Jha 2017
Jha 2017
A R T I C L E I N F O A BS T RAC T
Keywords: The existence of sunlight, air and other resources on earth must be used in an appropriate way for human
Renewable energy welfare while still protecting the environment and its living creatures. The exploitation of sunlight and air as a
Wind energy substantial Renewable Energy (RE) source is an important research and development domain over past few
Solar energy years. The present and future overtaking in RE mainly comprises of (i) the development of novel technology for
Geothermal energy
optimum production from the available natural resources (ii) environmental awareness, and (iii) the better
Hydro energy
Ocean energy
management and distribution system. Like other domains (food, health, accommodation, safety, etc.), Artificial
Bioenergy Intelligence (AI) could assist in achieving the future goals of the RE. Statistical and biologically inspired AI
Hydrogen energy methods have been implemented in several studies to achieve common and future aims of the RE. The present
Hybrid renewable energy study summarizes the review of reviews and the state-of-the-art research outcomes related to wind energy, solar
Artificial Intelligence energy, geothermal energy, hydro energy, ocean energy, bioenergy, hydrogen energy, and hybrid energy.
Particularly, the role of single and hybrid AI approaches in research and development of the previously
mentioned sources of RE will be comprehensively reviewed.
⁎
Corresponding author.
E-mail addresses: [email protected], [email protected] (S.K. Jha).
http://dx.doi.org/10.1016/j.rser.2017.04.018
Received 23 May 2016; Received in revised form 10 April 2017; Accepted 13 April 2017
1364-0321/ © 2017 Elsevier Ltd. All rights reserved.
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
countries in the world for the year 2014, from which it is obvious that
the world is focusing additional attention on the alternative ways of
energy production [13]. The research and development in the RE have
been in full swing within the past few years. A total of 24248 research
reports was published in the literature by the numerous research
groups worldwide, focusing on the different issues related to the RE
domain in between the years 2012–2014 [18]. It results in a total
citation count of 144911 [18]. Fig. 3(a) shows the total number of
publications related to the RE sustainability and environment in top
ten scientific journals in between the years 2012–2014. The citation
counts of top ten scientific journals in the similar period of time are
represented in the Fig. 3(b). The advancement in the RE domain
specifically in the sources, county-wise application, future use, envir-
onmental effect, production methods, storage, management, distribu-
tion, allied policies and limited technical limitations, etc. are detailed in
several review reports [7–12,19–35] available in the literature. The
most prioritized research in RE domain includes: life cycle assessment
(LCA) [27,36–40], search and analysis of novel sources [41–45], social,
economic and environmental effects [46–48], effective storage and
relocation system [49–51], planning and design of grid integration and
supply systems [52–54], electrification in rural area of developing
countries [55–57], data acquisition and monitoring systems [58–61],
country and region wise assessment of development and availability
[23,35,62–70] and many more. Decision systems have been developed
for several aspects of RE such as in the evaluation of prospective, using
geographical information system (GIS) database [71], structuring of
projects [72], planning for diffusion [73], and selection of project [74].
Fig. 1. (a) Total consumption of energy in last ten years (2005–2014) in the world, and
Optimization in RE is reported in several studies, like in control
(b) top ten countries with the maximum consumption of energy in the year 2014 [13].
strategy for hydrogen storage [75], a community-based hybrid system
[76], configuration of power generating system [77,78], scheduling of
Most of the countries in the world, including the top ten listed in
micro-grid [79]. Besides that, simulation and optimization of hybrid
Fig. 1(b) were trying to include RE as a major constituent of their total
RE system, including the solar, wind, and other sources are designed
energy production. RE sources in China are showing increasing in
and evaluated [80–82], modeling for high percentage of combined heat
growth, but their preliminary predictions are not even close to being
and power production (CHF) and wind power [83], solar radiation
fully used. Republic of China RE Law and associated conventions have
modeling [84], induction generator [85] are also described.
encouraged the additional utilization of RE resources [14]. The similar
Adaptability in any field is always mandatory for additional
trend is also followed by many small developing countries like (i)
advancement with the passage of time; it is also true for the RE.
Former Yugoslav Republic of Macedonia (FYROM), where the first
Since the scope of technology is developing day by day, the application
wind power plant was completely installed and operating successfully
of the previous becomes an essential part of each of the research and
with the total capacity of around 50 MW in 2014, the projected annual
development domain currently. Specifically, the use of a machine which
production is about 125 GW/h to supply the need of 60,000 people
acts intelligently to tackle the problems is preferred in most of the
(total population of the country is about 2.1 million) [15]; (ii) Uruguay
research domains. Artificial Intelligence (AI) focuses mainly on devel-
(population 3.4 million) is producing 94.5% of its energy demands
oping intelligent machines and software for specific problems [86]. It
from renewables [16]; (iii) Costa Rica (population 4.8 million) is using
has countless applications in most of the research domains, including
maximum renewable and target 100% renewables for the power
the food, health, safety, education, business, agriculture, art, etc. [87].
production by 2021. The European Union (EU) regulations on the
AI also plays a substantial role in the advancement of RE. Importance
RE decided to achieve a target of 20% of RE production in the total
of AI in RE specifically in solar radiation and wind speed prediction,
energy consumption of EU by 2020 and 27% by 2030 [17]. Fig. 2
prediction of energy intake of a solar building and heating loads of
exhibits the share of renewables in electricity production by top ten
buildings, modeling of room heater, load and short-term electric power
forecasting, sizing photovoltaic systems, wind and solar power model-
ing and forecasting, electrical load prediction of the city and super-
markets, etc. is summarized in the studies [88–95]. Though most of the
previous reports cover the application of artificial neural network
(ANN) based approaches in RE, therefore the main focus of the present
study is to review the applications of different AI techniques including
the ANN, applied in the RE in recent few years. Precisely, the
performance of AI methods in the progress of Wind Energy, Solar
Energy and other significant sources of RE is detailed. Besides, the
impact of hybrid AI approaches in single and hybrid RE system have
been thoroughly discussed and summarized.
298
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
Fig. 3. Research reports published (a) from 2012 to 2014 in top ten journals, and (b) citation of reports published from 2012 to 2014 in top ten journals [18].
representing different renewable energy sources is shown in the Fig. 4. A noticeable in the similar duration. An overall growth rate of 9.96% is
short description of some most significant types of RE is as follows. obvious in installed wind power capacity in the year 2015 compared
with last year capacity in EU [102]. Also, the annual installation is
2.1. Wind energy increased from 48 GW in the year 2006–141 GW in the year 2015 with
an annual growth rate more than 9% [102]. Vestas V161 is the largest
The motion of earth and unbalanced incidence of the sun rays on (height 220 m and diameter 164 m) and most powerful (8 MW) wind
the surface of the earth (more on equator than the pole) causes wind turbine in the world was installed at the Danish National Test Centre,
[96,97]. The application of the wind as a significant source of energy by Denmark in 2014 [103]. A detailed overview of research and develop-
converting its kinetic energy into the mechanical energy, with the ment in different aspects of wind energy [104–111] including the
windmills and the wind turbines, is ongoing for many centuries till available resources and uses [104], policies worldwide [105], existing
now. Past evidence was found in Persia and China (200 BCE), technology [106,107], environmental impact [108], influence of cli-
Netherlands (1300–1875 CE) to the modern advancement in the mate changes [109], storage schemes [110], and monitoring and error
USA (1850–1970) [98–100]. The first wind turbine of capacity diagnosis [111] were summarized in different review reports. In the
12 kW was installed in Ohio, USA in 1887–1888 [100]. Thereafter past decade (2005–2014), a tremendous amount (80.59%) in a number
numerous wind turbines of enhanced capacity were installed in of scientific reports in the wind energy research is noticed compared to
different countries to accomplish the demand of electricity. The (1995–2004) (Fig. 6).
installed wind power capacity of top ten countries of the world in
between the years 2006–2015 is shown in Fig. 5(a), while the installed 2.2. Solar energy
capacity of the EU in the similar duration is demonstrated in the
Fig. 5(b) [101,102]. Sun is a vital source of energy for the entire living creature on the
China has a maximum number of installed wind power units earth. Solar radiation is being used for several purposes by human being
between the years 2010–2015 (Fig. 5(a)). An obvious improvement since many centuries [112–114]. The first recorded evidence is available
in the installed capacity of wind power in USA and Germany is also from 7th century B.C. when the sun ray was used to make fire after
299
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
Fig. 5. Installed wind power capacity (a) different countries, and (b) EU in between the Fig. 7. Published scientific reports of solar energy research (web of science).
years 2006–2015 [101,102].
concentrating with glass. A detailed record of the historical development solar energy is used mainly with active systems (Photovoltaic, Thermal,
of solar energy from ancient time (7th Century BCE – 1200s CE) to the etc.) and passive systems. Photovoltaic is the process of transforming the
modern era (1767–2001) available in [112]. The major breakthrough was solar energy for electricity production [115], while in the Thermal process
attained in the year of 1839 with the discovery of photovoltaic effects. The first the solar energy is transformed into some mechanical energy
300
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
Fig. 8. Published scientific reports of geothermal energy research (web of science). Fig. 10. Published scientific reports of ocean energy research (web of science).
301
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
Fig. 11. Published scientific reports of bioenergy research (web of science). Fig. 13. Published scientific report on hybrid RE research (web of science).
2.7. Hydrogen energy obtained from the web of science in between the years 1945–2014.
Maximum research reports are available for the duration of 2005–
Each of the fuel products (hydrocarbon) contains hydrogen as an 2014.
integral constituent, which is separated from the previous for the
independent applications of the latter. Besides that, electrolysis process
of water and the biological process of bacteria and algae also discharges 3. Artificial Intelligence (AI)
hydrogen, which creates high energy after burning and used as a RE
source for electric power generation [181,182]. Fuel cells are com- Artificial Intelligence tries to understand human thinking in order
monly used for the latter process. Hydrogen energy continuously to build smart entities that will perform efficiently for some compli-
supplies the demand of electricity, which is a limitation of the wind cated problems [201–203], even, though the understanding of the
and solar energy based RE systems [181–183]. The most significant complex thinking of a human brain is a tough issue to be resolved. The
research and development outcomes in the field of hydrogen energy are development in the domain of AI reduced the burden of manual
summarized in several review reports [184–190], mainly, present computation [204–206]. Only a few areas outperform the natural brain
status [184], photo-production of hydrogen [185], influencing factors performance, whilst others have already been surpassed with the
in hydrogen production [186], storage process [187], hydrogen fuel cell development of the technology like computer machines, built to do
[188], present and future strategies of hydrogen [189], technical several thousand calculations per second while this would be impos-
situation and economic part [190]. Published research reports based sible for an average human brain [206]. The AI is applied in the several
on the hydrogen energy research in between 1945 and 2014, obtained fields, including the database, accounting, information retrieval, pro-
from the web of science is shown in the Fig. 12. A gradual improvement duct design, production planning and distribution economy and
in the number of published scientific reports is obvious from 1975 to industry, medicine, food quality monitoring, biometric and forensic,
2014. etc. [201–204]. AI is based on several learning theories like statistical
learning, neural learning, evolutionary learning, etc. [201–205].
