fuzzy controller in LABVIEW.

13.3.6 Fuzzy-Based PID Controller

It is very well known that stand-alone FLC is not the perfect solution to
tune the speed of a motor. To get better result, fuzzy PID controller is used.
Figure 13.10 shows the block diagram of the controller. This will give the
optimization result for the control parameters which needed to tune the
controller. The specialty of FPID is that in it separate rule base are defined
for gains in proportional, derivative, and integral part of PID which make
the response of the controller faster [12]. From the block diagram, it is seen
that fuzzy controller tune the gain constants of P, I, and D parts of PID
controller on the basis of error and change in error.
13.3.6.1 Fuzzy Block Design

Figure 13.11 shows the block diagram of fuzzy, it is already explained in
previous section, how crisp values are converted on the basis of member-
ship function to the fuzzy values and after that rule base is design to infer
the output fuzzy values. These fuzzy values are converted to the crisp val-
ues by defuzzification methods.
In the design of fuzzy controller individually design for below controllers:
• Fuzzy proportional
• Fuzzy integral
• Fuzzy derivative
• Fuzzy PID
r PID y
Plant
+ Controller
–
d/dt dKp dKi dKd
Fuzzy Controller
Figure 13.10 Fuzzy based PID controllers.
Fuzzy Block
Fuzzy rules
e(t) kp
Fuzzification Inference Defuzzification ki
Ce(t) kd
Figure 13.11 Flow chart of fuzzy block.

Table 13.5 FAM table for fuzzy tuned PID controller.

đêt VERY VERY
ê t
LOW HIGH LOW HIGH MEDIUM
VERY
LOW FAST SLOW SLOW FAST
FAST
VERY VERY
HIGH SLOW MEDIUM SLOW
SLOW SLOW
VERY VERY VERY
MEDIUM MEDIUM FAST
LOW FAST FAST
VERY VERY VERY VERY
SLOW SLOW
HIGH SLOW SLOW SLOW
MEDIUM FAST FAST FAST FAST FAST
In the FAM Table, each rule is designed between the input and output
variables directly and in a simple way. It’s a tabular way to express Fuzzy
Logic rules.
All rules for proportional control, integral control and derivative con-
trol are design by using FAM table. Table 13.5 shows the FAM table of
FPID controller.
13.3.6.2 Flowchart for Fuzzy-PID Controller

Figure 13.12 shows the process of FPID controller to tune the PID gain
parameters and to control the speed of DC motor.
13.3.6.3 Simulation Diagram of Fuzzy-PID Controller

Figure 13.13 shows the simulation diagram of FPID controller in LabVIEW.
(0.028)
TF
(4.47296 E 8s 2 3.18204 E 5S 0.0078994)
A fuzzy-based PID controller to tune the speed of DC motor was

designed. A comparison is done between performance parameters for
the same technique and proposed work by using same transfer function.
Start
kp, ki, and kd Calculation according to the Ziegler Nichols method
Reference Signal tends to the FLC as Error signal and change

in error signal to the fuzzy inference system
Fine tune the Fuzzy Controller by varying the parameter

and\or modifying the rules suitably
Fuzzy individually tune gain constant of PID Controller
Desired
Parameters
Obtain
No
Record Results as Gain Parameters, Rise Time, Settling Time
END
Figure 13.12 FPID controller flowchart.
Figure 13.14 shows the response of proposed FPID controller. Table 13.6
shows the response of the proposed FPID controller.
13.4 Experimental Work and Results Analysis
• Proposed work experimental analysis has been divided in

two parts. In which first part is development of controllers,
second is comparison with the previous work.
• Development of controllers is further divide in three sec-
tions in which first is PID controller design, second Fuzzy
Controller design and third fuzzy-PID controller design.
1 %Motor plant Model

0.028 K
2 num=[K];
1.8E-6 B 3 den=[J*L B*L+J*R B*R+K^2];
Equation
4 plant=tf(num,den); plant
9.64E-6 J
G
3.3 R
0.00464 L
file path FL Load Fuzzy System.vi
Path
Control & Simulation Loop
(
no Error
proptional 1
integral Integrator
Gain Summation 2
Summation 1 + Transfer Function Waveform Chart 2
Step Signal S + =
+ H(s)
k
derivative
G Derivative 2
k s
1
(
Summation 3 Transfer Function 2
+ = 1
– Summation 4 H(s)
MISO
+ =
–
( Derivative
Halt Simulation
1 Halt?
s
Transfer Function 3
H(s)
output value
Summation 5 ( 1.23
+ = 1 MISO
–
6 output value 2
+ = 1.23
–
( Derivative 3
MISO
1 s
output value 3
1.23
MISO
file path 1
Path
file path 2
Path
file path 3
Path
Figure 13.13 Fuzzy-based PID controller in LABVIEW.
• For comparison with the previous work based on same tech-

nique a Fuzzy-PID controller is designed using number of
rules with the new membership functions.
• In proposed FPID controller, the DC motor speed tuning
parameters are 0.007 sec, settling time 0.066 sec and peak
time is 0.0025 sec. Table 13.6 shows the same.
1.6
1.4
Step input
1.2
1 PID Response
Amplitude
0.8
FUZZY
0.6 Response
0.4 Fuzzy-PID
Response
0.2
0
Time (ms) 1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96
Figure 13.14 Step response for fuzzy-based PID controllers.
Table 13.6 Response of proposed fuzzy-PID.

Ref. Gain Rise Settling Peak Damping Gain
no. parameters time(s) time time ratio margin
[31] kp=24.0021 ki 0.007 0.0066 0.0025 0.60 ∞
= 24.9809
kd = 2.8855
• When compared with the previous work like FPGA-based

fuzzy controller settling time decreases 95% but rise time
increases 55% from proposed work, for a nonlinear sys-
tem rise time increases 77% but settling time improved by
86.11% and for the DSP-based controller rise time decreases
by 67.5%, settling time increases by 40.91% when compared
with the proposed work. Table 13.7 shows the same.
• Fuzzy-PID controller gives best result with comparison than
other controllers. The proposed FPID controller is simulated
in LabVIEW; therefore, the time response is better than
other simulation environment.
13.5 Conclusion and Future Scope

The future work of the proposed work is as follows:
• Some specific DC Motor can be taken to tune their speed.

• Other AI techniques, i.e., neural network, genetic algorithm
can be combined with fuzzy technique.
Table 13.7 Comparison between same technique and proposed work.

Proposed fuzzy-PID Researchers
Gain Rise Settling Gain Rise Settling

Ref. Transfer function Application parameters time time parameters time time Result
[9] 50000 Controlled kp = 2.3 0.7 0.45 kP = 0.6 6.5 0 Rise time reduced
Object ki = 3.0 ki = 0.5 to 0.7s from
(S 3 80S 2 15000S)
kd = 1.24 kd = 0.001 6.5s, settling
time obtained in
0.45s.rise time
reduced by 89%.
[8] 0. 6 FPGA kp = 3.0 0.009 0.0066 - 0.0004 0.16 Settling time

Based ki = 3.0 decreases 95%
(0.0001s 2 0.0051s 0.365)
DC kd = 2.3 but rise time
Motor increases 55%.
[16] 2 DSP Based kp = 24 0.4 0.31 kp = 20 0.13 0.22 Rise time decreases
DC ki= 3.0 ki = 1.35 by 67.5%,
S 2 3S 1
Motor kd = 2.3 kd = 3.5 settling time
increases by
40.91%.
[24] 5 Nonlinear kp = 2.9 0.9 0.72 kp= 4.82 0.2 10 Rise time increases
System ki = 3.1 ki = 3.5 to 77% but
s3 4.5s 2 5.5s 15
kd = 3.1 kd = 3.82 settling time
improved by
86.11%
References
1. Rubaai, A., Castrom Sitiriche, M.J., Ofoli, A.R., DSP-based laboratory imple-
mentation of hybrid fuzzy-PID controller using genetic optimization for
high-performance motor drives. IEEE Trans. Ind. Appl., 44, 6, 1977–1986,
IEEE, 2008.
2. Arulmozhiyal, R., Design and implementation of fuzzy PID controller for
BLDC motor using FPGA. 2012 IEEE International Conference on Power
Electronics, Drives and Energy Systems (PEDES), Bengaluru, IEEE, pp. 1–6,
2012.
3. Gowthwan, E., Surya, M., Cominos, P., Munro, PID controllers: Recent tun-
ing methods and design to specification. Control Theory Appl., 149, 1, 46–53,
2013.
4. Abhinav, R. and Sheel, S., An adaptive, robust control of DC motor using
fuzzy-PID controller. 2006 IEEE International Conference on Power
Electronics, Drives and Energy Systems (PEDES), IEEE, pp. 1–5, 2006.
5. Wu, H. and Su, W., PID controllers: Design and tuning methods. Industrial
Electronics and Applications (ICIEA), 2014 IEEE 9th Conference on, IEEE, pp.
808–813, 2014.
6. Kanojiya, R.G. and Meshram, P.M., Tuning of PID controller using Ziegler-
Nichols method for speed control of DC motor. IEEE-International
Conference on Advances in Engineering, Science and Management (ICAESM
-2012), Nagapattinam, Tamil Nadu, IEEE, pp. 117–122, 2012.
7. vas, P. and Bulale, M.J.E., Design and implementation of fuzzy based PID
controller. Industrial Technology, IEEE ICIT ‘02. 2007 IEEE International
Conference on, vol. 1, IEEE, pp. 220–225, 2007.
8. Ali, F.H. and Mahmood, H.M., LabVIEW FPGA implementation of PID con-
troller for DC motor speed control. 1st International Conference on Energy,
Power and Control (EPC-IQ), IEEE, pp. 1–6, 2010.
9. Kan, W.X. and Liang, S.Z., Design and research based fuzzy-PID parame-
ters self-tuning controller with MATLAB. Advanced Computer Theory and
Engineering, 2008. ICACTE ‘08. International Conference on, IEEE, pp. 996–
999, 2008.
10. Cesar, M. and Abdelrahman, A., Speed control of DC motor using fuzzy
logic controller. 2008 International Conference on Communication, Control,
Computing and Electronics Engineering (ICCCCEE), Khartoum, IEEE, pp.
1–8, 2008.
11. Chen, G.S., Particle swarm optimization for PID controllers with robust test-
ing. International Conference on Machine Learning and Cybernetics, Hong
Kong, IEEE, pp. 956–961, 2007.
12. Huang, Y. and Yasunobu, S., Design of neuro-fuzzy controller. 9th
International Symposium on Communications and Information Technology,
Icheon, IEEE, pp. 808–813, 2009.
13. Kamal, M.M., Mathew, L., Chatterji, S., Speed control of brushless DC
motor using genetic algorithm. 2014 Students Conference on Engineering and
Systems, Allahabad, IEEE, pp. 1–5, 2014.
14. Ming, Z.X. and Liu, L.S., Simulation study of fuzzy-PID controller for DC
motor based on DSP. 2012 International Conference on Industrial Control and
Electronics Engineering, IEEE, pp. 3401–3405, 2012.
15. Kadwane, S.G. and Gupta, S., Real time based PI-fuzzy controller for DC
servomotor. 2006 International Conference on Power Electronic, Drives and
Energy Systems, New Delhi, pp. 1–4, 2006.
16. Yongjuan, Z. and Yutitan, P., The design and simulation of hybrid fuzzy-
PID controller. Information Technology and Applications (IFITA), 2010
International Forum on, vol. 3, IEEE, pp. 95–98, 2010.
17. Rey, J.P. and Dalvi, M.M., A numerical-based fuzzy-PID controller applied
to a DC drive. Industrial Electronics, Symposium Proceedings, ISIE ‘, IEEE
International Symposium on, IEEE, pp. 429–434, 2009.
18. Surya Kalawati, M. and Mitra, R., Comparative analysis of robustness of
optimally offline tuned PID controller and fuzzy supervised PID controller.
Engineering and Computational Sciences (RAECS), 2014 Recent Advances in,
pp. 1–6, 2014.
19. Salim, J.O., Fuzzy based pid controller for speed control of dc motor using
labview. WSEAS Transactions on Systems and Control, 10, 154–159, 2015.
20. Kumar, S. and Kumar, V., Speed control DC motor using fractional fuzzy
based PID controller. 2012 IEEE International Conference on Engineering and
Technology (ICETECH), Coimbatore, IEEE, pp. 1–6, 2012.
21. Mehra, V. and Kala, V., Design and simulation of FOPID composite param-
eters controller with MATLAB. Computer Design and Applications (ICCDA),
2010 International Conference on, vol. 4, IEEE, pp. V4–308-V4-311, 2010.
22. Ahmadi, A. and Kumari, S.U., Optimal fuzzy-PID controller. Power Signals
Control and Computations (EPSCICON), International Conference on, IEEE,
pp. 1–6, 2012.
23. Vikhe, P.S. and Dhanasekara, R., Improvement of speed control performance
in BLDC motor using PID controller. International Conference on Advanced
Communication Control and Computing Technologies (ICACCCT), IEEE, pp.
380–384, 2018.
24. Haung, Y. and Chen, Design fuzzy-PID controllers with robust testing.
Machine Learning and Cybernetics, 2007 International Conference on, vol. 2,
pp. 956–961, 2007.
25. Duman, S., Reddy, B.A., Charan, K.R., Design of fuzzy PI+D and fuzzy PID
controllers using GSA algorithm. 2009 Second International Conference on
Emerging Trends in Engineering & Technology, IEEE, pp. 957–961, 2009.
26. Elsrogy, W.M., Fkirin, M.A., Hassan, M.A.M., Speed control of DC motor
using PID controller based on artificial intelligence techniques. 2013
International Conference on Control, Decision and Information Technologies
(CODIT), IEEE, pp. 196–201, 2013.
27. Rahmani, R., Mahmodian, M.S., Mekhilef, S., Shojaei, A.A., Fuzzy logic con-
troller optimized by particle swarm optimization for DC motor speed con-
trol. 2012 IEEE Student Conference on Research and Development (SCORD),
Pulau Pinang, IEEE, pp. 109–113, 2012.
28. Salim, J.O. and Naveen, Speed control of DC motor using fuzzy logic based
on LabView, International Journal of Scientific and Research Publications
(IJSRP), 3, 6, 1–5, June 2013.
29. Omar, A. and Khedr, M.S., Particle swarm optimization of fuzzy supervisory
controller for nonlinear position control system. Computer Engineering &
Systems (ICCES), 2013 8th International Conference on, IEEE, pp. 138–145,
2013.
30. Payakkawan, P., Klomkarn, K., Sooraksa, P., Dual-line PID controller based
on PSO for speed control of DC motors. 9th International Symposium on
Communications and Information Technology, IEEE, pp. 134–139, 2009.
31. Somwanshi, D., Bundele, M., Kumar, G., Parashar, G., Comparison of fuzzy-
PID and PID controller for speed control of DC motor using LabVIEW. Proc.
Comput. Sci., 152, 252–260, 2019.
14
Application of Intuitionistic Fuzzy
Network Using Efficient Domination
A. Meenakshi1, J. Senbagamalar1* and A. Kannan2
1
Department of Mathematics, Vel Tech Rangarajan Dr. Sagunthala R&D Institute
of Science and Technology, Chennai, Tamil Nadu, India
2
Department of Mathematics, Vel Tech Multi Tech Dr. Rangarajan Dr. Sakunthala
Engineering College, Chennai, Tamil Nadu, India
Abstract
The most secured computational technique is to find the secret information
using encryption and decryption defined in this paper. Mathematical modeling
of Iintuitionistic network is defined and constructed to elude the burgeoning
intruder. The studies of efficient domination of intuitionistic graph is initiated and
this domination parameter plays a nuance technique to decrypt the framed net-
work. The algorithm is framed to encrypt and decrypt the given secret number.
Keywords: Efficient domination, intuitionistic network, single valued,

mathematical modeling
14.1 Introduction
A network is a group of peoples (a set of nodes) interact with each oth-
ers, sharing their knowledge and information (link is a relation which
represents sharing the information) so as to develop their professional
skills or social contacts. It imparts the good relationship among to develop
their business skills and it plays a vital role when starting the new busi-
ness. Labeling is a tool of numbering the persons (nodes) and the relations
(links) between any two members. It helps to identify the set of significant
213
persons who plays a vital role in the given network. Developed a com-
binatorial technique involve with domination, intuitionistic graphs with
encryption and decryption concepts.
Although the policy is very illuminating and informative, we feel that
this traditional approach needs to be supplemented with basic mathemat-
ical and a framework that displays cryptology as a fully-fledged part of
computational technique. On the other hand, something that has been
done and accomplished in recent decades, although this study is in the
subject of cryptology, we occasionally find stenographic purposes in addi-
tion to cryptographic ones, as can be seen in the traditional parts. Each in
this research, we refer to a secret in as much as a concealment system in
which the existence of the message is concealed from the enemy and more
commonly, in the sense that the message is concealed by cipher, but its
existence is known, not hidden from the enemy or attacker.
Domination plays a vital role in decision making, monitoring, mini-
mize the cost of network etc.., Let ‘O’ denote the set of monitoring mem-
bers (a set of nodes) of the given network. Every member in the given
network except the monitoring members should be the neighbor of at least
one monitoring member of the given network. The minimum number of
monitoring members of this network is domination number of the given
network. Every member in the given network except the monitoring mem-
bers should be a neighbor of exactly one monitoring member of the given
network. The minimum number of monitoring members of this network
is efficient domination number of the given network.
L.A. Zadeh proposed a mathematical framework to characterize the
phenomenon of uncertainty in real-world situations in 1965 [18]. The idea
of fuzzy graphs and various fuzzy analogues of graph theory ideas with
connectedness were first presented by Rosenfeld [14]. Ore [13] and Berge
started researching graphs’ domination sets. Paired domination studies
begun by Teresa et al. [17]. Efficient domination was initiated by Biggs [2],
V.R. Kulli [5], begun the study of split domination of graph and also he [6]
wrote the theory of domination in graphs. The independent domination
number was first used in graphs by Cockayne [3]. Equitable domination
was introduced by Swaminathan and Dharmalingam [16]. Paired equitable
domination was introduced and studied by A. Meenakshi [7] and it contin-
ued by in inflated graph and its complement of a graph [8, 9]. Intuitionistic
fuzzy relations and Intuitionistic fuzzy graphs (IFGS) were developed by
K.T. Atanassov [1]. IFG was defined by M.G. Karunambigai et al. [4] which
is a special case of IFGS defined by A.Shannon and Atanassov of [15]. The
terms “order,” “degree,” and “size” of IFG were defined by A. Nagoor Gani
Intuitionistic Fuzzy Network Using Efficient Domination 215
and Shajitha Begum [12]. Split domination in Intuitionistic fuzzy graph

was introduced by A. Nagoor Gani and S. Anu priya [11]. Split domination
in neutrosophic graph was studied by Mullai et al. [10].
14.2 Efficient Domination in Intuitionistic Fuzzy

Graph (IFG)
Definition 14.2.1 An IFG is of the form HNG =(Vs, Es) where
(i) s = {o1,o2,…,on} such that TVs : Vs [0,1] ; and

V
FVs : Vs [0,1] denote the degree of truth member-
ship value, degree of indeterminacy membership value,
and degree of falsity membership value respectively and
0 TVs (v s ) FVs (v s ) 2 for every vs ∈ V.
(ii) E V V where TEs : V V [0,1] ; FEs : V V [0,1]
are defined by TEs {(ai , a j )} min{TVs (ai ),TVs (a j )} ;
FEs {(ai , a j )} max{FVs (ai ), FVs (a j )} denote the degree
of truth membership value and degree of falsity mem-
bership value of the edge (ai,aj) ∈ Es respectively where
0 TEs {(ai , a j )} FEs {(ai , a j )} 2 (ai , a j ) Es .
Definition 14.2.2 A subset T of V1 is said to be dominating set of a single

valued IFG if for every vertex in V1-T is dominated by at least one vertex
of V1. The dominating set T is said to be minimal if no proper subset of T
is a dominating set.
Definition 14.2.3 An arc (u-v) is said to be strong arc if its degree of edge
membership value is is equal to strength of connectedness between u and v.
Definition 14.2.4 Let e = (a, b) be an edge of a IFG. We say that a domi-

nates b if there exists a strong arc between them.
Definition 14.2.5 Intuitionistic Fuzzy Network (IFN) is defined as a group

of same category peoples (a set of nodes) they interact with each other
and work together (link is a relation which represents sharing work or
sharing information) such that every node (person) has true degree mem-
bership value (T), indeterminacy degree membership value (I) and falsity
degree membership (F). The relation (information, knowledge sharing,
etc.) between any two persons is represented by link. The link also has true
degree membership value (T) and falsity degree membership (F).
d1(0.2, 0.4)
a1(0.25, 0.15)
d2(0.4, 0.62)
a2(0.2, 0.5) a(0.4, 0.15) c(0.14, 0.2)

d(0.25, 0.35)
b(0.24, 0.25) d3(0.45, 0.2)
a3(0.4, 0.25)
d4(0.5, 0.4)
Figure 14.1 Efficient domination of IFG.
Definition 14.2.6 A IFG is said to be strong if it satisfies the following

TEs {(ai , a j )}; min{TVs (ai ),TVs (a j )}; FEs {(ai , a j )} max{FVs (ai ), FVs (a j )} (ai , a j ) Es
ai ), FVs (a j )} (ai , a j ) Es and ai & a j Vs.
Definition 14.2.7 A dominating set T of Vs is said to be efficient dominat-

ing set of a IFG if |T ∩ N[v]| = 1, for every vertex v in Vs -T, where N[v]
represents the closed neighborhood of v. The dominating set T is said to be
minimal efficient dominating if no proper subset of T is a efficient domi-
nating. The following IFG is strong.
Table 14.1 Membership values of vertex degree and edge degree.

Vertex degree membership values Edge degree membership values
a(0.4, 0.15) ab(0.24,0.25)
b(0.24, 0.25) bc(0.14,0.25)
c(0.14,0.2) cd(0.14,0.35)
d(0.25,0.35) aa1(0.25,0.15)
a1(0.25,0.15) aa2 (0.2,0.5)
a2(0.2,0.5) cc3 (0.4,0.25)
a3(0.4,0.25) dd1 (0.2,0.4)
d1 (0.2,0.4) dd2 (0.25,0.62)
d2 (0.4,0.62) dd3 (0.25,0.35)
d3 (0.45,0.2) dd4 (0.25,0.4)
d4 (0.5,0.4) ----
The vertex and edge degree membership values of the IFG in Figure 14.1
is given in Table 14.1. Every edge in the IFG in Figure 14.1 is strong, the
only efficient dominating set is T = {a,d} since every vertex in V-T is dom-
inated by exactly one vertex and this dominating set is unique.
Encryption and Decryption is the technique used to identify or break
the secret key or secret information present in the network. In this paper
we present the new nuance combinatorial technique of encryption and
decryption of single valued IFG using efficient domination.
14.3 Main Frame Work

This main frame work consists of this paper is
Construction of IFN from sub IFN

Secret key
Encryption Algorithm
Decryption Algorithm
14.3.1 Construction of IFN from Sub IFN

The secret number (numerical value) to be encrypted is non zero inte-
ger. Select the suitable numerical value (NV ≠ 0) (as we have to split this
under modulo r, r ≠ 0). Now NV is sub divided in to ‘r’ values say NV1,
NV2,…, NVr such that NV1 ≡ R1(mod r) (where R1 = 0), NV2 ≡ R2(mod r)
(where R2 = 1), NV3 ≡ R3(mod r) (where R3 = 2),…, NVr ≡ Rr(mod r) (where
Rr = r-1). Since we have ‘r’ subdivision values, have to frame ‘r’ sub network
and planned to assign ‘r’ efficient domination nodes in the constructed
network. Let the efficient dominating nodes (EDN) be o1, o2, o3,…,or . These
nodes are the center of the sub networks say SN1, SN2, SN3 ,…,SNr respec-
tively. Let the neighbors of o1, o 2, o 3,…, o r be o11 , o12 ,..., o1l1; o21 , o22 ,..., o2l2 ;
o31 , o32 ,..., o3l3 ,…., or1 , or 2 ,..., orlr respectively.
First SVN sub network is SN1 whose center is o1 and its neighbors are
o11 , o12 ,..., o1l1. o11 , o12 ,..., o1l1 First subdivision value NV1 ≡ R1(mod r). Set
NV1
V1 and D1 = Dv1/V1 (where Dv1 is the numerical value 1 followed
r
by the number of 0’s digits of integral part of V1) partitioned into sum of
l1 values say d11 , d12 ,...d1l1 respectively and assign these values are minimum
value of either o1 or o11 , o12 ,..., o1l1 , degree of truth membership value. IFN
sub network as shown in Figure 14.2.
Second SVN sub network is SN2 whose center is o2 and its neighbors are
NV2
o21 , o22 ,..., 02l2 . First subdivision value NV2 ≡ R2(mod r). Set V2 and
r
D2 = Dv2/V2 (where D is the numerical value 1 followed by the number of 0’s
digits of integral part of V2) partitioned into sum of l2 values say d21 , d22 ,..., d2l2
and assign these values are minimum value of either o2 or o21 , o22 ,..., o2l2 degree
of truth membership value. Repeat the process till to frame the sub network
SNr. IFN sub networks 2 and r are shown in Figures 14.3 and 14.4.
By the definition of SVNN, the degree of membership values of the
edges o1o11 , o1o12 ,..., o1o1l1 are (min{o1(t1),o11(t11)}, max{o1(f1),o11(f11)})
(min{o1(t1),o12(t12)}, max{o1(f1),o12(f12)}),…,( min{o1(t1 ), o1l1 (t1l1 )} , max{o1{ f1}, o1l1 { f1l1 }}
max{o1{ f1}, o1l1 { f1l1 }} ) respectively.
By the definition of IFN, the degree of membership values of the
edges o2o21 , o2o22 ,..., o2o2l2 are (min{o2(t2),o21(t21)}, max{o2(f2),o21(f21)})
(min{o2(t2),o22(t22)}, max{o2(f2),o22(f22)}),…,( min{o2 (t 2 ), o2l1 (t 2l2 )} , max{o2 ( f 2 ), o2l1 ( f 2l2
max{o2 ( f 2 ), o2l1 ( f 2l2 }}) respectively. Repeat the process till to frame the sub net-
work SNr and by the definition of IFN, the rest of the edge’s degree mem-
bership values will be defined.
o11(t11, f11)
o1l1(t1l1, f1l1)
o12(t12, f12)
o1(t1, f1)
o13(t13, f13)
o14(t14, f14)
o15(t15, f15)
Figure 14.2 IFN subnetwork-1.
o21(t21, f21)
o2l2(t2l2, f2l2)
o22(t22, f22)
o2(t2, f2)
o23(t23, f23)
o24(t24, f24)
o25(t25, f25)
Figure 14.3 IFN subnetwork-2.

or1(tr1, fr1)
orlr(trlr , frlr )
or2(tr2, fr2)
or3(tr3, fr3)
or(tr, fr)
or4(tr4, fr4)
or5(tr5, fr5)
Figure 14.4 IFN rth subnetwork.
By the definition of IFN, the degree of membership values of the

edges or or1 , or or 2 ,..., or orlr are (min{or(tr),or1(tr1)}, max{or(fr),or1(fr1)})
(min{or(tr),or2(tr2)}, max{or(fr),or2(fr2)}),…,(min{or (t r ), orlr (t rlr )} , max{or ( f r ), orlr ( f rlr
max{or ( f r ), orlr ( f rlr )} ) respectively.
14.4 Secret Key

The key is to break the encrypted SVNN is the efficient dominating set of
the SVN encrypted network. Once we find the efficient dominating set of
this IFN, we can decrypt it.
14.4.1 Encryption Algorithm

Input: NV ≥ r, r ≠ 0 is the secret number
Output: Encrypted IFN Network
begin
Step 1: Sub divide the secret number NV into “r’ values NV1 , NV2,…, NVr
such that NV1 R1(mod r ) (where R1 = 0), NV2 ≡ R2(mod r) (where R2 = 1),
NV3 ≡ R3(mod r) ( where R3 = 2),…, NVr ≡ Rr(mod r) (where Rr = r-1)
Step 2: Frame ‘r’ sub network and planned to assign ‘r’ efficient domina-
tion nodes in the constructed network. The efficient dominating nodes
(EDN) be o1, o2, o3 ,…,or . These nodes are the centers of the sub networks
say SN1, SN2, SN3 ,…, SNr respectively. The neighbors of o1, o2, o3,…, or are
o11 , o12 ,..., o1l1 ; o21 , o22 ,..., o2l2 ; o31 , o32 ,..., o3l3 ,…., or1 , or 2 ,..., orlr respectively.
(conveniently choose the number of neighboring vertices l1, l2, …, lr of o1,
o2,o3,…, or respectively where l1, l2, …, lr ≥ 1)
Step 3: The number of nodes present in the IFN network is r + l1 + l2 + … + lr.
Define Min E (Minimum no. of edges present in the constructed network is
1st sub network
o1l1(t1l1, f1l1) o11(t11, f11)

o12(t12, f12)
o1(t1, f1)
o13(t13, f13)
o14(t14, f14)
o15(t15, f15) 2nd sub network
o21(t21, f21)
o2l2(t2l2, f2l2)
rth sub network o22(t22, f22)
orlr(trlr , frlr )
or1(tr1, fr1)
o2(t2, f2)
or2(tr2, fr2) o23(t23, f23)
or(tr, fr)
or3(tr3, fr3) o24(t24, f24)
o25(t25, f25)
or4(tr4, fr4)
or5(tr5, fr5)
3rd sub network
(r-1) sub network
4th sub network
i-th sub network

1<i<lr
Figure 14.5 Encrypted IFN network with minimum edges.

1st sub network
o1l1(t1l1, f1l1)
o11(t11, f11)
o12(t12, f12)
o1(t1, f1)
o13(t13, f13)
o14(t14, f14) 2nd sub network

o15(t15, f15)
o21(t21, f21)
o2l2(t2l2, f2l2)
o22(t22, f22)
rth sub network

o2(t2, f2)
orlr(trlr , frlr )
or1(tr1, fr1) o23(t23, f23)
or2(tr2, fr2)
or(tr, fr) o24(t24, f24)

or3(tr3, fr3)
o25(t25, f25)
or4(tr4, fr4)
or5(tr5, fr5)
(r-1) sub network
4th subnetwork
i-th sub network

1<i<lr
Figure 14.6 Encrypted IFN with moderate edges.