Amongst these, neural learning is most commonly used in several
2.8. Hybrid renewable energy
applications. ANN is the most fundamental technique of neural
learning. The ANN established in 1943 by McCulloch and Pitts with
A hybrid RE system combines multiple RE sources with the
the hypothesis of the mathematical model for a primitive cell of the
objective to improve the efficiency and stability of power sources than
brain (neuron). The neuron is triggered when the weighted sum of
what could be achieved using a single RE source. Some of the
input exceeds a threshold value which results in an output as a
commonly used hybrid RE sources include PV-diesel, Wind-diesel,
response of some activated function. The ANN is able to adjust its
PV-hydrogen, Wind–hydrogen, etc. (Fig. 4) [191–193]. Research and
values to fix the error from the output, which makes it more powerful
development outcomes in the field of hybrid RE are reviewed in several
learning tool [207]. Fig. 15 shows the schematic representation of a
published reports [194–200], especially, applications [194], configura-
simple ANN model based on the mathematical neuron. Some of the
tion and control [195], optimal design [196], software tools for
core types of the ANNs are the Feed-forward neural networks, Radial
integration [197], current status and future potential [198], storage
basis function neural networks (RBFNN), Kohonen self-organizing
system [199], and mathematical modeling [200]. Fig. 13 represents the
network. Besides, the neural learning, statistical and evolutionary
number of published research reports based on the hybrid RE research
learning based techniques were also used in different practical applica-
tions. Some of the statistical learning techniques in AI are Bayesian and
naïve Bayes models, clustering, hidden Markov model, nearest neigh-
bor model, etc. [208]. Also, the popular evolutionary learning methods
include genetic algorithm (GA), particle swarm optimization (PSO), ant
colony optimization (ACO), bees algorithms, etc. [209]. For the past
few years, hybrid methods of AI are also being used in many
applications with the objective to get the better accuracy than what
could be achieved using a single method. Some of the hybrid AI
methods are (i) Neuro-fuzzy (combination of ANN and fuzzy inference
system); (ii) Neuro-genetic (combination of ANN and genetic algo-
rithm, the latter is used for the connection optimization of previous);
(iii) Fuzzy-genetic (combination of fuzzy inference system and genetic
algorithm, the latter is used in the optimization of the decision
boundary of the previous) and many other kinds will be available in
Fig. 12. Published scientific reports of hydrogen energy research (web of science). future [210]. In the present study, both the single and hybrid Artificial
302
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
303
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
briefly reviewed the combined techniques for short term wind speed Monfared et al. [225]. The proposed methods perform better than the
and power estimation [213]. traditional one (RMSE 3.30 and 3.27 for two methods respectively in
The prominent research and development in the application of AI in one of the case).
wind energy contain approaches from three main categories: neural, Some statistical approaches were discussed in [226,227]. Juban
statistical and evolutionary learning and their combination as hybrid et al. [226] proposed a probabilistic method for short-term wind power
AI techniques [214–243]. Most of the research work focus on the estimation. The procedure is based on kernel density estimation and
prediction of wind speed and wind power using the neural learning results in predictive probability density function for estimation. The
approaches of AI [214–216]. Feed-forward backpropagation neural reliability of the model lies in between (2–4%), which is comparable to
network (BPNN) is used in the estimation of wind power over a period that found in similar research. The support vector machines (SVM)
of 3 years from seven wind farms by Mabel et al. [214]. The BPNN has method was used by the Mohandes et al. [227] in wind speed
a worthy prediction accuracy (root mean square error (RMSE) 0.0070 prediction of the wind data from the Madina, Saudi Arabia. Also, the
for the training data set and 0.0065 for the test data set). Three performance of SVM is compared with the multilayer perceptron
different types of ANN methods BPNN, RBFNN, and adaptive linear (MLP) neural networks. SVM achieve less estimation accuracy (MSE
element network (ADALINE)) have been used in wind speed estimation 0.009) compared with the ANN method (MSE 0.0078).
from two different sites; also, the performance of three models is The adaptive neuro fuzzy inference system (ANFIS) (a hybrid of
compared [215]. The performance of ANN methods varying according neural and fuzzy methods) has been used in some studies [228–231]
to the location of wind farms, like BPNN results in the best perfor- with the objective to further improve the performance of ANN method.
mance for one site (minimum RMSE 1.254) while for another, the best ANFIS is used by Potter et al. [228] to estimate wind power in a very
performance is achieved with the RBF method (minimum RMSE short term basis utilizing wind power data from Tasmania, Australia.
1.444). Mabel et al. [216] have optimized the configuration of BPNN MAE is always less than 8 for analysis of wind data in a different
by trial and error in estimation of wind power. Using the wind speed, session of the year. Mohandes et al. [229] have estimated the wind
relative humidity, and generation hours as the inputs, a 3×5×1 ANN speed up to a height of 100 m using the wind speed information at
model results in the best estimation performance (mean square error heights 10, 20, 30, and 40 m by using ANFIS. The ANFIS predicted
(MSE) 7.6×10−3). wind speed at the height 40 m has 3% mean absolute percentage error
The performance of ANN methods is not consistent, therefore, (MAPE) compared with the actual wind speed at the same height.
some alteration is proposed in ANN with the objective to improve its ANFIS method is used by Yang et al. [230] in interpolating the missing
efficiency [217], as well as other methods were also included for wind data measured from the twelve wind farms in China. The RMSE
comparison in some studies [218–222]. Kariniotakis et al. [217] in between the ANFIS predicted and actual measured wind speed was
implemented an advanced version of ANN (recurrent high order neural 0.230. Meharrar et al. [231] have designed maximum-power-point-
networks) for wind power estimation. The performance of ANN model tracking (MPPT) based on ANFIS wind generator. The ANFIS is used in
is compared with the naïve Bayes (NB) method. The ANN results in the estimation of the rotational speed of wind turbines using wind
minimum RMSE 4.2 compared to the NB. BPNN method was used in speed as the input. The ANFIS has effective performance (error 0.005)
spatial forecasting of wind speed in the Marmara for the years 1993– in training.
1997 [218]. The performance of ANN model is compared with the Besides ANFIS, the combination of ANN is tried with some other
Trigonometric point cumulative semivariogram (TPCSV) method. ANN methods for prediction performance improvement [232–235], for
results in a better correlation coefficient between the actual and instance, BPNN in combination with the wavelet analysis (WT) is used
predicted wind speed for most of the months and sites, for instance for the fault diagnosis of the wind turbine gearbox by Yang et al. [232],
for Canakkale site and in the month of January, the correlation which successfully detected two normal cases, two gentle fault cases,
coefficients were 0.95 for ANN and 0.88 for TPCSV. Alexiadis et al. three fault cases and one bad fault case. Evolutionary algorithms (EA)
[219] have demonstrated the significant improvement (20–40%) in the (i) particle swarm optimization (PSO) and (ii) differential evolution
estimation accuracy of the wind speed and wind power by using the (DE) have been implemented by Jursa et al. [233] for the selection of
BPNN method compared to the persistence forecasting model. Li et al. input variables and parameters of ANN and nearest neighbor models
[220] have used Bayesian combination (BC) method, and ADALINE, used in the short term wind power estimation. The PSO optimized ANN
BPNN and radial basis function neural network (RBFNN) methods in results, 2.8% improvement in prediction accuracy compared with the
wind speed forecasting from the two wind farms. The BC method manually structured ANN. Guo et al. [234] have developed an
results in consistent and better estimation result (RSME 1.5) compared improved version of the empirical mode decomposition (EMD)-feed-
with the ANN methods. A detailed comparison of twelve estimation forward neural network (FNN) method for wind speed estimation.
techniques including the linear (ARMA) methods, neural logic network Modified EMD-FNN results better performance (MSE 0.1648) than the
(NLN) non-linear ANN methods in the analysis of hourly wind speed FNN (MSE 0.1511) and EMD-FMM (MSE 0.1296). An ANN-Markov
time series data has been reported [221]. NLN exhibit the best chain (MC) method has been proposed by Pourmousavi et al. [235] for
performance (RMSE 4.9%) compared with other methods. Cadenas short term wind speed estimation. The ANN-MC has less error (94.84)
et al. [222] have used BPNN in the wind speed forecasting of data compared with the ANN (96.05) for higher margins.
obtained from wind farm Chetumal, Quintana Roo in Mexico over the Some other hybrid AI approaches were also described [236–243].
duration of two years from 2004 to 2005. The performance of ANN is Damousis et al. [236] developed Fuzzy methods using the two GA
compared with the single exponential smoothing (SES) method. The algorithms (real coded GA and binary coded GA) for wind speed and
earlier method performs better (mean absolute error (MAE) 0.5251) power estimation. The wind energy data from the remote location were
compared with SES method (MAE 0.5617). received by using the wireless modems and analyzed with the Fuzzy
In some studies [223–225], fuzzy logic [223], as well as their method which results in 29.7% and 39.8% higher accuracy for the next
combination with the ANN methods was also studied in wind power hour and longtime respectively than the persistent method. A hybrid
forecasting. Fuzzy logic was used to design a wind generation system wind-forecasting technique is developed and examined by Hu et al.