1st sub network

o1l1(t1l1, f1l1)
o11(t11, f11)
o12(t12, f12)
o1(t1, f1)
o13(t13, f13)
o15(t15, f15)
o14(t14, f14)
2nd sub network

o21(t21, f21)
o2l2(t2l2, f2l2)
rth sub network o22(t22, f22)
orlr(trlr , frlr )
or1(tr1, fr1)
o23(t23, f23)
or2(tr2, fr2)
or(tr, fr)
or3(tr3, fr3) o24(t24, f24)
o25(t25, f25)
or4(tr4, fr4)
or5(tr5, fr5)
3rd sub network
(r-1) sub network
4th sub network
i-th sub network

1<i<lr
Figure 14.7 Encrypted IFN with more than moderate edges.
denoted by Min E ) = o1o1 j1 , o2o2 j2 ,..., or orjr / 1 j1 l1 ,1 j2 l2 ,...,1 jr lr {o1 j1 o2 j2 o2

...,1 j lr {o1 j1 o2 j2 , o2 j2 o3 j3 ,..., o(r 1) jr 1 orjr } for only one j1, j2, …, jr where 1 ≤ j1 ≤ l1,
1 ≤ j2 ≤ l2, …, 1 ≤ jr ≤ lr.
Hence the minimum number of edges present in the network is (r − 1) +
l1 + l2 + … + lr , which is shown is Figure 14.5. IFN with moderate edges and
more than moderate edges are shown in Figures 14.6 and 14.7.
Step 4: Define (Maximum no. of edges present in the constructed network

is denoted by Max E) Max E = {(a,b); 1 ≤ a,b ≤ l1 + l2 + … + lr + r; a ≠ b}
o1ok1 where 2 k1 r , o2ok2 where 3 k2 r , o3ok3 where 4 k3 r ,..., or 1or ,
o1o2 j2 , o1o3 j3 ...., o1orjr ; o2o3 j 3 , o2o4 j 4 ,...o2orjr ; o3o4 j 4 , o3o5 j 5 ,...o3orjr ;...; or 1orjr ,
where 1 j2 l2 , 1 j3 l3 ...., 1 jr lr .
NV1 NV2 NV3 NVr

Step 5: V1 ,V2 ,V3 ,Vr
r r r r
D1 = Dv1/V1 (where Dv1 -the numerical value 1 followed by the number of
0’s digits of integral part of V1)
D2 = Dv2/V2 (where Dv2 -the numerical value 1 followed by the number
of 0’s digits of integral part V2) ,…, Dr = Dvr/Vr
(where Dvr -the numerical value 1 followed by the number of 0’s digits
of integral part of Vr )
Step 6: Split D1 d11 d12 ... d1l1 ; D2 d21 d22 ... d2l2 ;….; Dr dr1 dr 2 ... drlr
Dr dr1 dr 2 ... drlr ;
Assign min{o1(t1),o11(t11)} = d11, min{o1(t1),o12(t12)} = d12,…,
min{ o1(t1 ), o1l1 (t1l1 )} d1l1 in the first sub network.
min{ o2 (t1 ), o2l2 (t 2l2 )} d2l2 in the second sub network, continuing the pro-
cess till to assign
min{or(t1),or1(tr1)} = dr1, min{or(t1),or2(tr2)} = d22,…., min{ or (t1 ), orlr (t rlr )} drlr
n{ or (t1 ), orlr (t rlr )} drlr , in the r th sub network.
Step 7: Rest of the edge’s membership values will be followed by the defi-
nition of IFN
end
14.4.2 Decryption Algorithm

Input: Encrypted IFN
Output: NV, the secret number
Begin
Step 1: Find the Efficient dominating members of SVNN o1, o2, o3,…, or
such that N[o1] ∩ N[o2] ∩ N[or] = ϕ where N[oi] represent the neighbours
of the vertex oi
l1
Step 2: V1 Dv1 d1 j1
j1 1
l2 lr
V2 Dv 2 d1 j2 ,…, Vr Dvr d1 jr
j2 1 jr 1
r
Step 3: NV r Vi
i 1
End
14.5 Illustration
The secret number is NV = 10810
14.5.1 Construction of IFN from Sub IFN

The suitable numerical value NV. we have to split this NV under modulo
r=5. Now NV is sub divided in to ‘5’ values say NV1, NV2, NV3 , NV4, NV5
such that NV1 = 3000 ≡ 0(mod r), NV2 = 3001 ≡ 1(mod r), NV3 = 1502 ≡
2(mod r), NV4 = 1503 ≡ 3(mod r) NV5 = 1804 ≡ 4(mod r). Since we have
‘5’ subdivision values, have to frame 5 sub network and planned to assign
‘5’ efficient domination nodes in the constructed network. Let the efficient
dominating nodes (EDN) be o1, o2, o3, o4, o5 . These nodes are the center of
the sub networks say SN1, SN2, SN3, SN4, SN5, respectively. Let the neighbors
of o1, o2, o3, o4, o5 be o11, o12, o13, o14, o15, o16; o21, o22, o23, o24; o31, o32, o33, o34;
o41, o42, o43, o44; o51, o52, o53, o54, o55, o56, respectively (selecting the neighbor-
ing nodes depends on our choice). First sub network is SN1 whose center
is o1 and its neighbors are o11, o12, o13, o14, o15, o16. First subdivision value
NV1 3000
NV1 = 3000 ≡ 0(mod r). Set V1 600 and D1 = Dv1/V1 =
r 5
1000/600 = 0.600 (where Dv1 is the numerical value 1 followed by the num-
ber of 0’s digits of integral part of V1) is partitioned into sum of 6 values say
d11 , d12 ,...d16 and assign these values are minimum value of either o1 or o11,
o12, o13, o14, o15, o16 truth degree membership value. The edge membership
values of the constructed IFN are given in Table 14.2. IFN sub network 1 is
Second sub network is SN2 whose center is O2 and its neighbors are
o21 , o22 , o23 , o24 , o25 .o26 . Second subdivision value NV2 = 3001 ≡ 1(mod r).
NV2 3001
Set V2 600.2 and D2 = Dv2/V2 = 1000/600.2 = 0.6002
r 5
(where Dv2 is the numerical value 1 followed by the number of 0’s of inte-
gral part of V2 ) is partitioned into sum of 6 values say d21, d22, … d26 and
Table 14.2 Degree membership values of the edges.

Degree of
Edges membership values Edges Degree of membership values
o1o11 (0.1,0.33) o33o34 (0.05,0.3)
o1o12 (0.1,0.35) o34o35 (0.05,0.25)
o1o13 (0.1,0.32) o4o41 (0.1,0.4)
o1o14 (0.1,0.3) o4o42 (0.05,0.4)
o1o15 (0.1,0.33) o4o43 (0.05,0.4)
o1o16 (0.1,0.3) o4o44 (0.05,0.4)
o11o21 (0.1,0.36) o4o45 (0.0506,0.4)
o12o26 (0.1,0.35) o41o45 (0.0506,0.36)
o13o26 (0.1,0.32) o41o42 (0.05,0.36)
o11o16 (0.1,0.33) o43o44 (0.05,0.3)
o11o12 (0.1,0.36) o41o53 (0.05,0.36)
o14o15 (0.1,0.25) o45o54 (0.05,0.2)
o15o16 (0.1,0.33) o41o54 (0.05,0.2)
o2o21 (0.1,0.36) o5o51 (0.05,0.4)
o2o22 (0.1,0.3) o5o52 (0.05,0.36)
o2o23 (0.1,0.3) o5o53 (0.05,0.4)
o2o24 (0.1,0.3) o5o54 (0.05,0.4)
o2o25 (0.1,0.3) o5o55 (0.05,0.4)
o2o26 (0.1002,0.3) o5o56 (0.0608,0.4)
o21o26 (0.1,0.36) o14o52 (0.05,0.25)
o25o26 (0.1,0.2) o14o51 (0.05,0.36)
o23o24 (0.1,0.3) o15o56 (0.0608,0.33)
o22o23 (0.1,0.35) o51o56 (0.05,0.36)
o3o31 (0.05,0.36) o51o52 (0.05,0.361)
(Continued)
Table 14.2 Degree membership values of the edges. (Continued)

Degree of
Edges membership values Edges Degree of membership values
o3o32 (0.05,0.4) o53o54 (0.05,0.35)
o3o33 (0.05,0.4) o54o55 (0.05,0.25)
o3o34 (0.05,0.4) o24o53 (0.05,0.35)
o3o35 (0.0504,0.4) o34o53 (0.05,0.35)
o31o35 (0.0504,0.36) o45o42 (0.05,0.25)
o32o33 (0.05,0.35)
o11(0.1, 0.33)
o16(0.1, 0.25)
o1(0.2, 0.3) o12(0.1, 0.35)

o15(0.1, 0.33)
o14(0.1, 0.25) o13(0.1, 0.32)
Figure 14.8 Illustration IFN subnetwork-1.
assign these values are minimum value of either o2 or o21, o22, …, o26 truth
degree membership value. IFN sub network 2 is shown in Figure 14.9.
By the definition of IFN, the degree of membership values of the edges
o2o21 , o2o22 ,..., o2o2l2 are (min{o2(t2),o21(t21)}, max{o2(f2),o21(f21)}) (min
{o2(t2),o22(t22)}, max{o2(f2),o22(f22)}),…,(min{o2(t2),o26(t26)}, max{o2(f2),
o26(f26)}) respectively.
Third subnetwork is SN3 whose center is O3 and its neighbors are o31, o32,
o33, o34, o35. Third subdivision value NV3 = 1502 ≡ 2(mod r). Set D3 = Dv3/V3 =
1000/300.4 = 0.3004 (where Dv3 is the numerical value 1 followed by the num-
ber of 0’s of integral part of V3 ) is partitioned into sum of 5 values say d31, d32,
…, d35 and assign these values are minimum value of either o3 or o31, o32, …, o35
true degree membership value. IFN sub network 3 is shown in Figure 14.10.
Fourth subnetwork is SN4 whose center is o4 and its neighbors are o41, o42,
o43, o44, o45. Fourth subdivision value NV4 = 1503 ≡ 3(mod r). Set D4 = Dv4/
o26(0.12, 0.15) o21(0.1, 0.36)
o2(0.2, 0.3) o22(0.1, 0.35)
o25(0.1, 0.26)
o24(0.1, 0.25) o23(0.1, 0.3)
o35(0.0504, 0.25) o31(0.1, 0.36)
o3(0.05, 0.4) o32(0.05, 0.35)
o34(0.05, 0.25)
o33(0.05, 0.3)
V4 = 1000/300.6 = 0.3006 (where Dv4 is the numerical value 1 followed by

the number of 0’s of integral part of V4 ) is partitioned into sum of 5 values
say d41, d42, …, d45 and assign these values are minimum value of either o4
or o41, o42, o43, o44, o45 truth degree membership value. IFN sub network 4 is
o45(0.0506, 0.2) o41(0.1, 0.36)
o4(0.06, 0.4) o42(0.05, 0.15)
o44(0.05, 0.25)
o43(0.05, 0.3)

o56(0.0608, 0.22) o51(0.05, 0.36)
o55(0.05, 0.25)
o52(0.1, 0.4) o52(0.05, 0.15)
o54(0.05, 0.2) o53(0.05, 0.35)
Fifth subnetwork is SN5 whose center is o5 and its neighbors are o51, o52,
o53, o54, o55. Fourth subdivision value NV5 = 1804 ≡ 4(mod r). Set D5 = Dv5/
V5 = 1000/360.8 = 0.3608 (where Dv5 is the numerical value 1 followed by
the number of 0’s of integral part of V5 ) is partitioned into sum of 6 values
say d41, d42, …, d45 and assign these values are minimum value of either o5
or o51, o52, o53, o54, o55, truth degree membership value. IFN sub network 5
is shown in Figure 14.12.
14.5.2 Secret Key

The key is to break the encrypted IFN is the efficient dominating members
of the IFN encrypted network which is shown in Figure 14.13. Once we
find the efficient dominating set of this network, we can decrypt it. The
efficient dominating members of this IFN is o1, o2, o3, o4, and o5.
14.5.3 Encryption Algorithm

Input: NV = 10810 is the secret number
Output: Encrypted IFN Network
begin
Step 1: Sub divide the secret number NV into ‘5’ values NV1, NV2, NV3,
NV4, NV5 such that NV1 = 3000 ≡ 0(mod r), NV2 = 3001 ≡ 1(mod r), NV3 =
1502 ≡ 2(mod r), NV4 = 1503 ≡ 3(mod r) NV5 = 1804 ≡ 4(mod r).
Step 2: Frame ‘5’ sub network and planned to assign ‘5’ efficient domi-
nation nodes in the constructed network. The efficient dominating nodes
(EDN) be o1, o2, o3, o4, and o5. These nodes are the centers of the sub net-
works say SN1, SN2, SN3, SN4, SN5 respectively. Let the neighbors of o1, o2,
o3, o4, and o5 be o11, o12, o13, o14, o15, o16; o21, o22, o23, o24, o25, o26; o31, o32, o33, o34;
o41, o42, o43, o44; o51, o52, o53, o54, o55, o56 respectively.
o11(0.1,0.33)
o16(0.1,0.25)
o1(0.2,0.3)
o15(0.1,0.33)
o12(0.1,0.35)
o14(0.1,0.25) o13(0.1,0.32)
o26(0.1002,0.15) o21(0.1,0.36)
o56(0.0608,0.22) o51(0.05,0.36)
o22(0.1,0.35)
o2(0.2,0.3)
o55(0.05,0.25) o5(0.1,0.4) o52(0.05,0.15) o25(0.1,0.26)
o24(0.1,0.25) o23(0.1,0.3)
o54(0.05,0.2) o53(0.05,0.35)
o35(0.504,0.25) o31(0.1,0.36)
o41(0.1,0.36)
o45(0.0506,0.2)
o3(0.1,0.4)
o4(0.2,0.4) o42(0.05,0.15)
o32(0.05,0.35)
o43(0.05,0.3) o34(0.05,0.25)
o44(0.05,0.25) o33(0.05,00.3)
Figure 14.13 Illustration encrypted IFN with secret number-10810.
Step 3: The number of nodes present in the SVN network is 33.

Minimum no. of edges present in the constructed network is denoted by
Min E
o1o1 j1 , o2o2 j2 ,..., o5o5 j 5 / 1 j1 6,1 j2 6,1 j3 5,1 j4 5,1 j5 6 {o1 j1 o2 j2 ,
,1 j 6 {o1 j1 o2 j2 , o2 j2 o3 j3 , o4 j4 o5 j5 } for only one j1, j2, …, j5 where 1 ≤ j1 ≤ 6, 1 ≤ j2 ≤
6, 1 ≤ j3 ≤ 5, 1 ≤ j4 ≤ 5, 1 ≤ j5 ≤ 6
Hence the minimum number of edges present in the network is 32.
Step 4: Maximum no. of edges present in the constructed network is
denoted by Max E
(a, b);1 a, b 33; a b
o1ok1 where 2 k1 6, o2ok2 where 3 k2 6 , o3ok3 where 4 k3 5,.o4o5 ,
o1o2 j2 , o1o3 j3 ...., o1o5 j5 ; o2o3 j 3 , o2o4 j 4 ,...o2o5rj 5 ; o3o4 j 4 , o3Oo5 j 5 ; o4o5 j 5 ,
where 1 j2 6, 1 j3 5,1 j4 5 ,1 j5 6.
NV1 3000 3001 1502 1503 1804

Step 5: V1 ,V2 ,V3 ,V4 ,V5
r 5 5 5 5 5
D1 = Dv1/V1 = 1000/600 = 0.600; D1 = Dv1/V1 (where Dv1 -the numerical value
1 followed by the number of 0’s digits of integral part of V1 ). Remaining
values of D2, D3, D4, D5 are calculated as like D1.
D2 = 1000/600.2 = 0.6002; D3 = 1000/300.4 = 0.3004; D4 = 1000/300.6 =
0.3006;
D5 = 1000/360.8 = 0.3608
Step 6: Split D1 = d11 + d12 + d13 + d14 + d15 = 0.1 + 0.1 + 0.1 + 0.1 + 0.1 + 0.1;
D2 = d21 + d22 + d23 + d24 + d25 = 0.1 + 0.1 + 0.1 + 0.1 + 0.1 + 0.1002:
D3 = d31 + d33 + d33 + d34 + d35 = 0.05 + 0.05 + 0.05 + 0.1 + 0.0504
D4 = d41 + d44 + d44 + d44 + d45 = 0.05 + 0.05 + 0.05 + 0.1 + 0.0506
D5 = d51 + d55 + d55 + d55 + d55 + d56 = 0.05 + 0.05 + 0.05 + 0.05 + 0.05 +
0.0608
min{ o1(t1 ), o16 (t161 )} d16 in the first sub network.
min{o2(t1),o21(t26)} = d26 in the second sub network, continuing the process
till to assign
min{o5(t1),o51(t51)} = d51, min{o5(t1),o52(t52)} = d52,…., min{o5(t1),o56(t56)} =
d56, in the 5 th sub network.
Step 7: Rest of the edge’s membership values will be followed by the defi-
nition of IFN
end
14.5.4 Decryption Algorithm

Input: Encrypted IFN
Output: NV, the secret number
Begin
Step 1: Find the Efficient dominating members of SVNN o1, o2, o3, o4, o5
such that N[o1] ∩ N[o2] ∩ N[o3] ∩ N[o4] ∩ N[o5] = ϕ
6
Step 2: V1 Dv1 d1 j1 1000(0.1 0.1 0.1 0.1 0.1 0.1) 600
j1 1
6 5
V2 Dv 2 d1 j2 1000(0.6004) 600.4 , V3 Dv 3 d1 j2 1000(0.3004

j2 1 j2 1
5
d1 2 1000(0.3004) 300.4 ,
j2 1
5
V4 Dv 4 d1 j2 1000(0.3006) 300.6 ,
j2 1
6
V5 Dv 5 d1 j2 1000(0.3608) 360.8
j2 1
5
Step 3: NV r Vi 5(2162) 10810 .
i 1
14.6 Conclusion
The secret key is to find the efficient dominating set of IFN. Repetition of
assigning the dij values and maximum number of edges presented in the
constructed IFN then it becomes more complicated. Moreover, according
to the construction of IFN, the efficient dominating set of the constructed
network is unique and also it is strong.
References
1. Atanassov, K.T., Intutionistic fuzzy set theory and applications, Physica, New
York, 1999.
2. Biggs, N., Perfect codes in graphs. J. Comb. Theory Ser. B, 15, 289–296, 1973.
3. Cockayne, E.J. and Hedetniemi, S.T., Towards a theory of domination in
graph. Networks, 7, 247–261, 1977.
4. Karunambigai, M.G., Parvathi, R., Bhuvaneswari, R., Constant intuitionistic
fuzzy graphs. NIFS, 17, 37–47, 2011.
5. Kulli, V.R. and Janakiram, B., The split domination number of a graph. Graph
Theory Notes of New York, New York Academy of Sciences XXXXII, pp.
16–19, 1997.
6. Kulli, V.R., Theory of domination in graphs, Vishwa International Publications,
University of Michigan, Michigan, 2012.
7. Meenakshi, A. and Baskar Babujee, J., Paired equitable domination in graphs.
Int. J. Pure Appl. Math., 109, 7, 75–81, 2016.
8. Meenakshi, A., Paired equitable domination in inflated graph. Int. J. Innov.
Technol. Exploring Eng., 8, 7, 1117–1120, 2019.
9. Meenakshi, A., Equitable domination in inflated graphs and its comple-
ments. AIP Conf. Proc., 2277, 100006, 2020, 100006-1-8.
10. Meenakshi, A., Equitable domination in inflated graphs and its comple-
ments. AIP Conf. Proc., 2277, 100006, 2020, 100006-1-8.
11. Nagoor Gani, A. and Anupriya, S., Split domination in intuitionistic fuzzy
graph. Adv. Comput. Math. Appl., 2, 2, 2–8, 2012.
12. Nagoor Gani, A. and Shajitha Begum, S., Degree, order and size in intuition-
istic fuzzy graphs. Int. J. Algorithms Comput. Math., 3, 3, 11–16, 2010.
13. Ore, O., Theory of graphs. Americal Mathematical Society Colloquium
Publications, vol. 38, Providence, RI, 1962.
14. Rosenfeld, A., Fuzzy graphs, in: Fuzzy Sets and Their Applications, L.A.
Zadeh, K.S. Fu, M. Shimura (Eds.), Academic, New York, 1975.
15. Shannon, A. and Atanassov, K., On a generalization of intutionistic fuzzy
graphs. NIFS, 12, 1, 24–29, 2006.
16. Swaminathan, V. and Dharmalingam, K.M., Degree equitable domination on
graphs. Kragujev. J. Math., 35, 1, 191–197, 2011.
17. Haynes, T.W. and Slater, P.J., Paired domination in graphs. Networks, 32,
199–206, 1998.
18. Zadeh, L.A., Fuzzy sets. Inf. Control, 8, 338–353, 1965.
15
Analysis of Parameters Related to
Malaria with Comparative Study
on Fuzzy Cognitive Maps and
Neutrosophic Cognitive Maps
Regan Murugesan1*, Sathish Kumar Kumaravel1, Kala Raja Mohan1,
Narmada Devi Rathinam1 and Suresh Rasappan2
1
Department of Mathematics, Vel Tech Rangarajan Dr. Sagunthala R&D Institute
of Science and Technology, Avadi, Chennai, Tamil Nadu, India
2
Mathematics Section, College of Technology and Information Sciences,
University of Technology and Applied Sciences - Ibri, Ibri, Sultanate of Oman
Abstract
In this paper, a comparative study is made on different parameters which are
related to malaria disease, such as fever, headaches, nausea etc., by using FCM
and NCM methods and conclusion are derived to predict the spread of disease.
As a result of this study, the parameters involved in the spread of these diseases
are identified, and the prevention parameters are concluded. The major benefits
of this comparative study are to identify the parameters which are inconclusive,
applying the concept of fuzzy and neutrosophic cognitive maps.
Keywords: Fuzzy matrix, fuzzy cognitive map, neutrosophic cognitive map,
malaria
15.1 Introduction
Malaria is a disease caused by the transmission of parasites. This trans-
mission is done by the mosquitoes which results in illness of humans. The
infected mosquitoes transmit the parasite known as plasmodium. A high
233
share of the burden on malaria burden is found at the African region. In

the region, around 95 percentages of malaria infected cases was found out
of which 96 percentage of death was recorded. In addition, the heartbreak-
ing statistics is that about 80 percentage of the mortality was the children
aged below 5. With an aim to provide a rescue to the disease, an interna-
tional team of scientists came up with a protein which acts as a deactivating
agent in reducing the vito growth of plasmodium [1–4].
Anopheles mosquito is the female mosquito which is involved in the
transmission. The symptoms of malaria are chills, fever, nausea, headache,
abdominal pain, diarrhea, joint pain etc. The blood meal suck from an
infected person is transmitted by the mosquito to another uninfected per-
son. This ends up with an addition of malaria infected person [5–6]. About
90 percentage of the infection gets reduced by anti-malarial drugs. Still it is
necessary to take few preventive measures like mosquito nets, fully covered
clothes etc.
Decision making in the field of medical diagnosis involves a degree of
uncertainty and the need to consider the patient’s clinical parameters, the
context of illness, and the physician’s medical knowledge in order to deter-
mine and confirm the diagnosis. Fuzzy cognitive maps (FCM) are a cogni-
tive process that, like intuitive human reasoning, applies the key features of
fuzzy logic and neural processors to imprecise and uncertain descriptions
[7–9].
In this chapter, a comparative study is made on the different parameters
related to Malaria such as travelling area, Fever, chest pain etc., The param-
eters are analyzed by taking any one as ON state by the concept FCM.
Simultaneously those parameters are analyzed through the NCM as like
FCM by considering the same state as ON state. The aim of this chapter
is to analyze the disease malaria through comparison between FCM and
NCM among the parameters related to malaria.
The following is the outline of the contents of this chapter. In section
15.2, parameters respect to malaria is presented. In section 15.3 the Fuzzy
cognitive maps and its corresponding matrix form are discussed. Section
15.4 describes the Neutrosophic Fuzzy cognitive maps and its matrix form.
Finally in section 15.5, a comparison of results from the FCM and NCM is
discussed also in section 15.6 conclusion is furnished from the observation
of the analysis.
Analysis of Malaria with Study on FCM and NCM 235
15.2 Parameters of Malaria

In order to analyze the parameters of malaria, data were collected from
the medical experts. Based on their opinions the following factor was
identified.
• M1 - Travelling Area
• M2 - Lack of Insect repellent
• M3 - Fever
• M4 - Use mosquito net
• M5 - Living near to stagnant water site
• M6 - Shivering
• M7 - Less possibility of malaria
• M8 - sweating
• M9 - Possibility of malaria
• M10 - Prevention for malaria
• M11 - Chest pain
• M12 - Use fully covered clothes
15.3 Fuzzy Cognitive Map

The expert’s opinions were collected and based on their opinion; a graph is
formed by mapping between the parameters. Based on the fuzzy cognitive
map concept, weight age was assigned by the casual relation between the
nodes, i.e., edge weight was assigned as 1 (positive causality between the
nodes), if the relationship between the nodes had majority of respondents,
at the same time majority of respondents was uncertain then it is denoted
by 0. The corresponding fuzzy cognitive maps for the parameters are given
below in Figure 15.1.
15.3.1 Matrix Representation of FCM

The matrix representation of the fuzzy cognitive map is designed based
on the connectives between the nodes which are the possibility among the
parameters. The entries of the matrix are noted either as 0 or 1. The num-
ber 0 denoted for unconnected and 1 represents for connected between the
nodes. The adjacent matrix of the Figure 15.1 is given below.
M1
M3
M2
M4
M9
M10
M5
M6
M8
M12
M7 M11
Figure 15.1 Fuzzy cognitive maps on malaria.
W M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 M11 M12

M1 0 0 1 0 0 0 0 0 1 0 0 0
M2 1 0 1 0 0 0 0 0 1 0 0 0
M3 0 0 0 0 1 0 0 1 0 0 0 0
M4 0 0 0 0 0 0 1 0 0 1 0 0
M5 0 0 1 0 0 0 0 0 1 0 1 0
E M6 0 0 1 0 0 0 0 1 1 0 0 0
M7 0 0 0 1 0 0 0 0 0 1 0 1
M8 1 0 0 0 0 1 0 0 0 0 0 0
M9 1 1 0 0 1 0 0 0 0 0 0 0
M10 0 0 0 1 0 0 0 0 0 0 0 1
M11 0 0 0 0 0 0 0 1 1 0 0 0
M12 0 0 0 0 0 0 1 0 0 1 0 0
Case 1
Consider M2 in on state. Let us take,
T1 = [0 1 0 0 0 0 0 0 0 0 0 0]
The effect of T1 on E,
T1 E [1 0 1 0 0 0 0 0 1 0 0 0]
[1 1 1 0 0 0 0 0 1 0 0 0]
T2
T2 E [2 1 2 0 2 0 0 1 2 0 0 0]
[11 1 1 0 1 0 0 1 1 0 0 0]
T3
T3 E [3 1 3 0 2 1 0 1 3 0 1 0]
[1 1 1 0 1 1 0 1 1 0 1 0]
T4
T4 E [3 1 4 0 2 1 0 3 5 0 1 0]
[1 1 1 0 1 1 0 1 1 0 1 0]
T5
T4 T5
In this case, the parameter lack of repellent is considered as ON state.

While doing the iteration process, the result arises when we are violating
M2 on travel there is a chance of getting fever and shivering also the chance
for getting affected by malaria while living near to stagnant water sites and
sweating.
Case 2
The parameter M7, less possibility is considered as ON state. Let
T1 [0 0 0 0 0 0 1 0 0 0 0 0]
T1 E [0 0 0 1 0 0 0 0 0 1 0 1]
[0 0 0 1 0 0 1 0 0 1 0 1]
T2
T2 E [0 0 0 2 0 0 2 0 0 3 0 2]
[0 0 0 1 0 0 1 0 0 1 0 1]
T3
T2 T3
While taking M7 as ON state, the iteration process brings that when vio-
lating the preventive measures of disease can cause high risk of malaria
which implies that a person should use mosquito net and fully covered
clothes.
Case 3
Considering M12 is ON state. i.e., the parameter, using of fully covered
clothes like a major factor.
Let T1 [0 0 0 0 0 0 0 0 0 0 0 1]
T1 E [0 0 0 0 0 0 1 0 0 1 0 0]
[0 0 0 0 0 0 1 0 0 1 0 1]
T2
T2 E [0 0 0 2 0 0 1 0 0 2 0 2]
[0 0 0 1 0 0 1 0 0 1 0 1]
T3
T3 E [0 0 0 2 0 0 2 0 0 3 0 2]
[0 0 0 1 0 0 1 0 0 1 0 1]
T4
T3 T4
On the calculating process, this brings the result that when we use fully
covered clothes then the chances of getting malaria is less. Also, if the pre-
vention for malaria is followed then there will be a less possibility of getting
affected by malaria.
Case 4
By choosing M10 as ON state and rest of the parameters as OFF state, the
following iteration process is furnished.
Let T1 [0 0 0 0 0 0 0 0 0 1 0 0]
T1 E [0 0 0 1 0 0 0 0 0 0 0 1]
[0 0 0 1 0 0 0 0 0 1 0 1]
T2
T2 E [0 0 0 1 0 0 2 0 0 2 0 1]
[0 0 0 1 0 0 1 0 0 10 1]
T3
T3 E [0 0 0 2 0 0 2 0 0 3 0 2]
[0 0 0 1 0 0 1 0 0 1 0 1]
T4
T3 T4
On the process of simplication, it implies that if we use mosquito nets

and follow the prevention measures then the chances for getting malaria is
low. Also as like that chances of affecting with malaria are low while using
of fully covered clothes on day time.
Case 5
Consider M9 parameter as ON state. i.e., possibility of malaria.
Let T1 [0 0 0 0 0 0 0 0 1 0 0 0]
T1 E [1 1 0 0 1 0 0 0 0 0 0 1]
[1 1 0 0 1 0 0 0 1 0 0 0]
T2
T2 E [2 1 3 0 1 0 0 0 3 0 1 0]
[1 1 1 0 1 0 0 0 1 0 1 0]
T3
T3 E [2 1 3 0 2 0 0 2 4 0 1 0]
[1 1 1 0 1 0 0 1 1 0 1 0]
T4
T4 E [3 1 3 0 2 1 0 2 4 0 1 0]
[1 1 1 0 1 1 0 1 1 0 1 0]
T5
T4 E [3 1 4 0 2 1 0 3 5 0 1 0]
[1 1 1 0 1 1 0 1 1 0 1 0]
T6
T5 T6
In this case, the iteration process arises the result that except the param-
eters M4 [use mosquito net], M7 [less possibility of malaria], M10 [preven-
tion for malaria], and M12 [use fully covered clothes] all the others are
possibility of malaria.
15.4 Neutrosophic Cognitive Map

The experts’ opinions were collected and based on their opinion, a graph is
formed by mapping between the parameters. Based on the Neutrosophic
cognitive map concept, weight age was assigned by the casual relation
between the nodes, i.e., edge weight was assigned as 1 (positive causality
between the nodes), if the relationship between the nodes had majority
of respondents, at the same time the respondents which are uncertain or
indeterminate then it is denoted by I. The number zero is assigned, when
there is no relationship between the parameters based on expert’s opinion.
In the neutrosophic graph, for the indeterminate case the edges between
the nodes are drawn by the dotted lines. The corresponding neutrosophic
cognitive maps for the parameters are given below in Figure 15.2.
15.4.1 Matrix Representation of NCM

The matrix representation of the neutrosophic cognitive map is designed
based on the connectives between the nodes which are the possibility
among the parameters. In addition to the FCM concept, here one more case
is occur, when there is an inconclusive possibility of relationship between
the parameters that is denoted as indeterminate I case. The entries of the
matrix are noted as 0 or 1 or I. The number 0 denoted for unconnected, 1
represents for connected between the nodes and I noted for the indetermi-
nate case between the nodes which is connected by the dotted lines. The
adjacent matrix of Figure 15.2 is given below.
M1
M3 M2
M4
M9
M10
M5
M6
M8
M12
M7 M11
Figure 15.2 Neutrosophic cognitive maps on malaria.