(3.5 kW) by Simoes et al. [223]. The developed system performs [237] by combining ensemble empirical mode decomposition (EEMD)
satisfactorily and has field application capability. Sideratos et al. and SVM methods. Average monthly wind speed from three different
[224] implemented the combination of ANN, RBFNN, and fuzzy logic sites in China was estimated using the proposed hybrid method. EEMD
techniques for estimation of wind power. The analysis outcomes are has MAE 0.12 compared with two traditional time series methods:
effective in the operational planning of wind farm 1–48 h ahead. The autoregressive integrated moving average (ARIMA) and seasonal
BPNN and fuzzy methods have been used in wind speed estimation by autoregressive integrated moving average (SARIMA), SVM, and
304
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
EMD-SVM. A different hybrid model using ARIMA and BPNN methods 4.2. AI in solar energy
was developed and used in wind speed forecasting for three different
locations in Mexico by Cadenas et al. [238]. The hybrid method has Importance of AI in solar energy applications is summarized in
MSE 0.49 compared with the ANN (MSE 5.65) and ARIMA (MSE 4.1). reviews [89,90,92,94,244,245]. Particularly, the detail applications of
Salcedo-Sanz et al. [239] have proposed hybridization of the 5th ANN methods in modeling and design, heating load of the building, etc.
generation mesoscale model (MM5) with the ANN method for short is summarized in [89]. Mellita et al. [90] have briefed detail research
term wind speed forecasts for the thirty-three wind turbine data sets. based on the application of AI in modeling of weather data, and sizing,
The output of MM5 is used in the ANN method which results in better modeling, simulation and control of PV systems. Mellita et al. [92] have
estimation accuracy with the MAE in between 1.45 and 2.2 m/s for the also reviewed the uses of AI techniques in the sizing of individual and
different number of neurons (9−15) in the hidden layer and locations grid-connected PV systems. Building energy consumption estimation
of the wind turbine. Liu et al. [240] have developed a hybrid AI method using statistical and AI methods has been compiled in [94]. Dounis
by using deep quantitative analysis, WT, GA and SVM methods. GA is et al. [244] summarized the application of agent-based intelligent
used in tuning the parameters of SVM. The WT-SVM-GA model control systems for energy management of buildings. AI techniques in
achieved better performance (MAE 0.6169) compared with the persis- modeling and forecasting of solar radiation data are discussed in [245].
tent method (MAE 0.8356) and SVM-GA (0.7843). A novel hybrid The research in solar energy contains the application of both the
model for wind speed prediction was developed by Kong et al. [241] by single and hybrid approaches of AI [246–285]. ANN is the most used
using the improved version of support vector regression (SVR) referred method in solar energy research [246–258]. ANN is used in solar
as reduced support vector machine (RSVM), principal component irradiance prediction for PV connected with the grid [246]. A correla-
analysis (PCA), and particle swarm optimization (PSO) for parameter tion of 98–99% for sunny days and 94–96% for cloudy days between
optimization of RSVM. The RSVM exhibits effective estimation accu- the actual and predicted solar irradiance is achieved.
racy. Rahmani et al. [242] developed a hybrid intelligent technique Global solar radiation (GSR) is forecasted with the BPNN using
based on the combination of two meta-heuristic techniques: ant colony temperature and humidity as inputs over the years (1998–2002) [247].
optimization (ACO) and particle swarm optimization (PSO) for hourly The RMSE value was 2.823×10−4 in between the actual and BPNN
wind power estimation of forty-three wind turbine data sets using wind predicted GSR for the year 2002. BPNN is used in performance
speed and temperature inputs. The hybrid method performs best estimation of a solar water heating system by Kalogirou et al. [248].
MAPE 3.5% compared with ACO (MAPE 5.8%) and PSO (MAPE The higher values of coefficient of determination (R2 0.9914 and
10.5%). Pousinho et al. [243] have proposed a hybrid method by using 0.9808 for extracted energy and the maximum temperature rise
WT, PSO and ANFIS for risk optimization in wind energy trading. This respectively) confirm the better performance of BPNN. Beam solar
hybrid approach is applied for wind farm data analysis in Portugal. The radiation was estimated using the BPNN by analyzing the data from
expected profit was estimated successfully in between 18719 € and eleven different stations. The RMSE between the actual and model
18487€ for different values of risk level in between 0 and 1. A complete predicted values of radiation lies in between 1.65 and 2.79% [249]. A
summary of describing research work in [214–243] is presented in BPNN of 3×6×1 is used in daily ambient temperature estimation with
Table 1. RMSE 1.96 [250]. Daily solar irradiation was estimated using the
BPNN with RMSE (5.0–7.5%) [251]. The maximum power of high
concentrator photovoltaic (HCPV) system was predicted using the
Table 1
Summary of reports for application of AI approaches in wind energy [214–243].
305
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
BPNN with RMSE 3.29% [252]. Monthly average daily global solar classical models, BPNN, RBFNN and MLP+IIR methods. The sizing
irradiation was estimated using the BPNN with the 0.97 correlation coefficients were estimated accurately (correlation 98%) with the RBF
between the actual and predicted solar irradiation [253]. Solar energy +IIR method. A combination of WT and BPNN was used in solar
output, and hot water quantity were estimated using the BPNN with R2 radiation values estimation [271]. The performance of WT+BPNN
0.9978 and 0.9973 respectively [254]. Solar radiation was estimated (accuracy 97%) was observed better than the classical methods (AR,
using the BPNN in Nigeria with R2 0.971 using latitude, longitude, ARMA, MTM), BPNN, recurrent and RBFNN methods. Solar power
altitude, month, mean temperature, mean sunlight duration, and output is predicted by using GA optimized BPNN without using the
relative humidity as input variables [255]. BPNN is used to estimate exogenous inputs [272]. The performance of GA+BPNN is compared
the energy intake of a passive solar building (wall thickness (15–60 cm) with the persistent model, ARIMA, k-nearest neighbor (KNN) and
with R2 0.9991) [256]. In another study, BPNN results in 94.8–98.5% BPNN methods. The GA+BPNN results in the minimum RMSE
prediction rate in the building energy consumption prediction for the 72.86 kW. Mandal et al. [273] have used the combination of WT and
insulation thickness of 0–2.5–5–10–15 cm, orientation angles 0–80° RBFNN in the prediction of PV system power and compared its
and the transparency ratios 15–20–25% [257]. performance with the WT+BPNN, RBF, and BPNN. The WT+RBF
In some studies [258–261], the performance of BPNN model is have minimum RMSE 0.23. Group method of data handling (GMDH)-
compared with the other methods. BPNN is used by Tasadduq et al. NN and GA are used in optimization of the economic benefits of solar
[258] in the estimation of ambient temperature 24 h ahead and the energy [274]. The optimal solution results in 3.1–4.9% increment in
performance of BPNN is compared with the batch learning ANN. The life cycle savings.
achieved values of mean percentage deviation (MPD) were 3.16, 4.17 ANFIS method is used in several studies [275–280] like in the
and 2.13 with BPNN for three years. Diffuse solar radiation is predicted modeling of PV power supply system with accuracy 98% [275],
by Alam et al. [259] using the BPNN on an hourly and daily basis with prediction of hourly global radiation using the satellite image data
RMSE of 4.5% compared to other empirical methods (EKD, Page, etc.) [276], clearness index and daily solar radiation prediction with RMSE
(RMSE 37.4%). Tymvios et al. [260] have used BPNN and Ångström's 0.0215–0.0235 [277], modeling of PS power supply [278], predicting
linear methods in global solar radiation prediction. The performance of solar radiation using the mean temperature and sunshine duration
BPNN method is comparable (RMSE 5.67–6.57%) to Ångström's linear [279], performance prediction of solar chimney power plant (SCPP)
method. BPNN method is used in global solar radiation estimation of [280].
the eight cities of China over the years 1995–2004, and the perfor- Several hybrid AI techniques were also used in solar energy systems
mance is compared with the empirical regression methods. The BPNN [281–285], like hybrid evolutionary optimization of ANN using the
performs better than the empirical regression methods with minimum PSO and GA in the estimation of PV power [281]; Genetic swarm
RMSE 0.867 [261]. Besides ANN, some other techniques were also optimization (GSO) of BPNN for PV system energy estimation [282];
implemented in solar energy analysis [262–265]. For instance, SVM solar radiation prediction using the combination of ARMA and time
method is used in the prediction of short-term solar power and its delay neural network (TDNN) [283]; power prediction of PV connected
performance is compared with the autoregressive (AR) and RBFNN to the grid using the hybrid of seasonal auto-regressive integrated
[262]. SVM method (MAE 33.7 W/m2) performs better than the RBF moving average (SARIMA) and SVM methods [284]; hybrid of SVM
(MAE 43 W/m2) and AR (MAE 62 W/m2) methods. Li et al. [263] have and Firefly algorithm (FFA) is developed for GSR estimation and
used SVR for solar PV energy production estimation and compared its performance is compared with the BPNN and genetic programming
performance with the ANN. The RMSE for the two methods were (GP) methods (RMSE 1.8661 for SVM-FFA). The findings of AI
almost similar. The performance of the RBF-SVM method is compared techniques for solar energy systems are summarized in Table 2
with the existing forecast methods (PPF and Cloudy) in the estimation [246–285].
of solar power generation. SVM exhibits 27% higher estimation
accuracy than other two methods [264]. 4.3. AI in geothermal energy
Some evolutionary AI methods were also used in solar energy
applications [265–267]. Mashohor et al. [265] suggested, GA in solar AI approaches have been used in geothermal applications, which
tracking for improved performance of PV systems. The GA with initial are summarized in reviews [286–290]. Particularly, the prospective of
population size 100, 50 epochs and probability of crossover and AI approaches with sensors and robots in geothermal well drilling
mutation 0.7 and 0.001 respectively results in the best GA-Solar design, control, and optimization is briefed in [286]. Computer
system. The low value of standard deviation (1.55) in generation gain simulation and modeling of geothermal reservoir and its effect of
also proves the better efficiency of the system. GA is used in the optimal geothermal energy progress are reviewed in [287]. Similarly, in other
design of a solar water heating system. Specifically, the plate collector reviews by Sanyal et al. [288] numerical simulations for enhanced
area is optimized with the GA to 63 m2 that results in solar fraction geothermal systems and for geothermal reservoir by O'Sullivan et al.
value 98% [266]. Kumar et al. [267] have used GA in maximum power [289] are reviewed. In another study, a brief history of numerical
point tracking (MPPT) of PV array connected to the battery. The modeling of geothermal reservoir is also presented [290].
performance of the GA is compared with the traditional perturb and Both the single and hybrid approaches of AI are used in geothermal
observe (PO) algorithm. The boost converter achieves the line voltage energy applications [291–310] summarized in Table 3, though the
of 400 V. ANN method is used in most of the studies [291–303]. Esen et al. [291]
The combination of AI methods was also reported to improve the have used BPNN (with Levenberg–Marguardt (LM), Pola–Ribiere
prediction efficiency [268–274]. The performance of an integrated conjugate gradient (CGP), and scaled conjugate gradient (SCG) algo-
collector storage (ICS) solar water heater is predicted using the rithms) in performance prediction of vertical ground coupled heat
combination of ANN and TRNSYS with the R2 value 0.9392 [268]. pump (VGCHP) system. The LM based BPNN with eight neurons in the
Monteiro et al. [269] have used GA in parameter optimization of hidden layer results in better prediction efficiency (RMS 0.0432).
HIstorical SImilar Mining (HISIMI) model for power prediction of PV Bassam et al. [292] have used LM based BPNN for the static formation
system. The performance of GA+HISIMI model (RMSE 283.89) is temperature (SFT) prediction of the geothermal well. The BPNN with
compared with the BPNN (RMSE 286.11), and classical persistence five neurons in the hidden layer results in prediction error < ± 5%.