W M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 M11 M12

M1 0 0 1 0 0 0 0 0 I 0 0 0
M2 I 0 1 0 0 0 0 0 I 0 0 0
M3 0 0 0 0 1 0 0 I 0 0 0 0
M4 0 0 0 0 0 0 1 0 0 1 0 0
M5 0 0 1 0 0 0 0 0 1 0 I 0
E M6 0 0 1 0 0 0 0 1 I 0 0 0
M7 0 0 0 1 0 0 0 0 0 1 0 I
M8 1 0 0 0 0 1 0 0 0 0 0 0
M9 I I 0 0 1 0 0 0 0 0 0 0
M10 0 0 0 1 0 0 1 0 0 0 0 1
M11 0 0 0 0 0 0 0 1 I 0 0 0
M12 0 0 0 0 0 0 0 0 0 1 0 0
Case 1
Let us take M2 in ON state side. i.e., lack of insect repellent.
Let us take
U1 [0 1 0 0 0 0 0 0 0 0 0 0]
U1 E [I 0 1 0 0 0 0 0 I 0 0 0]
[I 1 1 0 0 0 0 0 I 0 0 0]
U2
U2 E [I I 2 I I 1 0 1 I 0 0 I I 2 I 0 0 0]
[I 1 1 0 1 0 0 I I 0 0 0]
U3
U3 E [2 I I 2 I 2 2 I 0 1 I I 0 I I 2 I 1 0 I 0]
[I 1 1 0 1 I 0 I 1 0 I 0]
U4
U4 E [3I I 2 2 I 0 2 I 0 3I 1 I 3I 2 0 I 0]
[I 1 1 0 1 I 0 I 1 0 I 0]
U5
U4 U5
On the continuation of iteration process, it seems that except use of

mosquito net, less possibility of malaria and using fully covered clothes.
All the others are factors of M2. Here we see the indeterminate case.
Case 2
Consider M7 [less possibility of malaria] is ON state.
U1 [0 0 0 0 0 0 1 0 0 0 0 0]
U1 E [0 0 0 1 0 0 0 0 0 0 1 0 I]
[0 0 0 1 0 0 1 0 0 1 0 I]
U2
U2 E [0 0 0 2 0 0 2 0 0 2 I 0 1 I]
[0 0 0 1 0 0 1 0 0 1 0 1]
U3
U3 E [0 0 0 2 0 0 2 0 0 3 0 1 I]
[0 0 0 1 0 0 1 0 0 1 0 1]
U4
U3 U4
On the continuation of iteration process, when M7 is ON state, then

there is a less possibility of malaria and use fully covered clothes are also
under prevention for the disease malaria.
Case 3
Taking M12 as a positive state. i.e., use fully covered clothes as on state.
Then let
U1 [0 0 0 0 0 0 0 0 0 0 0 1]
U1 E [0 0 0 0 0 0 0 0 0 1 0 0]
[0 0 0 0 0 0 0 0 0 1 0 1]
U2
U2 E [0 0 0 1 0 0 1 0 0 1 0 1]
[0 0 0 1 0 0 1 0 0 1 0 1]
U3
U3 E [0 0 0 2 0 0 2 0 0 3 0 1 I]
[0 0 0 1 0 0 1 0 0 1 0 1]
U4
U3 U4
There is a less possibility of getting affected by malaria and it can be pre-

vented by using a mosquito net when we are taking fully covered clothes
is ON state.
Case 4
By choosing M10 in on state.
Let U1 [0 0 0 0 0 0 0 0 0 1 0 0]
U1 E [0 0 0 1 0 0 1 0 0 0 0 1]
[0 0 0 1 0 0 1 0 0 1 0 1]
U2
U2 E [0 0 0 2 0 0 2 0 0 3 0 1 I]
[0 0 0 1 0 0 1 0 0 1 0 1]
U3
U2 U3
When the prevention of disease is ON state, then by the above itera-

tion process, it can be prevented by using mosquito net and fully covered
clothes which lead to less possibility of getting affected.
Case 5
Taking 9th factor in on state. i.e., Possibility of diseases is on state.
U1 [0 0 0 0 0 0 0 0 1 0 0 0]
U1 E [I I 0 0 1 0 0 0 0 0 0 0]
[I I 0 0 1 0 0 0 1 0 0 0]
U2
U2 E [I 2 I I 2 I 1 0 2 0 0 I I 3I 2 1 0 I 0]
[I I 1 0 1 0 0 I 1 0 I 0]
U3
U3 E [I 2 I I 2II 1 0 2 0 0 I I 3I 2 1 0 I 0]
[I I 1 0 1 0 0 I 1 0 I 0]
U4
[I 2 I I 2 I 1 0 2 0 0 I I 3I 2 1 0 I 0]
[I I 1 0 1 I 0 I 1 0 I 0]
U5
U5 E [I 2 2I I 3I 1 0 2 I 0 3I 4 I 2 1 0 I 0]
[II I 1 0 1 I 0 I 1 0 I 0]
U6
U5 U6
M3 and M5 are major factor for the malaria possibility. Here we see some
indeterminate case which means uncertainty factors related to malaria.
With respect to the parameters of Malaria from the above iteration process
on FCM and NCM are briefly furnished with the comparative remarks in
the below Table 15.1.
Table 15.1 Comparison results of FCM and NCM on malaria.

On Comparison
state FCM iteration NCM iteration remarks
M2 T4 = T5; U4 = U5; state 1,6,8
[1 1 1 0 1 1 0 1 1 0 1 0] [I 1 1 0 1 I 0 I 1 0 I 0] and 11 are
having less
possibility
M7 T2 = T3; U3 = U4;
No changes
[0 0 0 1 0 0 1 0 0 1 0 1] [0 0 0 1 0 0 1 0 0 1 0 1]
M12 T3 = T4; U3 = U4;
No changes
[0 0 0 1 0 0 1 0 0 1 0 1] [0 0 0 1 0 0 1 0 0 1 0 1]
M10 T3 = T4; U2 = U3;
No changes
[0 0 0 1 0 0 1 0 0 1 0 1] [0 0 0 1 0 0 1 0 0 1 0 1]
M9 T5 = T6; U5 = U6; parameter
[1 1 1 0 1 1 0 1 1 0 1 0] [I I 1 0 1 I 0 I 1 0 I 0] 2,6,8, and
11 are
having low
possibility
level
15.5 Comparison and Discussion

The parameters related to the malaria are considered as M1, M2, M3, ···,
M12. The FCM and NCM among the parameters are designed based on
the experts’ opinions. The adjacent matrix of fuzzy and neutrosophic cog-
nitive maps are evaluated, and also it is used for the iteration process. The
comparison is made between the both FCM and NCM, by considering any
one state as state commonly. From which we have obtained the following
results.
In this chapter, a comparison results shows that the state 1, 6, 8, and 11
are causing malaria due to lack of insect repellent. No changes in FCM
and NCM for M7, M12 and M10, i.e., a same solution for less possibility of
malaria, use fully covered cloths and prevention for malaria due to inter-
link between the parameter. In the last case, we used possibility of malaria
we get living near stagnant water site and fever factors as ON state.
15.6 Conclusion
In this chapter, comparative study is made on different parameters of the
malaria, which are analyzed by FCM and NCM methods. This study helps
to identify the factors with uncertainty data having high possibilities. Based
on the comparison of parameters via FCM and NCM, we concluded that
positivity of malaria and host respond all the other are indeterminate case.
References
1. Kandasamy, W.V. and Smarandache, F., Fuzzy cognitive maps and neutro-
sophic cognitive maps. Book Review by Milan Mares: Kybernetika, 40, 1, 151,
2004.
2. Papageorgiou, E.I., and Salmeron, J.L., A review of fuzzy cognitive maps
research during the last decade. IEEE Trans. Fuzzy Syst., 21, 1, 66–79, 2012.
3. Uzoka, F.M.E., Akinnuwesi, B.A., Amoo, T., Debele, F., Fashoto, G., Nwafor-
Okoli, C., An expert system for malaria diagnosis using the fuzzy cognitive
map engine, in: 2018 IST-Africa Week Conference (IST-Africa), 2018, May,
IEEE, pp. 1–13.
4. Habib, S. and Akram, M., Medical decision support systems based on fuzzy
cognitive maps. Int. J. Biomath., 12, 06, 1950069, 2019.
5. Sutton, R.T., Pincock, D., Baumgart, D.C., Sadowski, D.C., Fedorak, R.N.,
Kroeker, K.I., An overview of clinical decision support systems: Benefits,
risks, and strategies for success. NPJ Digit. Med., 3, 1, 1–10, 2020.
6. Hoyos, W., Aguilar, J., Toro, M., Dengue models based on machine learn-
ing techniques: A systematic literature review. Artif. Intel. Med., 119, 102157,
2021.
7. Shakil, M.T.A., Ubaid, S., Sohail, S.S., Alam, M.A., A neutrosophic cognitive
map based approach to explore the health deterioration factors. Neutrosophic
Sets Syst., 41, 198, 2021.
8. Hoyos, W., Aguilar, J., Toro, M., A clinical decision-support system for den-
gue based on fuzzy cognitive maps. Health Care Manage. Sci., 25, 4, 666–81,
2022.
9. Douali, N., Papageorgiou, E.I., De Roo, J., Cools, H., Jaulent, M.C., Clinical
decision support system based on fuzzy cognitive maps. J. Comput. Sci. Syst.
Biol., 8, 2, 112, 2015 Mar 1.
16
Applications of Fuzzy Cognitive Maps
and Neutrosophic Cognitive Maps
on Analysis of Dengue Fever
Sathish Kumar Kumaravel1*, Regan Murugesan1, Nagadevi Bala Nagaram1,
Suresh Rasappan2 and G. Yamini1
Department of Mathematics, Vel Tech Rangarajan Dr. Sagunthala R&D

1
Institute of Science and Technology, Avadi, Chennai, Tamil Nadu, India

2
Mathematics Section, College of Technology and Information Sciences, University
of Technology and Applied Sciences - Ibri, Ibri, Sultanate of Oman
Abstract
In this chapter, a comparative study on dengue is made on different parameters.
The parameters involved in dengue and spread of this disease are identified, and
the prevention parameters are concluded. The major benefits of this comparative
study are to identify the parameters which are inconclusive, applying the concept
of FCM and NCM are applied to bring out the result.
Keywords: Fuzzy matrix, fuzzy cognitive map, neutrosophic cognitive map,

dengue
16.1 Introduction
The bite of a mosquito which carries infected Aedes species remains the
cause of spreading dengue viruses. These mosquitoes also spread few other
viruses like chikungunya, zika, etc. Nearly 50% of the population around
the world lives with the risk of getting affected by dengue. Almost 400 mil-
lion people gets infected with the disease every year. Around 40,000 people
lose their life by this dangerous disease [1–3].
249
One of the four viruses dengue, virus 1, 2, 3, 4, remains the reason for
the spread of the disease. Because of this, there is chance of getting affected
by the disease nearly four time for each human being in their life. For this
year (2022) 8278 peoples were affected by dengue in India and only one
death in Puduchery. In those 2,485 cases of dengue in Tamilnadu. Between
January and May, more than two lakhs’ people have been tested for dengue
in the state [4–6]. Those people who were infected were treated at different
hospitals following the standard protocol.
Once the mosquito is bitten a person with the virus, then it affects
the body which shows the symptoms up to 7 days from the beginning.
Symptoms can range from mild infection to severe infection. The diseases
with severe symptoms involve the dengue hemorrhagic fever and dengue
shock syndrome, also muscle pain, headache, vomiting, etc., are the few
more symptoms of the dengue disease [7]. These usually require hospi-
talization. As a preventive measure to avoid the bite of mosquito, insect
repellent, cleaning stagnant areas, using mosquito net for bed, door and
windows can be done, seek early consultation when signs and symptoms of
dengue occur, wear light-colored clothes, etc. Treatment can be positively
done if the diagnosis takes place before the patient gets developed with
DSS or DHF.
Dengue vaccine: For children with an age group from 9 to 16, a new
vaccine for dengue has been approved for usage. The vaccine is laboratory-
confirmed and tested with previously infected persons. It is being used at
the areas at which the virus is endemic. Those areas include some of the
territories of the U.S, and the states associated with it. The vaccine is not
allowed for usage at the areas where dengue is not that much affected and
also for the travellers who visits the U.S.
The most serious obstacle that develops by the disease is plasma leak-
age. This remains the symptom of severe stage of dengue. The fluid like
component from blood vessels gets leaked into the tissues surrounding it.
This process is known as plasma leakage. In some severe infected dengue
patients, the fever begins to disappear.
In this chapter, a comparative study is made on the different parame-
ters related to the disease dengue, such as travelling area, low antibody,
vomiting, high fever, etc. The parameters are analyzed by taking any one
as ON state by the concept FCM. Simultaneously those parameters are
analyzed through the NCM as like FCM by considering the same state
as ON state.
Applications of FCM and NCM on Analysis of Dengue 251
The aim of this chapter is to analyze the disease dengue through com-
parison between FCM and NCM among the parameters related to dengue.
The following is the outline of the contents of this chapter. In section 16.2,
parameters respect to dengue is presented. In the rest of the chapter, the
comparative study is presented. In section 16.3 the fuzzy cognitive maps,
and its matrix form is discussed. Section 16.4 describes the neutrosophic
cognitive maps and its matrix form. Finally, in section 16.5, the compari-
son of results from the FCM and NCM are discussed.
16.2 Parameters of Dengue

In order to analyze the parameters of dengue, data were collected from the
medical experts. Based on their opinions, the following factor were identi-
fied and collected.
D1-Travelling area
D2-Low antibody
D3-Possibility of dengue
D4-Vomiting
D5-High risk of dengue
D6-Wear fully covered cloths
D7-Host response
D8-High fever
D9-Muscle pain
D10-Headache
D11-Use mosquito repellent
D12-Less possibility of dengue
16.3 Fuzzy Cognitive Maps

The expert’s opinion was collected and based on their opinion, a graph is
formed by mapping between the parameters. Based on the fuzzy cognitive
map concept, weight age was assigned by the casual relation between the
nodes, i.e., edge weight was assigned as 1 (positive causality between the
nodes), if the relationship between the nodes had majority of respondents,
at the same time majority of respondents was uncertain then it denoted by
0. The corresponding fuzzy cognitive maps for the parameters are given
below in Figure 16.1.
D1
D3
D7
D10
D8
D2
D4
D6 D5
D12
D11
D9
Figure 16.1 Fuzzy cognitive maps on dengue.
16.3.1 Matrix Representation of FCM