(RMSE 445.48) methods. The combination of RBFNN and infinite BPNN (with LM, CGP, and SCG) is used in the determination of an
impulse response (IIR) filter is used for size optimization of PV system optimum working condition of geothermal well [293]. The BPNN with
in the Algeria [270]. Optimal sizing coefficients were determined using seven neurons in the hidden layer results in the best predicted values of
the RBF+IIR method and its performance is compared with the generated and circulation pump power, using the vapor fraction of
306
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
Table 2
Summary of reports for application of AI approaches in solar energy [246–285].
geothermal water and its temperature, and the ammonia fraction as the generating and required pump circulation power. The input variable of
input (RMSE 1.5289). ANN is used in the optimization of the power the cycle s1 is similar to that described in [293] though for the cycle s2
cycle like ORC-Binary using the BPNN (with LM, CGP, and SCG) [294]. an additional input variable outlet pressure is included in the analysis.
The LM based BPNN with 14–16 neurons in the hidden layer result in BPNN is used in the generation of geothermal map at different depth
best accuracy (RMSE 0.0001 for s1 and s2 cycles) for prediction of with less than 5% deviation with the actual values for the 96.5% data
Table 3
Summary of reports for application of AI approaches in geothermal energy [291–310].
307
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
points [295]. The LM based BPNN is used in the prediction of thermal two different seasons for seventeen years [315]. The steam flow is most
performance and exergy destructions of the Afyonkarahisar geothermal significant factor in the hydroelectric power production. Kisi O [316]
district heating system (AGDHS) with good accuracy (RMSE 0.0053) has also studied the river flow modeling using the BPNN with gradient
[296]. Void fraction (VF) values of the geothermal well were predicted descent (GD) and the performance is compared with the autoregressive
with the BPNN based on the LM training algorithm using the eight (AR) method. BPNN estimates more precisely than the AR method.
different input parameters. Six neurons in the hidden layer of BPNN Estoperez et al. [317] have used BPNN in scheduling of micro-hydro
result in best prediction accuracy (RMSE 0.0966) [297]. The BPNN power plant by estimating the power discharge for one month ahead
(with LM, Quasi-Newton (QN), and Bayesian Regularization (BR) (minimum RMSE 0.061). GA and [318–320] Fuzzy [321] approaches
algorithms) is used to predict the biochemical oxygen demand have been also used in the hydro energy study; like Carneiro et al. [318]
(BOD), ammonia-nitrogen, nitrate-nitrogen, and ortho-phosphatepho- have used GA in the scheduling of hydrothermal power system in Brazil
sphorus of geothermal energy treating storm water. Best accuracy is and compared with the outcomes from traditional non-linear program-
obtained for Ammonia-nitrogen prediction with the QN based BPNN ing (NP) optimization method. The GA has less operating cost
[298]. BPNN is used to test the effectiveness of PID controller of (726,742.2 MW) than NP (745,020 MW) for the years 1971–1973.
AGDHS which enhances the energy efficiency by 13% [299]. In Gil et al. [319] have implemented a new GA (with a set of proficient
modeling of ORC-Binary geothermal plant, BPNN with LM (twenty operators) for a similar application and compared the performance
neurons in the hidden layer) for the o2 and o3 cycles and (twenty-two with previously used GA. Yuan et al. [320] have developed a novel
neurons in the hidden layer) for b3 type cycle results in better accuracy version of GA (chaotic hybrid (CH)-GA) to solve the issue of the
[300]. BPNN based on the LM and SCG algorithms is used for the site existence of water delay time as a constraint in the short term
location planning model using the geographical information data hydrogenation scheduling. The CHGA results in a better profit com-
[301]. BPNN is used in the conductivity map creation of ground with pared with the standard (S)-GA and NP. The application of fuzzy logic
better accuracy (83% of predicted data have deviation less than 10%) based approach in the selection of optimal penstock material from
[302]. BPNN based on the LM algorithm exhibits better prediction Steel, Asbestos cement and GRP for hydro turbine is addressed by
efficiency of pressure drop in the geothermal well by using the wellbore Adhikary et al. [321]. The GRP was declared as the ideal material with
production database [303]. a maximum degree of index.
In some studies, EA and fuzzy logic were also used in the The contribution of ANFIS and hybrid AI approaches in hydro
geothermal system analysis [304–307] like Sayyaadi et al. [304] have energy generation has also been discussed in some studies [322–327].
used the single objective-thermodynamic and thermoeconomic (TE) The ANFIS method is used in control of Shihmen reservoir in Taiwan
and multi-objective optimizations of vertical ground source heat pump (in the prediction of water release); also the performance is compared
(VGSHP) using the (EA), in another study six EAs (two DE, PSO, GA, with the M-5 rule curves [322]. The ANFIS exhibits better performance
Monte-Carlo random search, etc.) were used to locate the optimal (less water shortage) than the M-5 rule curves. The ANFIS model is
position of borehole heat exchangers (BHEs) [305]; a fuzzy logic used efficiently in flow estimation of the Menderes River in Turkey by
controller (FLC) system has been designed for geothermal heat in the Firat et al. [323]. The performance of ANFIS is compared with the ANN
recirculation aquaculture systems (RAS) [306] and to control the water and multiple regressions (MR) (minimum relative error 0.073 for
temperature for the maximum RAS production [307]. ANFIS). The integration of ANN with the expert system is used in
ANFIS and hybrid AI approaches were also implemented in a few acoustic prediction (AP) and predictive maintenance (PM) of hydro-
studies of geothermal energy analysis [308–310] like ANFIS is used for power plant by using the Learning Vector Quantization (LVQ) and
the VGSHP performance evaluation and compared with the BPNN ART-MAP respectively [324]. More accurate predictions are obtained
methods (LM, SCG, CGP algorithms), in which ANFIS results in better by the AP and PM. Sinha et al. [325] have developed GA and PSO tuned
efficiency than the BPNN methods [308]; ANFIS is used in the FLC for the automatic generation control (AGC) in hydropower system.
evaluation of the AGDHS system (exergy and energy rates prediction) The GA-FLC and PSO-FLC perform better (less peak overshoot, and
and performance is compared with the BPNN methods (LM, SCG, CGP settling time) than the FLC.
algorithms) [309]. Again, ANFIS performs better than the BPNN A hybrid AI approach (referred as case-based reasoning (CBR))
methods; GA and singular value decomposition (SVD) based GMDH- using the hierarchical clustering (HC), Fourier frequency transform
NN is used in geothermal reservoir temperature prediction [310]. (FFT), Elman ANN and Modular ANN have been developed for the
river flow estimation [326]. The performance of CBR is compared with
4.4. AI in hydro energy the BPNN, Elman ANN, RBFNN, etc. (minimum MAE 17. 11 for CRB).
BPNN in combination with the artificial bee colony (ABC) algorithm
Application of AI approaches in hydro energy domain is summar- (particularly the BPNN is trained with ABC) is used to predict the
ized in reviews [311,312]. Particularly, the design and control of hydraulic energy production in Turkey (relative error (RE) 0.23 [327].
hydropower plants using traditional methods and modern AI ap-
proaches like GA, ANN, Fuzzy, ANFIS, etc. has been briefly presented 4.5. AI in ocean energy
by Kishor et al. [311]. In another review study by Nourani et al. [312]
described the significance and application of wavelet pre-processor The use of AI approaches in ocean energy is summarized in the
based hybrid AI approaches in hydro-climatology, specifically in the reviews [328–331]. Mainly, the role of AI in investigating the sea for
estimation of significant hydrologic cycle processes. the development of power supply system is discussed in [328]; the
The application of single and hybrid AI approaches in hydro energy impact on AI in the ocean is briefed by Aartrijk et al. [329]. Several
applications [313–327] is summarized in Table 4. BPNN approach is applications of ANN in the ocean engineering are presented by the Jain
used in optimal scheduling the activities of hydropower plants from ten et al. [330]. Iglesias et al. [331] have discussed in detail about the
reservoirs in Taiwan [313]. The BPNN is more cost effective than the k- availability of the renewable energy resources, especially the potential
nearest neighbor (KNN) and differential dynamic programming (DDP). of ocean energy wave farm in Canary Islands (which will be a first
Smith et al. [314] have implemented BPNN method in the modeling of Island in the future having 100% renewable energy).
rainfall-runoff process to estimate the discharge peak and time of peak The role of some single and hybrid AI approaches in ocean energy
of linear and non-linear reservoirs. Better accuracy of BPNN is were described in the studies [332–341] and main outcomes are
achieved for non-linear reservoirs in the prediction of peak discharge summarized in Table 5. A three layer BPNN method is used in the
and linear reservoir in the prediction of time to peak. BPNN model is estimation of sea level variation on the coast of Western Australia
used effectively in the steam flow prediction of San Juan River basin in (correlation coefficient 0.7–0.9) [332]. One day forecast of ocean wave
308
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
Table 4
Summary of reports for application of AI approaches in hydro energy [313–327].
[313] BPNN, DDP, KNN Scheduling of hydropower plant BPNN 0.011% cost effective
[314] BPNN Modeling of rainfall-runoff process RMSE 0.097–0.260
[315] BPNN (LM) Stream flow prediction MAPE < 5%
[316] BPNN (GD), AR River flow prediction Error 0.2% with BPNN
[317] BPNN Power discharge estimation MAPE < 5%
[318] GA, NP Scheduling of hydropower plant Less active cost 2.9% by GA
[319] GA Scheduling of hydropower plant Population cost 2.05$
[320] CHGA, SGA, NP Hydrogenation scheduling 190,301$ profit with CHGA
[321] Fuzzy logic Selection of optimal material GRP, degree of index 2.07
[322] ANFIS, GA, M-5 Water release prediction Water shortage 0.0 for ANFIS
[323] ANFIS, ANN, MR River flow prediction RMSE 7.1 for ANFIS
[324] Hybrid method (LVQ-ART-MAP) Acoustic and maintenance prediction False alarm rate < 10%
[325] Hybrid method (FLC-PSO, FLC-GA) FLC design for AGC Scaling factor for hydro area 4.731 with FLC-PSO
[326] Hybrid method (HC-FFT-ANN) River flow prediction Std. 26. 48
[327] Hybrid method (ANN-ABC) Hydraulic energy prediction MAPE 4.6%
Table 5 ocean energy. Chen et al. [336] have developed a FLC to reduce the
Summary of reports for application of AI approaches in ocean energy [332–340]. effect of the external ocean wave force. The FLC exhibits good stability.
Sea level is predicted using the GP and ANN by Ghorbani et al. [337].
Ref. no. Method/methods Application Outcome
The GP prediction accuracy was better than the BPNN based on LM
[332] BPNN Sea level variation RMSE 10% of tidal algorithm (MSE 230.5–236.2).
prediction range ANFIS [338] and hybrid AI approaches [339,340] have been
[333] BPNN Sea wave height 84% for a lead time of implemented to achieve better prediction accuracy. Karimi et al.
prediction 6h
[338] have used ANFIS (five types with different membership func-
[334] BPNN Wave parameters RMSE 0.53
prediction tions) in sea level forecasting and compared the performance with the
[335] BPNN, RBFNN, Dispersion coefficient RMSE 27.9 BPNN BPNN (LM), BPNN (CG), BPNN (GD) and eleven types of ARMA
GRNN prediction models. ANFIS and ANN methods result, almost similar but better
[336] Fuzzy logic Reducing effect of Stability 0.0 for small
than the ARMA models. A hybrid approach using the combination
ocean wave amplitudes
[337] GP, BPNN (LM) Sea level prediction R2 0.972–0.973 numerical wave model (NWM) and BPNN is used for wave hindcasting
[338] ANFIS, BPNN, Sea level prediction RMSE 0.055for ANFIS [339]. The hybrid approach performs better than the BPNN and NWM
ARMA method. De-Paz et al. [340] have developed a hybrid intelligent system
[339] Hybrid method Wave hindcasting Correlation 0.93 based on case-based reasoning (CVR) and support vector regression
(NWM-BPNN)
(SVR) for improved prediction of CO2 flux to explore the understanding
[340] Hybrid method CO2 flux prediction Mean accuracy 96.3%
(CVR-SVR) of interaction between the air and ocean.
4.6. AI in bioenergy
condition was done by the Londhe et al. [333] using the BPNN method
(six different architectures for the number of neurons in the hidden A brief review of deterministic and stochastic mathematical model-
layer) with good accuracy (67% correlation for the predicted wave ing for optimization of forest biomass (specifically the optimum design
height for lead times of 12 h). Three different architects used the BPNN of the supply chain) in RE generation is presented by Shabani et al.
method in the prediction of wave parameters using the coastal [341]. The use of single and hybrid AI approaches for bioenergy
environment variables as input by analyzing the data collected from analysis is described in several research reports [342–354] and
Tasmania during 1985–1993 (R2 0.92) [334]. Toprak et al. [335] have summarized in Table 6. ANN is applied in several studies [342–347]
used BPNN, RBFNN and generalized regression neural network related with the bioenergy: like forecasting the cetane number (CN)
(GRNN) to forecast the longitudinal dispersion coefficient in streams and density of diesel fuel using the GRNN by Yang et al. [342];
for 65 data sets from 30 rivers in the USA (MSE 13275 for BPNN). detection of trace compounds like H2S and NH3 up to 93 ppm (ppm) in
Fuzzy [336] and GP [337] methods have also been used in the study of biogas using the BPNN (RMSE 416, 5.1 ppm, respectively) [343];
Table 6
Summary of reports for application of AI approaches in bioenergy [342–354].
309
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
detection of CN in biodiesel using the BPNN, RBFNN, GRNN and brane fuel cell (PEMFC) with high accuracy [361]; BPNN with the
recurrent neural network (RNN) using the fatty acid composition (the twelve different training algorithms were implemented for the predic-
best performance is achieved with BPNN) [344]; estimation of methane tion of hydrogen engine characteristics (mass air flow (MAF), air
concentration in the biomass from bioreactors using alkalinity, BOD, pressure, fuel pulse width, exhaust gas and engine temperature, and
chloride, conductivity, pH, sulfate, and temperature as input parameter NOx emission) using two inputs engine speed and throttle position
for ten types of BPNN (according to different training algorithms) [362]. BPNN is also implemented in another studies [363–366] to
(RMSE 0.00263-0.00250) [345]; estimation of biodiesel properties predict the hydrogen engine parameter and emissions [363] (RMSE ±
(density, viscosity and water and methanol content) using the multiple 4%); for the tensile strength prediction of hydrogen-functionalized
linear regression (MLR), principal component regression (PCR), poly- graphene [364]; to predict the stack voltage of the solid oxide fuel cell
nomial and Spline partial least squares regression (PLS), BPNN (SOFC) [365]; and in the power density prediction of MFC (RMSE
methods and their performance comparison (the best performance is 4.89×10−4 for one configuration) [366].
achieved with the BPNN compared with the rest methods) [346]; Fuzzy logic methods [367–369] and EU approaches [370–372]
performance estimation of biodiesel engine (thermal efficiency and have also been used in hydrogen energy analysis: like Fuzzy logic
energy consumption of break, exhaust temperature, and engine emis- method is used in prediction of ignition time of hydrogen car using
sions) using the load, compression ratio, blend, injection timing, three different types of membership functions [367]; recurrent fuzzy
pressure as inputs of RBFNN (accuracy 69–96%) [347]. system is used to model the current density characteristics of SOFC
Besides ANN, some other methods, including SVM and KNN [348], [368]; Fuzzy logic controller based on parameter optimization with the
PSO [349] and GP [350] methods have been also implemented in GA is used to manage the hydrogen consumption in fuel cell hybrid
bioenergy analysis. Balabin et al. [348] have implemented regularized vehicles (FCHV) [369]. Besides the fuzzy logic and GA, PSO is also used
discriminant analysis (RDA), PLS, KNN and SVM methods to classify in the energy optimization of FCHV [370]. BPNN, GA and PCA in
the biodiesel into ten different classes (according to their origin) using hydrogen production modeling is reviewed by Nath et al. [371].
the near-infrared (NIR) data. SM results in better classification Askarzadeh et al. have proposed the bird mating optimization (BMO)
accuracy than the rest three methods. A modified version of PSO is approach to model the PEMFC system [372].
implemented in the optimization of biomass supply chain (flows from Application of ANFIS [373–377] and other hybrid AI approaches
sources of production) [349]. GP and BPNN have used the higher [378–382] were described in many studies [378–382]: ANFIS is used
heating value (HHV) estimation of biomass fuels and performance is to predict the SOFC parameters (stack current and voltage) and the
compared with the existing HHV models [350]. GP and BPNN exhibit performance is compared with the ANN method (RMSE < 2 for ANFIS
better prediction accuracy than the conventional models (RMSE 0.942- in current prediction) [373]; ANFIS is used in prediction of several
0.987). hydrogen safety parameters (like explosive limit, hydrogen pressure,
Some research reports the application of hybrid AI approaches in and flow rate) using the ten input conditions [374], performance of
the bioenergy analysis [351–354]. Koutroumanidis et al. [351] have ANFIS is compared with the eleven types of BPNN based on different
used ARIMA, ANN and hybrid of ANN-ARIMA for estimation of training algorithms (RMS 1.4 in hydrogen pressure prediction with
fuelwood prices in Greece for the years 1964–2006. The ANN- ANFIS); ANFIS and BPNN (LM) were implemented for emissions (HC,
ARIMA model predicts better estimation than the ANN and ARIMA CO, CO2, NOx) prediction from the hydrogen car, BPNN shows better
methods independently (MAPE 14%). A hybrid system based on the prediction than the ANFIS (RMSE 1.58% of HC emission with the
combination of Fuzzy logic and ANN is used for improving the biomass BPNN) [375]; ANFIS is used in the performance (H2 flow rate, system
boiler cleaning and maximizing heat transfer which saves 12 GW h/ and stack efficiencies) prediction of PEM electrolyzer (1.06% prediction
year [352]. BPNN and GA based hybrid AI method is developed for the error for hydrogen flow rate) [376]; ANFIS used in prediction of cell
methane production from the waste digester [353]. The hybrid method voltage of PEMFC efficiently [377], and performance is compared with
with optimized parameters results in 6.9% increment in methane RBFNN, and BPNN; a hybrid AI approach based on wavelet and fully
production. In another study, a similar hybrid method is used for the logic method is implemented for energy controlling of HEV (fuel
optimization of biogas production (from the banana stem, cow dung, consumption of 0.06962 kMol H2) [378]; SVR and PSO based hybrid
paper waste, rice bran, saw dust) [354] which result in the biogas approach is used in temperature forecasting oh hydrogen reactor with
production of 10.280L. high accuracy and performance is compared with the SVR and BPNN
[379]; BPNN in combination with the GA is used in the biohydrogen
4.7. AI in hydrogen energy yield optimization (54 ml/g improvement with proposed approach)
[380]; in another study [381] similar combination of methods is used
Petrone et al. [355] have presented briefly, a review of model based to optimize the cell parameters of SOFC (standard error of prediction
AI approaches for the diagnosis of proton exchange membrane fuel cell 1.705%); a hybrid ABC algorithm is used in the parameter prediction of
systems (PEMFCs). Similarly, in another study, three categories of non- PEMFC and performance is compared with the PSO and GA, hybrid
model based approaches, including AI, statistical, and signal proces- ABC performs better than the other methods with the minimum sum of
sing methods for a similar problem is detailed in [356]. squared error (SSE) [382].
The research application of AI approaches is described in several
studies [357–382], summarized in Table 7. The ANN is the widely 4.8. AI in hybrid renewable energy
implemented method in the hydrogen energy [357–366] like three AI
approaches, including the BPNN, SVR and multi-gene genetic pro- Applications of AI approaches in the hybrid RE were briefly
gramming (MGGP) is used in the prediction of output voltage of described in the reviews [383–385]. The development of approaches
microbial fuel cell (MFC) in which MGGP results in the best accuracy for the optimal sizing is briefly presented by Luna-Rubio et al. [383].
[357]; BPNN is used to predict CO2 hydrogenation activity [358]; Specifically, the design methodologies of solar-wind hybrid RE system
BPNN with eleven training algorithm is used to predict the effect of are presented by Zhau et al. [384]; application of different EA
hydrogen car engine operating conditions on the emission of CO2, CO, approaches in optimization is summarized in [385].
NOx, and hydrocarbons [359] (CO emission is predicted with 100% Few single and hybrid AI approaches in hybrid RE applications
accuracy); BPNN trained with LM and Bayesian algorithm is used for [386–397] are summarized in Table 8. BPNN is used in the power use
monitoring the stability and detection of error in the PEM fuel cell and generator status (on/off) prediction for a water power supply based
[360]; BPNN based on LM training algorithm is used to predict the hybrid RE system (prediction accuracy 97%) [386]. Chavez-Ramirez
voltage and cathode temperature of the polymeric electrolyte mem- et al. [387] implemented BPNN method for power prediction of hybrid
310
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
Table 7
Summary of reports for application of AI approaches in geothermal energy [357–382].