The matrix representation of fuzzy cognitive map is designed based on
connectives between the nodes which are the possibility among the param-
eters. The entries of the matrix are noted either as 0 or 1. The number is 0
denoted for unconnected and 1 represents connected between the nodes.
The adjacent matrix of Figure 16.1 is given below.
W D1 D2 D3 D4 D5 D6 D7 D8 D9 D10 D11 D12
D1 0 0 1 0 0 0 1 0 0 0 0 0
D2 0 0 0 0 0 0 0 0 1 0 0 0
D3 1 0 0 0 0 0 0 1 0 0 0 0
D4 0 0 0 0 0 0 0 0 0 1 0 0
D5 1 0 0 0 0 0 1 0 0 0 0 0
E D6 0 0 0 0 0 0 0 0 0 0 0 1
D7 0 0 0 0 1 0 0 1 0 0 0 0
D8 0 1 1 0 0 0 0 0 0 0 0 0
D9 0 0 1 0 0 0 0 0 0 0 0 0
D10 0 0 0 0 0 0 0 0 0 0 0 1
D11 0 0 0 0 0 0 0 0 0 0 0 1
D12 0 0 0 1 0 1 0 0 0 0 1 0
Case 1
Let us take D12 is ON state. Then
A1 [0 0 0 0 0 0 0 0 0 0 0 1]
A1 E [0 0 0 1 0 1 0 0 0 0 1 0]
[0 0 0 1 0 1 0 0 0 0 1 1]
A2
A2 E [0 0 0 1 0 1 0 0 0 1 1 2]
[0 0 0 1 0 1 0 0 0 1 1 1]
A3
A3 E [0 0 0 1 0 1 0 0 0 1 1 3]
[0 0 0 1 0 1 0 0 0 1 1 1]
A4
A3 A4
When we are taking less possibility of dengue in the state of ON, it gives
that we need to use fully covered clothes and mosquito repellent. Headache
and vomiting also have a low possibility level of effect.
Case 2
Considering Muscle pain (M9) as ON state.
Let A1 [0 0 0 0 0 0 0 0 1 0 0 0]
A1 E [0 0 1 0 0 0 0 0 0 0 0 0]
[0 0 1 0 0 0 0 0 1 0 0 0]
A2
A2 E [1 0 1 0 0 0 0 1 1 0 0 0]
[1 0 1 0 0 0 0 1 1 0 0 0]
A3
A3 E [1 1 3 0 0 0 1 1 1 0 0 0]
[1 1 1 0 0 0 1 1 1 0 0 0]
A4
A4 E [1 1 3 0 0 0 1 2 2 0 0 0]
[1 1 1 0 0 0 1 1 1 0 0 0]
A5
A4 A5
A person may have a possibility of dengue if they had lower antibody

level, also with high fever and it affect the travelling area then it causes host
response.
Case 3
By choosing (A6) wear fully covered clothes in ON state. Let
A1 [0 0 0 0 0 1 0 0 0 0 0 0]
A1 E [0 0 0 0 0 0 0 0 0 0 0 1]
[0 0 0 0 0 1 0 0 0 0 0 1]
A2
A2 E [0 0 1 0 1 0 0 0 0 0 1 1]
[0 0 1 0 1 0 0 0 0 0 1 1]
A3
A3 E [0 0 0 1 0 1 0 0 0 1 1 2]
[0 0 0 1 0 1 0 0 0 1 1 1]
A4
A4 E [0 0 0 1 0 1 0 0 0 1 1 1]
[0 0 0 1 0 1 0 0 0 1 1 1]
A5
A4 A5
It clearly shows that there is less possibility of disease by using fully

covered dresses and mosquito net. Otherwise, it causes headache and
vomiting.
Case 4
Let us take D7 (vomiting) as ON state. Then
A1 [0 0 0 1 0 0 0 0 0 0 0 0]
A1 E [0 0 0 0 0 0 0 0 0 1 0 0]
[0 0 0 1 0 0 0 0 0 1 0 0]
A2
A2 E [0 0 0 0 0 0 0 0 0 1 0 1]
[0 0 0 1 0 0 0 0 0 1 0 1]
A3
A3 E [0 0 0 1 0 1 0 0 0 1 1 1]
[0 0 0 1 0 1 0 0 0 1 1 1]
A4
A4 E [0 0 0 1 0 1 0 0 0 1 1 1]
[0 0 0 1 0 1 0 0 0 1 1 1]
A5
A4 A5
Vomiting is a less possibility of causing disease. Headache, using insect

repellent and fully covered dresses are also less possibility of dengue.
Case 5
Let D1 is ON state. Then
A1 [1 0 0 0 0 0 0 0 0 0 0 0]
A1 E [0 0 1 0 0 0 1 0 0 0 0 0]
[1 0 1 0 0 0 1 0 0 0 0 0]
A2
A2 E [1 0 1 0 1 0 1 2 0 0 0 0]
[1 0 1 0 1 0 1 1 0 0 0 0]
A3
A3 E [2 1 2 0 1 0 2 2 1 0 0 0]
[1 1 1 0 1 0 1 1 1 0 0 0]
A4
A4 E [2 1 3 0 1 0 2 2 2 0 0 0]
[1 1 1 0 1 0 1 1 1 0 0 0]
A5
A4 A5
Except preventive measures all the others are having connection with D1.
16.4 Neutrosophic Cognitive Map

The expert’s opinion was collected and based on their opinion; a graph is
formed by mapping between the parameters. Based on the neutrosophic
cognitive map concept, weight age was assigned by the casual relation
between the nodes, i.e., edge weight was assigned as 1 (positive causality
between the nodes), if the relationship between the nodes had majority
of respondents, at the same time the respondents which are uncertain or
indeterminate then it is denoted by I. The number zero is assigned, when
there is no relationship between the parameters based on expert’s opinion.
In the neutrosophic graph, for the indeterminate case the edges between
the nodes are drawn by the dotted lines. The corresponding neutrosophic
cognitive maps for the parameters are given below in Figure 16.2.
D1
D3
D7
D10
D8
D2
D4
D5
D6
D11
D12
D9
Figure 16.2 Neutrosophic cognitive maps on dengue.
16.4.1 Matrix Representation of NCM

The matrix representation of the neutrosophic cognitive map is designed
based on the connectives between the nodes which are the possibility
among the parameters. In addition to the FCM concept, here one more
case is occurred, when there is an inconclusive possibility of relationship
between the parameters that is denoted as indeterminate I case. The entries
of the matrix are noted as 0 or 1 or I. The number 0 denoted for uncon-
nected, 1 represents for connected between the nodes and I noted for the
indeterminate case between the nodes which is connected by the dotted
lines. The adjacent matrix of Figure 16.2 is given below.
W D1 D2 D3 D4 D5 D6 D7 D8 D9 D10 D11 D12

D1 0 0 1 0 0 0 I 0 0 0 0 0
D2 0 0 0 0 0 0 0 0 1 0 0 0
D3 1 0 0 0 0 0 1 I 0 0 0 0
D4 0 0 0 0 0 0 0 0 0 1 0 0
D5 I 0 0 0 0 0 1 0 0 0 0 0
E D6 0 0 0 0 0 0 0 0 0 0 0 1
D7 0 0 0 0 1 0 0 I 0 0 0 0
D8 0 0 0 I 0 0 0 0 1 0 0 0
D9 0 0 I 0 0 0 0 0 0 0 0 0
D10 0 0 I 0 0 0 0 0 0 0 0 1
D11 0 0 0 0 0 0 0 0 0 0 0 1
D12 0 0 0 0 0 1 0 0 0 0 1 0
Case 1
Taking in D12 ON state, i.e., the factor less possibility of dengue is on state.
Let B1 [0 0 0 0 0 0 0 0 0 0 0 1]
B1 E [0 0 0 0 0 1 0 0 0 0 1 0]
[0 0 0 0 0 1 0 0 0 0 1 1]
B2
B2 E [0 0 0 0 0 1 0 0 0 0 1 2]
[0 0 0 0 0 1 0 0 0 0 1 1]
B3
B2 B3
Using fully covered dresses and insect repellent are less possibility of
dengue.
Case 2
By choosing muscle pain (D9) is ON state. Let
B1 [0 0 0 0 0 0 0 0 1 0 0 0]
B1 E [0 0 I 0 0 0 0 0 0 0 0 0]
[0 0 I 0 0 0 0 0 1 0 0 0]
B2
B2 E [I 0 I 0 0 0 I I 2 0 0 0 0]
[I 0 I 0 0 0 I I 1 0 0 0]
B3
B3 E [I 0 2I I 2 I 0 I2 I I2 I I 0 0 0]
[I 0 I I I 0 I I 1 0 0 0]
B4
B4 E [I I 0 2 I 0 I 0 I 2 2 I I I I 0 0 0]
[I 0 I I I 0 I I 1 0 0 0]
B5
B4 B5
Muscle pain did not cause any serious risk because it may occur due to
weakness, older age, etc.
Case 3
Choosing (D6) fully covered clothes usage as on state.
Let B1 [0 0 0 0 0 1 0 0 0 0 0 0]
B1 E [0 0 0 0 0 0 0 0 0 0 0 1]
[0 0 0 0 0 1 0 0 0 0 0 1]
B2
B2 E [0 0 0 0 0 1 0 0 0 0 1 1]
[0 0 0 0 0 1 0 0 0 0 1 1]
B3
B3 E [0 0 0 0 0 1 0 0 0 0 1 2]
[0 0 0 0 0 1 0 0 0 0 1 1]
B3 B4
B3 B4
Preventions of malaria have a less chance of malaria positive. Prevention

is using fully covered dress and repellent.
Case 4
By taking B4 as on state. Then we can take,
B1 [0 0 0 1 0 0 0 0 0 0 0]
B1 E [0 0 0 0 0 0 0 0 0 1 0 0]
[0 0 0 1 0 0 0 0 0 1 0 0]
B2
B2 E [0 0 I 0 0 0 0 0 0 1 0 0]
[0 0 I 1 0 0 0 0 0 1 0 0]
B3
B3 E [I 0 I 0 0 0 I I2 0 1 0 0]
[I 0 I 1 0 0 I I2 0 1 0 0]
B4
B4 E [I I2 2I I2 I 0 I2 I 2I2 1 0 0]
[I I I 1 I 0 I I I 1 0 0]
B5
B5 E [I I 2 I2 I2 2I I2 I 0 I2 2I 2I2 2I 1 0 0]
[I I I 1 I 0 I I I 1 0 0]
B6
B5 B6
when we are taking vomiting as on state then it causes headache. Except

prevention, all others are indeterminate.
Case 5
By choosing travelling area as on state, let
B1 [1 0 0 0 0 0 0 0 0 0 0 0]
B1 E [0 0 1 0 0 0 I 0 0 0 0 0]
[0 0 1 0 0 0 I 0 0 0 0 0]
B2
B2 E [1 0 1 0 I 0 I 1 I I2 I I2 0 0 0]
[1 0 1 0 I 0 1 I I 0 0 0]
B3
B3 E [I 2 I 2 I2 1 I I2 0 0 1 I 2I I 0 0 0]
[1 I 1 I 0 0 1 I I 0 0 0]
B4
B4 E [I 2 I 2 I2 1 I I2 0 0 1 I 2I 2I I 0 0]
[1 I 1 I 0 0 1 I I I 0 0]
B5
B5 E [I 2I2 I2 1 2I I 0 0 1 I I I I 0 0]
[1 I 1 I 0 0 1 I I I 0 0]
B6
B5 B6
While we are taking travelling area in the ON state, then it is possible of

getting dengue and host response. Except the prevention all the others are
in the state of indeterminate, which means uncertainty factors related to
the disease dengue.
16.5 Comparison and Discussion

The parameters related to the dengue are considered as D1, D2, D3, ..., D12
The FCM and NCM among the parameters are designed based on the
experts’ opinions. The adjacent matrix of fuzzy and neutrosophic cognitive
maps are evaluated and it is used for the iteration process. The comparison
is made between the both FCM and NCM, by considering any one state as
state commonly. From which, the results obtained as given below.
In first case, by taking D12 as on state, the result obtained that D4 and D10
as OFF state in NCM while it is ON in FCM. Next case show that possibil-
ity of diseases is indeterminate state. For, third case same as first case 4th
and 10th parameters in OFF state. Now, for D4 headache is ON state. Due
to our travelling history, we get a possibility of dengue and host respond
all the other are in indeterminate case. The parameters of Dengue from the
above iteration process on FCM and NCM are briefly explained with the
comparative remarks in the below Table 16.1.
Table 16.1 Comparison results of FCM and NCM on dengue.
On Comparison
state FCM iteration NCM iteration remarks
D12 A3 = A4; B2 = B3; Differs in state
[0 0 0 1 0 1 0 0 0 1 1 1] [0 0 0 0 0 1 0 0 0 0 1 1] 4 and 10
D9 A4 = A5; B4 = B5; The possibility
[1 1 1 0 0 0 1 1 1 0 0 0] [1 0 1 1 1 0 1 1 1 0 0 0] of disease is
very low
D6 A4 = A5; B3 = B4; Off states are
[0 0 0 1 0 1 0 0 0 1 1 1] [0 0 0 0 0 1 0 0 0 0 1 1] 4 and 10
D4 A4 = A5; B5 = B6; Here
[0 0 0 1 0 1 0 0 0 1 1 1] [1 1 1 1 1 0 1 1 1 1 0 0] possibility
of headache
is vomiting
only
D1 A4 = A5; B5 = B6; Most
[1 1 1 0 1 0 1 1 1 0 0 0] [1 1 1 1 0 0 1 1 1 1 0 0] possibility
of is 3rd
and 7th
states.
16.6 Conclusion
In this chapter, comparative study is made on different parameters of the
dengue, which are analysis by FCM and NCM methods. Based on the cal-
culation process on FCM and NCM separately by considering different
parameter as ON state, we concluded that positivity of dengue and host
respond all the other are indeterminate case. The obtained results on dif-
ferent cases of FCM and NCM are analyzed by comparative study.
References
1. Fernández, E., Smieja, M., Walter, S.D., Loeb, M., A predictive model to dif-
ferentiate dengue from other febrile illness. BMC Infect. Dis., 16, 1, 1–7, 2016.
2. Gambhir, S., Malik, S.K., Kumar, Y., PSO-ANN based diagnostic model for
the early detection of dengue disease. New Horiz. Transl. Med., 4, 1–4, 1–8,
2017.
3. Qsim, M., Ashfaq, U.A., Yousaf, M.Z., Masoud, M.S., Rasul, I., Noor, N.,
Hussain, A., Genetically modified Aedes aegypti to control dengue: A review.
Crit. Rev.™ Eukaryot. Gene Expr., 27, 4, 331–340, 2017.
4. Caicedo-Borrero, D.M., Tovar, J.R., Méndez, A., Parra, B., Bonelo, A., Celis,
J., Osorio, L., Development and performance of dengue diagnostic clinical
algorithms in Colombia. Am. J. Trop. Med. Hyg., 102, 6, 1226, 2020.
5. Wong, P.F., Wong, L.P., AbuBakar, S., Diagnosis of severe dengue: Challenges,
needs and opportunities. J. Infect. Public Health, 13, 2, 193–198, 2020.
6. Sutton, R.T., Pincock, D., Baumgart, D.C., Sadowski, D.C., Fedorak, R.N.,
Kroeker, K.I., An overview of clinical decision support systems: Benefits,
risks, and strategies for success. NPJ Digit. Med., 3, 1, 1–10, 2020.
7. Hoyos, W., Aguilar, J., Toro, M., Dengue models based on machine learning
techniques: A systematic literature review. Artif. Intell. Med., 119, 102157,
2021.
17
A Comprehensive Review and
Analysis of the Plethora of Branches
of Medical Science and Bioinformatics
Based on Fuzzy Logic
Partha Sarker1 and Siddhartha Roy2*
1
Department of Computer Science, The University of Burdwan, West Bengal, India
2
Department of Computer Science, The Heritage College (Affiliated to University of
Calcutta), Kolkata, India
Abstract
This chapter represents comprehensive and analytical views of the latest research
work on bioinformatics and various branches of medical science using fuzzy set
theory. Fuzzy set theory is felicitous for developing based on knowledge systems
in various fields of medicine for example drug discovery, finding diseases, the best
way of medical treatments, and obtaining real-time information based on the data
provided by the patients. It is very difficult to achieve efficient and effective med-
ical work without expert system-based fuzzy logic (ESFL). This chapter reviews
the application of the ESFL in medical sciences in general and will consider the
ways of in-depth analysis of the applied fuzzy logic and implementation process
in those fields.
Keywords: ESFL, disease diagnosis, uncertainty, fuzzy mathematics,

bioinformatics, drug discovery, medical science
17.1 Introduction
In fuzzy mathematics, the fuzzy concept is a multi-valued formed with
truth values between 0 and 1 both inclusive. In 1956 at the University of
Berkeley, California, Professor Lotfi Ahmad Zadeh, in his paper which was
*Corresponding author: [email protected]; [email protected]
267
presented in a seminar “fuzzy sets” first introduced the basic concept of

fuzzy logic. The term fuzzy defines are neither known nor vague. In the
real field, we come across various situations when the truth values are not
Boolean, i.e., true (value is 1.1) or false (value is 0.0), but we may obtain
some intermediaries values in between that using fuzzy logic. In a fuzzy
world, there is the value from the intermediate position, i.e., partially true
or partially false value, there is no such of absolute truth (1) and absolute
false (0) value. Figure 17.1 illustrates the concept which is Boolean, as well
as fuzzy systems.
Fuzzy based system is used for the most diagnosis processes which are
based on the probability measurement of medical finding. Researchers
develop various models in the field of health care based on fuzzy methods
to demonstrate the uncertainty of various applications for different kinds
of illnesses. In fuzzy-based system, symptoms of illness are used for input,
and the disease of a patient for output. Artificial intelligence technology is
used in the fuzzy system to create fuzzy classes to act intelligently. The fuzzy
classes are used to define the transformation of the input values to new
values based on a specified function. Architecturally, there are four com-
ponents which are fuzzy logic system fuzzifier, rule-based, and defuzzifier.
The fuzzifier is used to transform a crisp value into a fuzzy value with the
use of knowledge base information. The resultant value of the fuzzy sets
is processed through the inference engine of the fuzzy logic system. The
knowledge-based architecture is the main component of a fuzzy logic sys-
tem. The knowledge base of a fuzzy system consists of a rule-based and
a database. A rule-based approach is constructed through a self-learning
algorithm. In this approach whole dataset is split into two parts, the first
part is used for training purposes whereas another part is used for valida-
tion purposes which are tested through a neuro-fuzzy system. The final
Yes (1.0)
Is it raining Boolean
today? logic
No (0.0)
Heavy (0.9)
Medium (0.5)
Is it raining Fuzzy
today? Low (0.25) logic
Very low (0.1)
Figure 17.1 Structure of boolean and fuzzy logic.

Fuzzy Logic in Medical Science & Bioinformatics 269
component of the fuzzy process is defuzzifier. Generally, the resultant out-

come processed by the inference engine is also a fuzzy set, which is not
useful in a real-world application. Thus, this crisp value needs to be con-
verted into a suitable real-life application. Defuzzification is the process of
making production a quantifiable result in Crisp logic, given fuzzy. Figure
17.2 depicts the architectural components of fuzzy logic.
The diagnosis of disease comprises of indecision and ambiguity. A sin-
gle disease may have different impact which will depend on the patient’s
immunity and distinguishable intensities. Different diseases may occur due
to a variety of symptoms. The summary report of the disease may be inac-
curate and ambiguous. Disease, illness, and sickness are three forms of loss
of health. Fuzzy is used in the medical decision-making process. A physi-
cian can predict diseases and accordingly diagnosis through linguistic per-
ception. Fuzzy set theory helps the physician to handle linguistic concepts.
A recent study shows that fuzzy logic is applied successfully in each area of
health care services. Through the fuzzy system, the preliminary diagnosis
stage can be anticipated. There is always uncertainty in the diagnosis of
diseases. The initial source of the patient’s medical history is obtained from
various laboratory tests. Complexity and vagueness of diagnosis increase
due to the high volume of data in spite of modern medical equipment and
tools which may affect the physicians to take the correct decision. It is
impossible to take a correct decision without gathering sufficient infor-
mation at a right time. Besides that, errors in diagnosis are unavoidable. A
patient suffering from various diseases has multiple symptoms is difficult
to diagnose the root cause of the disease. The solution to the diagnosis of
the disease is to assume some basic one or more possible hypotheses. To
reduce the complication of the diagnosis process and to improve accuracy,
Knowledge
Base
Fuzzifier Inference Defuzzifier

Engine
Process output & Crisp control

state signal
Process
Figure 17.2 Architecture of fuzzy process.

artificial intelligence has been employed augmented with the fuzzy system.
Handling uncertain situations and finding errors in disease diagnosis is
very vital as it can be helpful to medical practitioners to diagnose a patient.
Due to the lack of sufficient information, the fuzzy method was developed
as the most effective method to handle diagnosis uncertainty. In applied
fuzzy logic, the diagnosis of a disease can be defined in terms of the degree
of membership. In classical logic, the sickness can only be circumscribed
with a true (1) and false (0), in a fuzzy system; the sickness can be drawn
as a sketch with partial truth values. The following figure (Figure 17.3)
depicts the Fuzzy process to diagnose the disease using a flowchart.
After introductory section we discuss previous work in section 17.2.
Section 17.3 presents the deployment of fuzzy logic and fuzzy system in
the area of medical and bioinformatics. A review of some previous work,
methods, and findings is made a discussable talk in Section 17.4. Finally,
this paper ends with some concluding remarks and scope.
Start
Identify Fuzzy Method and

Select Model
Set input symptoms
Fuzzyfying all input values

Fuzzy membership
function
Generate Fuzzy Rule
Inference Engine
Defuzzyfying the fuzzy

output value
Get Output as a disease
Stop
Figure 17.3 Flow diagram of the fuzzy process to diagnosis the disease.
17.2 Previous Work

The term “fuzzy” refers to “confusion.” In fact, there will be always a cer-
tain degree of uncertainty and vagueness in all types of real-world prob-
lems. None can take a decision correctly without complete information.
The effectiveness of medicine possesses a high rate of uncertainty and is
highly complex in human health and diseases. Recently fuzzy logic model
has been widely applied in various types of health care treatment, such as
diabetes, cancer, etc. [1]. In contemporary health care, many diseases, such
as cancers, and diabetes, are an amalgamation of various influences and
are classified as complex diseases. This complex disease is easily carried
out by a fuzzy system, published by Sadegh-Zadeh [2]. Present percep-
tion and measures the disease to a certain degree to a fuzzy concept. A
fuzzy logic controller (FLC), which is based on a fuzzy set, is used as the
control unit in different medical equipment [3]. For human behavior and
to evaluate facial expression, fuzzy logic is extensively used [4]. Fuzzy cog-
nitive map (FCM) is used to investigate the immune system of humans
in a diagrammatic representation discussed in the paper by Faran et al.
[5]. How coronary heart disease can be predicted using a fuzzy decision
support system (FDSS) is discussed in the paper published by Alfons et al.
[6] Allahverdi [7], who designed a fuzzy expert system deployed in various
medical systems.
17.3 Fuzzy Logic in Medical Fields

and Bioinformatics
Several medical fields are widely deployed by the fuzzy-based expert
system. For the last few decades, increasing the accuracy of diagno-
sis and complexity has decreased due to the integration of medical
diagnostic methods and computational methods. Various researchers
show a significant contribution toward the effectiveness of the fuzzy
system in the health care system, especially in the medical fields and
bioinformatics.
17.3.1 Applied Fuzzy Logic in Medical Areas

In medical areas, the deficiency of useful data, its vagueness, and the
self-contradictory nature of the disease is very common. As discussed
previously adopting fuzzy logic, in this case, plays a very crucial role.
Using neuro fuzzy optimization techniques, various kinds of protein
motifs can be extracted [8, 9]. Analyzing diabetic neuropathy and detec-
tion of early diabetic retinopathy can be implemented using fuzzy logic.
It has enormous applications and uses in various medical areas, such as
determining appropriate lithium dosage. A fuzzy system is used to pre-
dict brain diseases for medical image processing. A fuzzy system is also
used to take the decision to improve radiation therapy and predict pros-
tate cancer, lung cancer, and breast cancer. Discriminating benign with
malignant cells and estimating drug discovery are extensively used by
fuzzy systems.
17.3.2 Applied Fuzzy Logic in Bioinformatics

Bioinformatics is the field to analyze biological data using algorithms
implemented using a computer. Bioinformatics is a field that combines
biology and computer science for the investigation and analysis of huge
sets of biological data. This biological data are stored in the genetic code.
Bioinformatics decreases the cost of research by accurately predicting
models. Fuzzy set theories are now frequently used in the domain of bioin-
formatics where the information is incomplete or inaccurate.
A fuzzy set is used to develop the flexibility of the short-conserved
sequence pattern of DNA. The objective of the fuzzy is to extract a simi-
larity pattern in the motif of a protein belonging to sequences of the same
family. There are four steps in motif extraction, namely sequence prepro-
cessing, motif generation, motif selection, and motif optimization [9, 10].
Fuzzy polynucleotide which is an n-dimensional metric space is used to
study the differences between polynucleotides. Phylogenetic associations
among organisms are established in a molecular sequence. Data collection,
sequence analysis, and phylogenetic analysis are three parts of this phase
[11]. The clustering method is used to analyze experimental data by fuzzy
adaptive resonance. Understanding the functioning of uncharacterized
genes using the Fuzzy ART algorithm [12] become easier. A significant
problem in computational biology is identifying differences and similar-
ities between organisms using the DNA sequence alignment technique.
The fuzzy logic method plays an important role in DNA sequencing [13].
The fuzzy logic is used to analyze gene expression which helps biologists
for monitoring the expression of thousands of genes in a single small chip
called a biochip, gene chip, or DNA chip [14]. A genetic network process
is used to increase the efficiency of fuzzy logic-based gene expression [15].
A fuzzy rule-based classifier is used to extract feature selection and catego-
rize amino acid sequences [16].
17.4 Review of Published Work and In-Depth

Analysis
In the review section, we briefly discuss the application of the fuzzy system
in various diseases, its research goal, methodology, and findings in Table
17.1. The information is collected from various research papers, including,
WoS, Wiley Online Library, Elsevier, IEEE, etc.
From the above table, it is clear that in most cases to detect diagnosis the
accuracy rate is higher than 80% using a fuzzy system. But in most cases,
we need more resources and data to get a more accurate result.
17.5 Conclusion
Recent advancements in soft computing techniques, fuzzy logic make suc-
cessful applications in health care. In this paper, we discuss the effective-
ness of the fuzzy set theory in detecting the diagnosis of disease, and its
utilization in the field of bioinformatics, and medical science. To achieve
the target, we have searched various scientific journals from the years 2000
to 2021. All the research journals are based on published years, methodol-
ogy used in various kinds of diseases, outcomes, and finally the influence
of the fuzzy logic that is deployed to identify disease. The most important
medical areas we have studied were related to heart daises Ebola, diabetes,
and cholera, respectively, using various fuzzy logic set methods. It has been
observed that more than 90% of fuzzy system has a positive impact on
disease analysis. Our future work will focus on fuzzy artificial neural net-
works to establish an intelligent system to detect and identify diseases more
accurately. Future steps are to develop an intelligent fuzzy expert system to
diagnose patient disease more accurately and would estimate parameters,
such as specificity and sensitivity.
Table 17.1 Review of various diseases using the fuzzy system.

SL. no. Symptom Research objective Procedure Performance/uses Conclusion
1 Heart To make a Neuro-Fuzzy The results show an Apart from the accuracy
Disease measurement of Integrated System accuracy rate of 93% rate can be higher it
the accuracy of (NFIS) and a tool will be used in another
the system for with database system in order.
heart disease. design, are used in
MATLAB [17].
2 Ebola To make web The “Root Sum Ebinformatics software The multi-language
informatics Square” (RSS), was was used to fight facility should be
named software adopted Using Fuzzy against Ebola-affected incorporated in future
using the fuzzy Inference and the medical version software
application to Fuzzy logic Expert
diagnose the Ebola System was used
virus especially [18].
recommended in
West Africa -
(Continued)
Table 17.1 Review of various diseases using the fuzzy system. (Continued)
3 Brain To make detection k-means clustering, By this algorithm brain, The number of MRI plates
Tumor of a brain tumor a fuzzy-based tumors (for the is limited. A large
in the MRI plate approach is used to human only) in MRI number of MRI plates
by improving the locate the tumor in plates can be located are required for proper
segmentation the MRI of the brain with segmentation segmentation.
process. [19]. based
4 Parkinson To detect Parkinson’s The proposed With this fuzzy The c-means clustering
disease disease a fuzzy- algorithm, the approach a higher approach is higher
based algorithmic C-means-based accuracy rate of 96% than another one. So
system was fuzzy inference to 97%. this diagnosis is very
made from the system was used to dependent on only one
biomedical make detection of fuzzy-based approach
measurement of Parkinson’s disease to get higher accuracy.
voice signals. [20].
5 COVID To detect early A smart fuzzy inference It gives a rate of 90% More countries should be
19 symptoms of system was proposed accuracy from 11 incorporated mobile
Covid 19 disease to detect covid 19 countries app-based service
based on the disease [21]. systems should be
symptoms with incorporated in the
the help of a fuzzy future.
algorithm.
(Continued)
Table 17.1 Review of various diseases using the fuzzy system. (Continued)
6 Diabetes To diagnose Mamdani’s fuzzy This approach has given No. of patients’ data
Disease diabetes disease inference system 96% accuracy to the set is very small (In
a fuzzy inference (MFIS) was used to 50 patients. the future huge no.
approach was diagnose diabetic of patients should
implemented disease [22]. participate in this
survey), and Web-based
and mobile-based apps
should be incorporated.
7 Dental To detect dental Hybrid Fuzzy inference Hybrid FIS gave an After comparing between
Disease disease fuzzy systems (FIS) accuracy rate of 82% two approaches the
logic and an and Evolutionary and the Evolutionary Evolutionary algorithm
evolutionary algorithms are used algorithm gave an is more accurate than
algorithm were to detect dental accuracy rate of 70%. then the other, only 100
used. diseases [23]. patients were used for
this survey.
References
1. Sadegh-Zadeh, K., Fuzzy health, illness, and disease. J. Med. Philos., 25, 5,
605–638:612, 2000.
2. Lancaster, S., A fuzzy logic controller for the application of skin pressure,
Creighton University, 2004.
3. Mahfouf, M., Abbod, M.F., Linkens, D.A., A survey of fuzzy logic monitoring
and control utilization in medicine, Department of Automatic Control and
System Engineering, University of Sheffield, Sheffield, England, 2000.
4. Yue, H., Yue, G., Yi, G., Application study in decision support with fuzzy
cognitive map. Int. J. Comput., 1, 140–144, 2007.
5. Faran, B., Saleem Khan, M., Noor, Y., Imran, M., Design model of fuzzy
logic medical diagnosis control system. Int. J. Comput. Sci. Eng. (IJCSE), 3, 5,
2093–2108, 2011.
6. Schuster, A., Adamson, K., Bell, D.A., Fuzzy decision support system in the
domain of coronary heart disease risk assessment. Fuzzy Log. Med., 177–204,
2002.
7. Allahverdi, N., Design of fuzzy expert systems and its applications in some
medical areas. Int. J. Appl. Math. Elec. Comput., 2, 1, 1–8, 2014.
8. Angela, T. and Nieto, J.J., Fuzzy logic in medicine and bioinformatics. J.
Biomed. Biotechnol., 2006, 1–7, 2006.
9. Chang, B.C. and Halgamuge, S.K., Protein motif extraction with neuro-fuzzy
optimization. Bioinformatics, 18, 8, 1084–1090, 2002.
10. Xu, D. et al., Applications of fuzzy logic in bioinformatics. Series on Advances
in Bioinformatics and Computational Biology, vol. 9, pp. 196–221.
11. Torres, A. and Nieto, J.J., The fuzzy polynucleotide space: Basic properties.
Bioinformatics, 19, 5, 587–592, 2003.
12. Tomida, S., Hanai, T., Honda, H., Kobayashi, T., Analysis of expression pro-
file using fuzzy adaptive resonance theory. Bioinformatics, 8, 8, 1073–1083,
2002.
13. Cordón, O., Gomide, F., Herrera, F., Hoffmann, F., Magdalena, L., Ten years
of genetic fuzzy systems: Current framework and new trends. Fuzzy Sets
Syst., 41, 1, 5–31, 2004.
14. Belacel, N., Čuperlović-Culf, M., Laflamme, M., Ouellette, R., Fuzzy J-Means
and VNS methods for clustering genes from microarray data. Bioinformatics,
20, 11, 1690–1701, 2004.
15. Ressom, H., Reynolds, R., Varghese, R.S., Increasing the efficiency of fuzzy
logic-based gene expression data analysis. Physiol. Genomics, 13, 2, 107–117,
2003.
16. Bandyopadhyay, S., An efficient technique for superfamily classification of
amino acid sequences: Feature extraction, fuzzy clustering and prototype
selection. Fuzzy Sets, 152, 1, 5–16, 2005.
17. Padmavathi, K., Meenakshi, K., Swaraja, K., Rajani, A., KafiulIslam, Md,
Detection of cardiac arrhythmia using fuzzy logic. Inf. Med. Unlocked, 17,
1–7, 2019.
18. Oluwagbemi, O., Oluwagbemi, F., Abimbola, O., Ebinformatics, Ebola fuzzy
informatics systems on the diagnosis, prediction and recommendation of
appropriate treatments for Ebola. 12–37, 2016.
19. Neha, M., Yogesh Kumar, M., Shruti, M., Divya, M., Detection of brain tumor
in MRI image through fuzzy. Intech-Open Sci., 4, 47–62, 2018.
20. Atanu, C. and Bhaskar, M., Detection of Parkinson’s disease using fuzzy
inference system, in: Intelligent Systems Technologies and Applications, vol. 1,
pp. 79–90, 2015.
21. Shatnawi, M., Anas, S., Zakaria Al, S., Ghaith, H., Symptoms-based fuzzy-
logic approach for COVID-19 diagnosis. Int. J. Adv. Comput. Sci. Appl.
(IJACSA), 12, 4, 444–452, 2021.
22. Niswati, Z., Mustika, F.A., Paramita, A., Fuzzy logic implementation for
diagnosis of diabetes mellitus disease at Puskesmas in East Jakarta, WMA-
Mathcomtech 2018 IOP Conf. Series. J. Phys.: Conf. Ser., 2018.
23. Andi, M., Abdul, K., Wayan, F., Ramdhani, F., PriyoAnggodo, Y., Dental
disease detection using hybrid fuzzylogic and evolution strategies. J.
Telecommun. Electron. Comput. Eng., 10, 1–8, 2018.
Index
ACE (area control error), 113, 114 Car-like mobile robot, 146
Acceleration, 146 CASO (Chaotic Atomic Search
Accuracy in quantification is raised, 16 Optimization) algorithm, 107,
Activation functions, 129, 132–133 108, 110, 111, 116, 118, 119
Adaptive networks: architecture, 176 Centroid points, for fuzzy numbers,
classification of images, 176 53
Adjacent matrix, 235, 241, 246, 252, Chaotic Atomic Search Optimization
258, 263 (CASO) algorithm, 107, 108,
ADNET, 125–138 110, 111, 116, 118, 119
AGC (automatic generation control), College, 162, 164
108, 116 Commodity market, 90, 91, 93
Aggregation, 149, 153 Communication via remote
Ambiguity (A), of fuzzy number, connections and a web server,
defined, 57 33–34
ranking method, 53, 54, 77, 79, 80 Comparative study, 233, 234, 247,
designed method, 56, 58, 60, 249–251, 264–265
61–62, 64, 66, 67–68 Competitive examination, 102
reasonable properties, 68–75 Compounding, 90
Analysis, 160, 162–163 Conditional attribute, 100, 182
ANFIS, 191, 194 Confusion matrix, 137–138
Area control error (ACE), 113, 114 Conventional PID, 108
Artificial intelligence, 268, 270 Convolutional neural networks
AUC, 135–138 (CNN), 125–134
Automatic braking system, 144 Core, 101, 102, 183
Automatic generation control (AGC), Cost, 164–168
108, 116 Crisp logic, 269
Autonomous bay parking, 148 Cryptology, 214
Basis, 101–104, 183, 185 Dataset, 161