[357] BPNN, SVR, MGGP MFC output voltage prediction R2 0.9872 for the MGGP
[358] BPNN CO2 hydrogenation activity 32% conversation rate for Mn-catalyst
[359] BPNN Prediction of emissions from hydrogen car RMSE 0.0002 for the CO
[360] BPNN (LM, and Bayesian algorithm) Error detection in PEM fuel cell Error 0.0
[361] BPNN (LM) Voltage and cathode temperature prediction of (PEMFC) Correlation 0.973–0.983
[362] BPNN (LM) Hydrogen engine characteristics prediction RMSE 0.4106 for MAF
[363] BPNN (LM) Hydrogen engine parameter and emissions prediction R2 0.99
[364] BPNN Tensile strength prediction of graphene R2 0.9867
[365] BPNN Stack voltage prediction of fuel cell MSE < 0.08
[366] BPNN Power density prediction R2 0.99
[367] Fuzzy logic Ignition time prediction of hydrogen engine RMSE ± 5%
[368] Fuzzy logic SOFC current density prediction RMSE 0.4
[369] Fuzzy logic with GA Hydrogen consumption management 37.9–43.8% improvement
[370] PSO FCHV energy management Optimal path for 50% swarm
[371] BPNN with GA, PCA Hydrogen production modeling Maximum production 6897 ml H2/L
[372] BMO Modeling of PEMFC system Std. 3.24×10−4
[373] ANFIS Stack current and voltage prediction of SOFC MRE < 0.25% with ANFIS for current
[374] ANFIS, BPNN Hydrogen safety parameter prediction RMSE 0.6 for explosion limit
[375] ANFIS, BPNN Prediction of emission from hydrogen car RMSE 5.51%, 2.29%, for CO2 respectively
[376] ANFIS PEM electrolyzer performance prediction Error 0.95%, 0.7% for stack and system efficiency
[377] ANFIS, RBFNN, BPNN Cell voltage prediction of PEMFC R2 0.99 for ANFIS
[378] Hybrid method (Fuzzylogic +WT) HEV energy management Fuel saving (∼8%)
[379] Hybrid method (SVR+PSO) H2 reactor temperature prediction 4.89% error
[380] Hybrid method (BPNN+GA) Biohydrogen yield optimization MSE of 9.1×10-8
[381] Hybrid method (BPNN+GA) SOFC cell parameter optimization RMSE 0.0108
[382] Hybrid method (PSO) PEMFC parameter prediction SSE 15.66
RE systems and FLC for energy management. In another study, FLC hybrid PV-wind-battery system with the objective to reduce the
and cuckoo search (CS) algorithm and PSO were used in energy production cost; also, the performance is compared with the hybrid
management of hybrid RE system (levelized energy cost (LEC) 2.01 $ optimization model for electric renewables (HOMER) and hybrid
with the CS) [388]. PSO is used in size optimization of hybrid RE optimization (HO)-GA (ANFIS achieve better performance) [396].
system by Hakimi et al. [389] with the objective to make it more cost ANN and fuzzy logic based controller is developed as a hybrid AI
effective. An improved GA is used in operation optimization of hybrid approach to control the flow of power between the hybrid RE system
RE system, which performs better than the traditional GA method and the energy storage unit, resulting, a high storage of charge (SOC)
[390]. Bee algorithm is used in performance parameters (net present [397].
cost (NPC), cost of energy (COE) and generation cost (GC)) optimiza- Some recent studies, [398–406] proposed the implementation of hybrid
tion of hybrid RE system [391]. Khatib et al. [392] have implemented and improved AI methods for different RE systems, like ANFIS in wind
GA in the optimization of the hybrid PV/wind system for size of PV power estimation [398,401], modeling of biodiesel [399], and solar
array and wind turbine and storage capacity. A multi-objective (MO)- radiation [400]; SVR+ARIMA for tidal current estimation [402]; empirical
ABC algorithm is used in hybrid (photo voltaic/wind turbine/fuel cell) decomposition, wavelet decomposition, ANN and autoregressive methods
energy system in size and distribution optimization [393], which in solar radiation estimation [403]; improved and hybrid ANN in load
results in a high voltage stability index (VSI). Markov based GA is estimation of PV system [404], and in wind speed and power prediction
used in size optimization of hybrid wind-PV-diesel system [394]. [405]; and data mining method based efficient energy management system
The performance of four techniques (PSO, tabu search (TS), [406]. The detailed applications of AI methods have been also discussed in
simulated annealing (SA), and harmony search (HS)) for the size some latest review studies [407–415], specifically, for power tracking of PV
optimization of PV/wind/battery and PV/wind/FC systems is described system [407,412,413], solar energy and wind energy estimation
in [395]. PSO results in better performance than rest three methods. [408,409,414], decision system in RE [410], controllers for PV systems
In hybrid AI approaches, ANFIS is used for size optimization of the [411], and energy management [415].
Table 8
Summary of reports for application of AI approaches in hybrid energy [386–397].
311
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
312
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
[65] Nomura N, Akai M. Willingness to pay for green electricity in Japan as estimated Renew Sust Energy Rev 2000;4(4):315–74.
through contingent valuation method. Appl Energy 2004;78(4):453–63. [107] Herbert GMS, Iniyan S, Sreevalsan E, Rajapandian S. A review of wind energy
[66] Sadrul-Islama AKM, Islam M, Rahman T. Effective renewable energy activities in technologies. Renew Sust Energy Rev 2007;11(6):1117–45.
Bangladesh. Renew Energy 2006;31(5):677–88. [108] Leung DYC, Yang Y. Wind energy development and its environmental impact: a
[67] Ringe M. Fostering the use of renewable energies in the European Union: the race review. Renew Sust Energy Rev 2012;16(1):1031–9.
between feed-in tariffs and green certificates. Renew Energy 2006;31(1):1–17. [109] Pryor SC, Barthelmie RJ. Climate change impacts on wind energy: a review.
[68] Lund H, Mathiesen BV. Energy system analysis of 100% renewable energy Renew Sust Energy Rev 2010;14(1):430–7.
systems—the case of Denmark in years 2030 and 2050. Energy [110] Hasan NS, Hassan MY, Majid MS, Rahman HA. Review of storage schemes for
2009;34(5):524–31. wind energy systems. Renew Sust Energy Rev 2010;14(7):1744–62.
[69] Pillai IR, Banerjee R. Renewable energy in India: status and potential. Energy [111] Lu B, Li Y, Wu X, Yang Z. A review of recent advances in wind turbine condition
2009;34(8):970–80. monitoring and fault diagnosis. USA: IEEE PEMWA; 2009.
[70] Ku S, Yoo S. Willingness to pay for renewable energy investment in Korea: a choice [112] Kreider JF, Kreith F. Solar energy handbook. NY: McGraw-Hill; 1981.
experiment study. Renew Sust Energy Rev 2010;14(8):2196–201. [113] Green MA. Solar cells: operating principles, technology, and system applications.
[71] Voivontas D, Assimacopoulos D, Mourelatos A, Corominas J. Evaluation of NJ: Prentice-Hall; 1982.
renewable energy potential using a GIS decision support system. Renew Energy [114] US Department of Energy, Energy Efficiency and Renewable Energy. 〈https://
1998;13(3):333–44. www1.eere.energy.gov/solar/pdfs/solar_timeline.pdf〉.
[72] Haralambopoulos DA, Polatidis H. Renewable energy projects: structuring a [115] Parida B, Iniyan S, Goicc R. A review of solar photovoltaic technologies. Renew
multi-criteria group decision-making framework. Renew Energy 2003;6:961–73. Sust Energy Rev 2011;15(3):1625–36.
[73] Beccali M, Cellura M, Mistretta M. Decision-making in energy planning. [116] Balcomb JD. Passive solar building. Cambridge: MIT Press; 1992.
Application of the Electre method at regional level for the diffusion of renewable [117] Schaller RD, Klimov VI. High efficiency carrier multiplication in PbSe nanocrys-
energy technology. Renew Energy 2003;28(13):2063–87. tals: implications for solar energy conversion. Phys Rev Lett 2004;92(18):186601.
[74] Cristóbal JRS. Multi-criteria decision-making in the selection of a renewable [118] Mor GK, Varghese OK, Paulose M, Shankar K, Grimes CA. A review on highly
energy project in Spain: the Vikor method. Renew Energy 2011;36(2):458–502. ordered vertically oriented TiO2 nanotube arrays: fabrication, material properties,
[75] Dufo-Lópeza R, Bernal-Agustína JL, Contreras J. Optimization of control strate- and solar energy applications. Sol Energy Mater Sol Cells 2006;90(14):2011–75.
gies for stand-alone renewable energy systems with hydrogen storage. Renew [119] Granqvist CG. Transparent conductors as solar energy materials: a panoramic
Energy 2007;32(7):1102–26. review. Sol Energy Mater Sol Cells 2007;17:1529–98.
[76] Ashok S. Optimised model for community-based hybrid energy system. Renew [120] Mahian O, Kianifar A, Kalogirou SA, Pop I, Wongwises S. A review of the
Energy 2007;32(7):1155–64. applications of nanofluids in solar energy. Int J Heat Mass Trans
[77] Senjyu T, Hayashi D, Yona A, Urasaki N, Funabashi T. Optimal configuration of 2013;57(2):582–94.
power generating systems in isolated island with renewable energy. Renew Energy [121] Solangi KH, Islam MR, Saidur R, Rahim NA, Fayaz H. A review on global solar
2007;32(11):1917–33. energy policy. Renew Sust Energy Rev 2011;15(4):2149–63.
[78] Haidar AMA, John PN, Shawal M. Optimal configuration assessment of renewable [122] Ekechukwu OV, Norton B. Review of solar-energy drying systems II: an overview
energy in Malaysia. Renew Energy 2011;36(2):881–8. of solar drying technology. Energy Convers Manag 1999;40(6):615–55.
[79] Morais H, Kádár P, Faria P, Vale ZA, Khodr HM. Optimal scheduling of a [123] Blanco J, Malato S, Fernández-Ibañez P, Alarcón D, Gernjak W, Maldonado MI.
renewable micro-grid in an isolated load area using mixed-integer linear Review of feasible solar energy applications to water processes. Renew Sust
programming. Renew Energy 2010;35(1):151–6. Energy Rev 2009;13(6):1437–45.
[80] Iqbal MT. Modeling and control of a wind fuel cell hybrid energy system. Renew [124] Mekhilef S, Saidur R, Safari A. A review on solar energy use in industries. Renew
Energy 2003;28(2):223–37. Sust Energy Rev 2011;15(4):1777–90.
[81] Deshmukh MK, Deshmukh SS. Modeling of hybrid renewable energy systems. [125] N’Tsoukpoe KE, Liu H, Le Pierrès N, Luo L. A review on long-term sorption solar
Renew Sust Energy Rev 2008;12(1):235–49. energy storage. Renew Sust Energy Rev 2009;13(9):2385–96.
[82] Bernal-Agustín JL, Dufo-López R. Simulation and optimization of stand-alone [126] Sartori I, Hestnes AG. Energy use in the life cycle of conventional and low-energy
hybrid renewable energy systems. Renew Sust Energy Rev 2009;13(8):2111–8. buildings: a review article. Energy Build 2007;39(3):249–57.
[83] Lund H, Münster E. Modelling of energy systems with a high percentage of CHP [127] Khatib T, Mohamed A, Sopian K. A review of solar energy modeling techniques.
and wind power. Renew Energy 2003;28(14):2189–93. Renew Sust Energy Rev 2012;16(5):2864–9.
[84] Myers DR. Solar radiation modeling and measurements for renewable energy [128] Fariba B, Dehghan AA, Faghih AR. Empirical models for estimating global solar
applications: data and model quality. Energy 2005;30(9):1517–31. radiation: a review and case study. Renew Sust Energy Rev 2013;21:798–821.
[85] Singh GK. Modeling and experimental analysis of a self-excited six-phase [129] Yadav AK, Chandel SS. Tilt angle optimization to maximize incident solar
induction generator for stand-alone renewable energy generation. Renew Energy radiation: a review. Renew Sust Energy Rev 2013;23:503–13.
2008;33(7):1605–21. [130] Verma P, Singal SK. Review of mathematical modeling on latent heat thermal
[86] Russel SJ, Norvig P. Artificial intelligence: a modern approach. New Jersey: energy storage systems using phase-change material. Renew Sust Energy Rev
Prentice Hall; 1995. 2008;12(4):999–1031.
[87] Turban E, Frenzel LE. Expert systems and applied artificial intelligence. New [131] White DE. Geothermal energy. Bull Volcanol 1966;29(1):481–3.
Jersey: Prentice Hall; 1992. [132] Armstead HCH. Geothermal energy: its past, present and future contributions to
[88] Kalogirou SA. Applications of artificial neural-networks for energy systems. Appl the energy needs of man. NY: Halsted Press; 1978.
Energy 2000;67:17–35. [133] Dickson MH, Fanelli M. Geothermal energy: utilization and technology. Paris:
[89] Kalogirou SA. Artificial neural networks in renewable energy systems applications: UNESCO; 2003.
a review. Renew Sust Energy Rev 2001;4(5):373–401. [134] Muffler LJP. Assessment of geothermal resources of the United States. Geol Surv
[90] Mellita A, Kalogirou SA. Artificial intelligence techniques for photovoltaic Circ 1978;790.
applications: a review. Prog Energy Combust Sci 2008;34(5):574–632. [135] Barbier E. Nature and technology of geothermal energy: a review. Renew Sust
[91] Lei M, Shiyan L, Chuanwen J, Hongling L, Yan Z. A review on the forecasting of Energy Rev 1997;1(1):1–69.
wind speed and generated power. Renew Sust Energy Rev 2009;13(4):915–20. [136] Barbier E. Geothermal energy technology and current status: an overview. Renew
[92] Mellita A, Kalogirou SA, Hontoria L, Shaarid S. Artificial intelligence techniques Sust Energy Rev 2002;6(1):3–65.
for sizing photovoltaic systems: a review. Renew Sust Energy Rev [137] Lund JW, Freeston DH. World-wide direct uses of geothermal energy 2000.
2009;13(2):406–19. Geothermics 2001;30(1):29–68.
[93] Baños R, Manzano-Agugliaro F, Montoya FG, Gil C, Alcayde A, Gómez J. [138] Fridleifsson IB. Geothermal energy for the benefit of the people. Renew Sust
Optimization methods applied to renewable and sustainable energy: a review. Energy Rev 2001;5(3):299–312.
Renew Sust Energy Rev 2011;15(4):1753–66. [139] Lund JW, Freeston DH, Boyd TL. Direct application of geothermal energy: 2005
[94] Zhao H, Magoulès F. A review on the prediction of building energy consumption. worldwide review. Geothermics 2005;34(6):691–727.
Renew Sust Energy Rev 2012;16(6):3586–92. [140] Menjoz A, Sauty JP. Characteristics and effects of geothermal resources exploi-
[95] Foley AM, Leahy PG, Marvuglia A, McKeogh EJ. Current methods and advances in tation. J Hydrol 1982;56:49–59.
forecasting of wind power generation. Renew Energy 2012;37(1):1–8. [141] Rybach L. Geothermal energy: sustainability and the environment. Geothermics
[96] Burton T, Sharpe D, Jenkins N, Bossanyi E. Wind energy handbook. UK: John 2003;32:463–70.
Wiley & Sons; 2001. [142] Haehnlein S, Bayer P, Blum P. International legal status of the use of shallow
[97] Manwell JF, McGowan JG, Rogers AL. Wind energy explained: theory, design and geothermal energy. Renew Sust Energy Rev 2010;14(9):2611–25.
application. UK: John Wiley & Sons; 2009. [143] Hochstein MP. Assessment and modelling of geothermal reservoirs (small
[98] Fleming PD, Probert SD. The evolution of wind-turbines: an historical review. utilization schemes). Geothermics 1988;17(1):15–49.
Appl Energy 1984;18:163–77. [144] Pruess K. Modeling of geothermal reservoirs: fundamental processes, computer
[99] Pasqualetti MJ, Righter R, Gipe P. Wind energy, history of, encyclopedia energy; simulation and field applications. Geothermics 1990;19(1):3–15.
2004, p. 419–33. [145] O'Sullivan MJ, Pruess K, Lippmann MJ. State of the art of geothermal reservoir
[100] Kaldellis JK, Zafirakis D. The wind energy (r)evolution: a short review of a long simulation. Geothermics 2001;30(4):395–429.
history. Renew Energy 2011;36(7):1887–901. [146] Sherman Josepha. Hydroelectric power. USA: Capstone Press; 2004.
[101] World Wind Energy Reports 2006–2015. 〈http://www.wwindea.org〉. [147] Wagner H, Mathur J. Introduction to hydro energy systems: basics, technology
[102] EWEA. Wind in power european statistics {C}2006–2015{C}. 〈http://www.ewea. and operation. Berlin: Springer-Verlag; 2011.
org/statistics〉. [148] Rupert CE. Hydropower: types, development strategies and environmental
[103] Wittrup S. Power from vestas giant turbine. 〈http://ing.dk/artikel/saa- impacts (energy science, engineering and technology). NY: Nova Science; 2014.
producerer-vestas-gigantmoelle-stroem-165903〉; 2014. [149] Deane JP, Ó Gallachóir BP, McKeogh EJ. Techno-economic review of existing and
[104] Sesto E. Wind energy in the world: reality and prospects. Renew Energy new pumped hydro energy storage plant. Renew Sust Energy Rev
1999;16(7):888–93. 2010;14(4):1293–302.
[105] Saidur R, Islam MR, Rahim NA, Solangi KH. A review on global wind energy [150] Yanga C, Jackson RB. Opportunities and barriers to pumped-hydro energy storage
policy. Renew Sust Energy Rev 2010;14(7):1744–62. in the United States. Renew Sust Energy Rev 2011;15(1):839–44.
[106] Ackermann T, Söder L. Wind energy technology and current status: a review. [151] Yeh W. Reservoir management and operations models: a state-of-the-art review.
313
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
Water Resour Res 1985;21(12):1797–818. stand-alone applications: a review. Renew Sust Energy Rev 2012;16(5):2926–39.
[152] Labadie J. Optimal operation of multireservoir systems: state-of-the-art review. J [195] Nehrir MH, Wang C, Strunz K, Aki H, Ramakumar R, Bing J, Miao Z, Salameh Z.
Water Resour Plan Manag 2004;130(2):93–111. A review of hybrid renewable/alternative energy systems for electric power
[153] Khan MJ, Bhuyan G, Iqbal MT, Quaicoe JE. Hydrokinetic energy conversion generation: configurations, control, and applications. IEEE Trans Sust Energy
systems and assessment of horizontal and vertical axis turbines for river and tidal 2011;2(4):392–403.
applications: a technology status review. Appl Energy 2009;86(10):1823–35. [196] Erdinc O, Uzunoglu M. Optimum design of hybrid renewable energy systems:
[154] Khan MJ, Iqbal MT, Quaicoe JE. River current energy conversion systems: overview of different approaches. Renew Sust Energy Rev 2012;16(3):1412–25.
progress, prospects and challenges. Renew Sust Energy Rev 2008;12(8):2127–93. [197] Nema P, Nema RK, Rangnekar S. A current and future state of art development of
[155] Padhy MK, Saini RP. A review on silt erosion in hydro turbines. Renew Sust hybrid energy system using wind and PV-solar: a review. Renew Sust Energy Rev
Energy Rev 2008;12(7):1974–87. 2009;13(8):2096–103.
[156] Mishra S, Singal SK, Khatod DK. Optimal installation of small hydropower plant— [198] Connolly D, Lund H, Mathiesen BV, Leahy M. A review of computer tools for
a review. Renew Sust Energy Rev 2011;15(8):3862–9. analysing the integration of renewable energy into various energy systems. Appl
[157] Sovacool BK, Dhakal S, Gippner O, Bambawale MJ. Halting hydro: a review of the Energy 2010;87(4):1059–82.
socio-technical barriers to hydroelectric power plants in Nepal. Energy [199] Etxeberria A, Vechiu I, Camblong H, Vinassa JM, Camblong H. Hybrid energy
2011;36(5):3468–76. storage systems for renewable energy sources integration in microgrids: a review.
[158] Mailmana M, Stepnuk L, Cicek N, Bodaly RA. Strategies to lower methyl mercury In: Proceedings IPEC; 2010, p. 1947–1262.
concentrations in hydroelectric reservoirs and lakes: a review. Sci Total Environ [200] Bhandari B, Poudel SR, Lee K, Ahn S. Mathematical modeling of hybrid renewable
2006;368(1):224–35. energy system: a review on small hydro-solar-wind power generation. Int J Precis
[159] Verma P, Varuna , Singal SK. Review of mathematical modeling on latent heat Eng Manuf Green Technol 2014;1(2):157–73.
thermal energy storage systems using phase-change material. Renew Sust Energy [201] Michalski RS, Carbonell JG, Mitchell TM. Machine learning: an artificial
Rev 2008;12(4):999–1031. intelligence approach. Berlin: Springer-Verlag; 1984.
[160] Charlier RH, Finkl CW. Ocean energy: tide and tidal power. Berlin: Springer- [202] Hayes-Roth F, Waterman D, Lenat D. Building expert systems. USA: Addison-
Verlag; 2009. Wesley; 1984.
[161] McCormick ME. Ocean wave energy conversion. NY: Dover Publication; 2007. [203] Jackson P. Introduction to expert systems. USA: Addison-Wesley; 1986.
[162] Westwood A. Ocean power: wave and tidal energy review. Refocus [204] Poole D, Mackworth A. Artificial intelligence foundation of computational agents.
2004;5(5):50–5. UK: Cambridge University Press; 2010.
[163] Ferro BD. Wave and tidal energy: its emergence and the challenges it faces. [205] Russsell S, Norvig P. Artificial intelligence, a modern approach, 3rd ed.. UK:
Refocus 2006;7(3):46–8. Pearson; 2014.
[164] Denny E. The economics of tidal energy. Energy Policy 2009;37(5):1914–24. [206] Smolensky P. Connectionist AI, symbolic AI, and the brain. AI Rev
[165] Drew B, Plummer AR, Sahinkaya MN. A review of wave energy converter 1987;1(2):95–109.
technology. J Power Energy 2009;223(8):887–902. [207] Reed RD, Marks RJ. Neural smithing: supervised learning in feedforward artificial
[166] Estebana M, Leary D. Current developments and future prospects of offshore wind neural networks. USA: MIT Press; 1998.
and ocean energy. Appl Energy 2012;90(1):128–36. [208] Vapnik V. The nature of statistical learning theory. NY: Springer-Verlag; 2013.