Bioinformatics, 272 Decision, 100, 104
Boundary region, 101, 102, 104, 182, Decision attribute, 101–103, 182–184
183, 185 Decision making, 160–163
Brand width at various scales, 17–18 Decision theory, 2, 6
279
280 Index
Decryption algorithm, 217, 223, 230 Fuzzification, 32, 164

Defuzzification, 163–164, 168–169 Fuzziness (F), of fuzzy number,
Defuzzifier, 268 defined, 57
Dengue, 249–251, 253–254, 256, 259, ranking method, 54, 79, 80
263–264 designed method, 56, 58, 60–61,
Deoxyribonucleic acid (DNA), 127 63, 64–65, 66, 67–68
Diagnostic biomarker, 127 reasonable properties, 68–75
Difficulties to get visa, 180, 184 Fuzzy, 4, 5, 7–11
Diseases, 233, 245, 250, 264 artificial neural network, 273
DNA sequence, 272 logic, 267–273
Domination number, 214 set, 267–269, 271–273
Drug discovery, 272 system, 269–272
Fuzzy associated memory table, 153,
Efficient domination, 214–217, 219, 155
221, 223, 227–229, 231 Fuzzy cognitive map (FCM), 233–236,
Electric vehicles, 148 247, 249, 251–252
Encryption algorithm, 217–219, 228 Fuzzy control system, 144, 146,
Errors in scaling, 17 148–149
ESFL, 267 Fuzzy controller, 196
Essentiality, 164, 167–169 Fuzzy-Fuzzy set, 112
Estimation of a curve, 15–16 Fuzzy inference, 32–33
Expert system, 267, 271, 273 Fuzzy inference system, 144, 147, 148,
159
FAM table, 197, 200, 204 Fuzzy logic (FL), 42
FCM, 271 Fuzzy logic controllers (FLC),
FDSS, 271 111–114, 117–118
Feeling like an outsider, 180 Fuzzy matrix, 233, 249
Finance, 159–164, 167, 169 Fuzzy number, defined, 54–55
FL application in marketing, 43 Fuzzy output, 164, 168–169
FLC, 271 Fuzzy rule base, 167
FLC (fuzzy logic controllers), Fuzzy set, 4–6, 8
111–114, 117–118
Footprint of uncertainty (FOU), Gabor transform, 175
112 Gold fund, 90, 92, 93
FOPID (fractional order PID) Governor block, transfer function of,
controller, 107, 108, 111, 116, 109
117–118, 119–121
FPID, 191, 203, 205 Handling emergency situation, 180
FPID flow chart, 205 Hippocampus, 126, 129
Fractional order PID (FOPID) Hybrid multi-area power system,
controller, 107, 108, 111, 116, type-2 fuzzy controller for. see
117–118, 119–121 Type-2 fuzzy controller, design
Fundamental definitions, 18 of
Index 281
Hybrid two area power system, LMA (Levenberg–Marquardt

116–117 algorithm), 107, 115–116
Hydro-electric power unit, 109, 117 Load frequency control (LFC), 108,
113
Image enhancement, 174, 175 Logical thinking, 3
Image transformation, 175 Long term bond, 90, 92, 93
Indeterminate, 241, 243, 245, 247, Lower approximation, 101, 102, 104,
257, 258, 262–265 182, 183, 185
Infected, 233, 234, 249–251
Information system, 100–103, Magnitude of loss, 87
182–185 Malaria, 234–241, 243–247
Input membership function, 165 Mathematicians, 103
Interconnected power system, MATLAB, 117
108–109, 117–120 Matrix representation, 235, 241, 252,
Interval type-1 and type 2 FLCs 258
(IT1FLCs and IT2FLCs), 112, Maximum crisp value, fuzzy
117–118, 119–120 numbers’, 53
Interval type-2 fuzzy inference systems Membership functions, 164–165,
(IT2FIS), 107, 112–114 169
Intuitionistic graphs, Mosquito, 234, 235, 238–240,
degree, 215–219, 225–228, 232 243–245
membership value, 215–219, Multicriteria decision making
223–228, 230 (MCDM), 1
Investing, 90
IT1FLCs and IT2FLCs (interval NDA chip, 273
type-1 and type 2 FLCs), 112, Necessity, 164–169
117–118, 119–120 Neural networks, 146
IT2FIS (interval type-2 fuzzy Neuro-fuzzy system, 268
inference systems), 107, Neutrosophic cognitive map (NCM),
112–114 233, 240, 241, 246, 247, 251, 257,
Iteration process, 237–240, 243, 245, 259
246, 264 Nodes, 235, 240, 241, 251–252,
257–258
Jacobian matrix, 115 Normal PI controller, 108
Normal triangular fuzzy number, 55
Kaggle dataset, 134
Obstacle, 250
LabVIEW, 187, 191, 202, 204, OFF state, 239, 264
206–207 ON state, 234, 236–240, 242–246,
Levenberg–Marquardt algorithm 250, 253–257, 259–265
(LMA), 107, 115–116 Optical flow technique, 148
Likelihood of failure, 87 Ordinary fuzzy sets (OFSs), 112
Liquidity, 90 Overshoot (OS) condition, 117–118
282 Index
Pandemic crisis, 181 comparative analysis and numerical

Parameters, 233–235, 239–241, examples, 75–87
245–247, 249–252, 257–258, definitions and concepts, 54–56
264 designed method, 56–68
Percentage, 164–168 graphical representation, 78, 80, 81,
PID (proportional integral derivative) 84, 87
controller, 110, 117–118, 119–121. overview, 51–54
see also FOPID (fractional order rank order of fuzzy numbers, 76,
PID) controller 77–78, 79, 80–83, 85–86,
Planning, 159–164, 169 88–89
PMF (primary membership function), reasonable properties, 68–75
112, 114 Real easte, 90, 92, 93
Point wise calculations on a curve, 15 Reheat block, transfer function of, 109
Positioning statement, 19 Respondents, 103
Power system, type-2 fuzzy controller Risk analysis, fuzzy, 87, 90–94
for hybrid multi-area. see Type-2 commodity market, 90, 91, 93
fuzzy controller, design of decision matrix of candidates vs.
Prevention, 233, 235, 238–240, 243, decision makers opinions, 91
245, 246, 249, 261–262, 264 fuzzy numbers for five investment
Primary membership function (PMF), avenues, 93
112, 114 gold fund, 90, 92, 93
Proportional integral derivative (PID) linguistic variables for ratings and
controller, 110, 117–118, 119–121. weights, 90
see also FOPID (fractional order long term bond, 90, 92, 93
PID) controller real easte, 90, 92, 93
stock market, 90, 92, 93
Qualitative matrix, 17 Risk elimination, 19
Quality, 160, 164–165, 167–168 Risk under the level of tolerance, 19
Rough set, 181
Ranking fuzzy numbers, Rough topology, 100–104, 181–183,
application, 87, 90–94 185, 186
commodity market, 90, 91, 93 Rule base, 143–147, 153
decision matrix of candidates vs.
decision makers opinions, 91 Scilab, 144, 145, 149
fuzzy numbers for five investment Secondary membership function
avenues, 93 (SMF), 112, 114
fuzzy risk analysis, 87, 90–94 Secret key, 219
gold fund, 90, 92, 93 Self-tuning algorithm, 146
linguistic variables for ratings and Self-tuning gain controller, 108
weights, 90 Sensor for measuring,
long term bond, 90, 92, 93 humidity and temperature, 31
real easte, 90, 92, 93 soil moisture, 30–31
stock market, 90, 92, 93 Sensors, 144–147
Index 283
Settling time (ST), 117–118 Type-2 fuzzy controller, design of,

Significance of pre-processing, 174 CASO algorithm, controller
Smart cars, 143–147, 149, 151 parameters using, 107, 108, 110,
Smart grid, 148 111, 116, 118, 119
SMF (secondary membership controller design, 109–114
function), 112, 114 FLC, 111–114, 117–118
Solar photovoltaic unit, 109 FOPID controller, 107, 108, 111,
Solar power plant, 109, 117 116, 117–118, 119–121
Stages of dementia, PID controller, 110, 117–118,
mild demented, 126–127, 131–137 119–121
moderately demented, 127, 137 LMA, 115–116
nondemented, 127, 134, 137 overview, 108
very mild demented, 127, 133, 137 plant model, 108–109
Steering angle, 146, 148 result and analysis
Step response, 207 with disturbances, 119–120
Stock market, 90, 92, 93 without disturbances, 116–119
Support vector machines (SVM), 127, Type-2 fuzzy inference systems
130 (T2FIS), 107, 111–114
Synthetic minority over-sampling
technique (SMOTE), 134–135, Undershoot (US) condition,
137 117–118
System architecture, 163 Universal set, 101, 102, 104
Upper approximation, 101, 102,
T2FIS (type-2 fuzzy inference 104, 182, 183, 185
systems), 107, 111–114
The problems with the basic Vaccine, 250
quantification approach, 16–17 Vagueness (T), of fuzzy number,
Thermal power plant, 109, 117 defined, 57
TNSET, 100, 102 ranking method, 54, 80
Transmission, 233, 234 designed method, 56, 58, 61, 63,
Trapezoidal fuzzy numbers (TrFNs), 65, 66, 67–68
defined, 55, 56 reasonable properties, 68–75
graphical representation of, 78, Value (V), of fuzzy number,
80 defined, 57
ranking of, 52–53, 57–58, 59–65 ranking method, 53, 54, 77, 79,
TrFNs (trapezoidal fuzzy numbers), 80, 81, 87
defined, 55, 56 designed method, 56, 58, 59,
graphical representation of, 78, 61–62, 64, 66–68
80 reasonable properties, 68–75
ranking of, 52–53, 57–58, 59–65 Vehicle dynamics, 146
Triangular fuzzy number, normal, 55
Turbine block, transfer function of, Wind power unit, 109, 117
109 Worthiness, 167, 169
WILEY END USER LICENSE AGREEMENT
Go to www.wiley.com/go/eula to access Wiley’s ebook EULA.

Fuzzy Logic Applications in Computer Science and Mathematics

Uploaded by

Copyright:

Available Formats

Fuzzy Logic Applications in Computer Science and Mathematics

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Fuzzy Logic Applications in Computer Science and Mathematics

Uploaded by

Copyright:

Available Formats

Fuzzy Logic Applications

Advances in Learning Analytics for Intelligent Cloud-IoT Systems

Series Editor: Dr. Souvik Pal and Dr. Dac-Nhuong Le

Wiley Global Headquarters

Limit of Liability/Disclaimer of Warranty

Library of Congress Cataloging-in-Publication Data

Cover image: Pixabay.Com

Printed in the USA

2.5.3 Risk Elimination 19

4.4 Further Studies 47

7.2 Plant Model 108

9.3.2 Membership Functions 150

12 A Study on Decision Making of Difficulties Faced by Indian

13.3 Design of Fuzzy-Based PID Controller 196

15.3 Fuzzy Cognitive Map 235

The prime objective of developing this book was to provide meticulous

a) The myriad modern research information in the field of

Department of Cyber Security and Digital Forensics, National Forensic

Sciences University, Tripura Campus,Tripura, India

*Corresponding author: [email protected]

Many cognitive processes exhibit an unpredictability that defies the predic-

1.2 Fuzzy Logic

a well-defined method when it comes to solving problems. As a result, it is

1.3 Decision Making

variety of alternative courses of action, or “courses of action.” It is the pur-

1.4 Literature Review

Multi-attribute decision-making process

In both management and decision science, multiattribute decision mak-

When l is 1, 2, …, n, the X and W powers (l) are determined.

Step 5: Constructing the aggregated decision matrix

Step 10: Calculating the relative merits of several options

The primary concept was to allocate weights to decision makers based on

settlement, previously unresolved issues are made public. Flexible, elastic,

in laptop vendor selection. 2020 Fifth International Conference on Informatics

Department of LAW, Brainware University, Barasat, West Bengal, India

*Corresponding author: [email protected]

Figure 2.1 Evaluation of risk as it occurs.

Risk evaluation and assessment can be done by maintaining proper

• Coverage under insurance.

2.2 Objectives of Risk Management

2.3 Improved Risk Estimation

2.3.1 Point-Wise Calculations on a Curve

2.3.2 Estimation of a Curve

Figure 2.2 Curve used to estimate risk.

risk quantification. If necessary, we can apply more complex strategies to

2.3.3 Accuracy in Quantification is Raised

2.3.4 The Problems with the Basic Quantification Approach

that attempt to quantify Inherent Risk were widely used, but

2.4 Threat at Quantification Matrix

2.4.1 Qualitative Matrix

2.4.2 Errors in Scaling

2.4.3 Band Width at Various Scales

Insignificant Minor Moderate Major Catastrophic

Figure 2.3 Scaling and bandwidth approximation errors.

2.5 Fundamental Definitions

2.5.1 Positioning Statement

Figure 2.4 Risk under the tolerance limit.

2.5.2 Risk Under the Level of Tolerance

2.5.3 Risk Elimination