[167] Large WG, McWilliams JC, Doney SC. Oceanic vertical mixing: a review and a [209] Fogel DB. Evolutionary computation: toward a new philosophy of machine
model with a nonlocal boundary layer parameterization. Rev Geophy intelligence. New Jersey: IEEE Press; 2006.
1994;32(4):363–403. [210] Cordón O. Genetic fuzzy systems: evolutionary tuning and learning of fuzzy
[168] Khanal SK, Surampalli RY, Zhang TC, Lamsal BP, Tyagi RD, Kao CM. Bioenergy knowledge bases. Singapore: World Scientific; 2001.
and biofuel from biowastes and biomass. USA: ASCE Publication; 2010. [211] Colak I, Sagiroglu S, Yesilbudak M. Data mining and wind power prediction: a
[169] Lee S, Shah YT. Biofuels and bioenergy: processes and technologies. Boca Raton literature review. Renew Energy 2012;46:241–7.
FL: CRC Press; 2013. [212] Zhang Y, Wang J, Wang X. Review on probabilistic forecasting of wind power
[170] Dahiya A. Bioenergy: biomass to biofuels. USA: Academic Press; 2015. generation. Renew Sust Energy Rev 2014;32:255–70.
[171] Balat M, Balat H. Recent trends in global production and utilization of bio-ethanol [213] Tascikaraoglu A, Uzunoglu M. A review of combined approaches for prediction of
fuel. Appl Energy 2009;86(11):2273–82. short-term wind speed and power. Renew Sust Energy Rev 2014;34:243–54.
[172] Mata TM, Martins AA, Caetano NS. Microalgae for biodiesel production and other [214] Mabel MC, Fernandez E. Analysis of wind power generation and prediction using
applications: a review. Renew Sust Energy Rev 2010;14(1):217–32. ANN: a case study. Renew Energy 2008;33(5):986–92.
[173] Du Z, Li H, Gu T. A state of the art review on microbial fuel cells: a promising [215] Li G, Shi J. On comparing three artificial neural networks for wind speed
technology for wastewater treatment and bioenergy. Biotechnol Adv forecasting. Appl Energy 2010;87(7):2313–20.
2007;25(5):464–82. [216] Mabel MC, Fernandez E. Estimation of energy yield from wind farms using
[174] Karthikeyan OP, Visvanathan C. Bio-energy recovery from high-solid organic artificial neural networks. IEEE Trans Energy Convers 2009;24(2):459–64.
substrates by dry anaerobic bio-conversion processes: a review. Rev Environ Sci [217] Kariniotakis GN, Stavrakakis GS, Nogaret EF. Wind power forecasting using
Biotechnol 2013;12:257–84. advanced neural networks models. IEEE Trans Energy Convers
[175] Mohan D, Pittman CU, Steele PH. Pyrolysis of wood/biomass for bio-oil: a critical 2002;11(4):762–7.
review. Energy Fuel 2006;20(3):848–89. [218] Öztopal A. Artificial neural network approach to spatial estimation of wind
[176] McKendry P. Energy production from biomass (part 2): conversion technologies. velocity data. Energy Convers Manag 2006;47(4):395–406.
Bioresour Technol 2002;83(1):47–89. [219] Alexiadis MC, Dokopoulos PS, Sahsamanoglou HS. Wind speed and power
[177] Gold S, Seuring S. Supply chain and logistics issues of bio-energy production. J forecasting based on spatial correlation models. IEEE Trans Energy Convers
Clean Prod 2011;19(1):32–42. 1999;14(3):836–42.
[178] Demirbas MF. Biorefineries for biofuel upgrading: a critical review. Appl Energy [220] Li G, Shi J, Zhou J. Bayesian adaptive combination of short-term wind speed
2009;86(1):S151–S161. forecasts from neural network models. Renew Energy 2011;36(1):352–9.
[179] Faaij APC. Bio-energy in Europe: changing technology choices. Energy Policy [221] Sfetsos A. A comparison of various forecasting techniques applied to mean hourly
2006;34(3):322–42. wind speed time series. Renew Energy 2000;21(1):23–35.
[180] Smeets EMW, Faaij APC, Lewandowski IM, Turkenburg WC. A bottom-up [222] Cadenas E, Jaramillo OA, Rivera W. Analysis and forecasting of wind velocity in
assessment and review of global bio-energy potentials to 2050. Prog Energy chetumal, quintana roo, using the single exponential smoothing method. Renew
Combust 2007;33(1):56–106. Energy 2010;35(5):925–30.
[181] Busby RL. Hydrogen and fuel cells: a comprehensive guide. Oklahoma: Pennwell; [223] Simoes MG, Bose BK, Spiegel RJ. Design and performance evaluation of a fuzzy-
2005. logic-based variable-speed wind generation system. IEEE Trans Ind Appl
[182] Drennen TE, Rosthal JE. Pathways to a hydrogen future. UK: Elsevier; 2007. 1997;33(4):956–65.
[183] Sherif SA, Goswami DY, Stefanakos EK, Steinfeld A. Handbook of hydrogen [224] Sideratos G, Hatziargyriou ND. An advanced statistical method for wind power
energy. NW: CRC Press; 2014. forecasting. IEEE Trans Power Syst 2007;22(1):258–65.
[184] Momirlan M, Veziroglu TN. Current status of hydrogen energy. Renew Sust [225] Monfared M, Rastegar H, Kojabadi HM. A new strategy for wind speed forecasting
Energy Rev 2002;6(1):141–79. using artificial intelligent methods. Renew Energy 2009;34(3):845–8.
[185] Bolton JR. Solar photoproduction of hydrogen: a review. Sol Energy [226] Juban J, Siebert N, Kariniotakis GN. Probabilistic short-term wind power
1996;57(1):37–50. forecasting for the optimal management of wind generation. In: Proceedings IEEE
[186] Wang J, Wan W. Factors influencing fermentative hydrogen production: a review. Power Tech; 2007, p. 683–5.
Int J Hydrog Energy 2009;34(2):799–811. [227] Mohandes MA, Halawani TO, Rehman S, Hussain AA. Support vector machines
[187] Sakintuna B, Lamari-Darkrim F, Hirscher M. Metal hydride materials for solid for wind speed prediction. Renew Energy 2004;29(6):939–47.
hydrogen storage: a review. Int J Hydrog Energy 2007;32(9):1121–40. [228] Potter CW, Negnevitsky M. Very short-term wind forecasting for Tasmanian
[188] Cheng X, Shi Z, Glass N, Zhang L, Zhang J, Song D, Liu Z, Wang H, Shen J. A power generation. IEEE Trans Power Syst 2006;21(2):965–72.
review of PEM hydrogen fuel cell contamination: impacts, mechanisms, and [229] Mohandes M, Rehman S, Rahman SM. Estimation of wind speed profile using
mitigation. J Power Sources 2007;165(2):739–56. adaptive neuro-fuzzy inference system (ANFIS). Appl Energy
[189] Midilli A, Aya M, Dincer I, Rosen MA. On hydrogen and hydrogen energy 2011;88(11):4024–32.
strategies: i: current status and needs. Renew Sust Energy Rev 2005;9(3):255–71. [230] Yang Z, Liu Y, Li C. Interpolation of missing wind data based on ANFIS. Renew
[190] Nicoletti G. The hydrogen option for energy: a review of technical, environmental Energ 2011;36(3):993–8.
and economic aspects. Int J Hydrog Energy 1995;20(10):759–65. [231] Meharrar A, Tioursi M, Hatti M, Stambouli AB. A variable speed wind generator
[191] Hunter R, Elliot G. Wind-diesel systems: a guide to the technology and its maximum power tracking based on adaptative neuro-fuzzy inference system.
implementation. NY: Cambridge University Press; 1994. Expert Syst Appl 2011;38(6):7659–64.
[192] Al-Hallaj S, Kiszynski K. Hybrid hydrogen systems: stationary and transportation [232] Yang S, Li W, Wang C. The intelligent fault diagnosis of wind turbine gearbox
applications. London: Springer-Verlag; 2011. based on artificial neural network. In: Proceedings CMD; 2008, p. 1327–1330.
[193] Agrawal A, Wies R, Johnson R. Hybrid electric power systems: modeling, [233] Jursa R, Rohrig K. Short-term wind power forecasting using evolutionary
optimization and control. Germany: VDM Publishing; 2007. algorithms for the automated specification of artificial intelligence models. Int J
[194] Bajpai P, Dash V. Hybrid renewable energy systems for power generation in Forecast 2008;24(4):694–709.
314
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
[234] Guo Z, Zhao W, Lu H, Wang J. Multi-step forecasting for wind speed using a [268] Souliotis M, Kalogirou S, Tripanagnostopoulos Y. Modelling of an ICS solar water
modified EMD-based artificial neural network model. Renew Energy heater using artificial neural networks and TRNSYS. Renew Energy
2012;37(1):241–9. 2009;34(5):1333–9.
[235] Pourmousavi-Kani SA, Ardehali MM. Very short-term wind speed prediction: a [269] Monteiro C, Santos T, Fernandez-Jimenez LA, Ramirez-Rosado IJ, Terreros-
new artificial neural network–Markov chain model. Energy Convers Manag Olarte MS. Short-term power forecasting model for photovoltaic plants based on
2011;52(1):738–45. historical similarity. Energies 2013;6:2624–43.
[236] Damousis IG, Petros D. A fuzzy expert system for the forecasting of wind speed [270] Mellit A, Benghanem M, Arab AH, Guessoum A. An adaptive artificial neural
and power generation in wind farms. In: Proceedings 22nd IEEE PES; 2001, p. network model for sizing stand-alone photovoltaic systems: application for
63–69. isolated sites in Algeria. Renew Energy 2009;34(5):1333–9.
[237] Hu J, Wang J, Zeng G. A hybrid forecasting approach applied to wind speed time [271] Mellit A, Benghanem M, Kalogirou SA. An adaptive wavelet-network model for
series. Renew Energy 2013;60(11):185–94. forecasting daily total solar-radiation. Appl Energy 2006;83(7):705–22.
[238] Cadenas E, Rivera W. Wind speed forecasting in three different regions of Mexico, [272] Pedro HTC, Coimbra CFM. Assessment of forecasting techniques for solar power
using a hybrid ARIMA–ANN model. Renew Energy 2010;35(12):2732–8. production with no exogenous inputs. Sol Energy 2012;86(7):2017–28.
[239] Salcedo-Sanz S, Pérez-Bellido AM, Ortiz-García EG, Portilla-Figueras A, Prieto L, [273] Mandal P, Madhira STS, Ul-haque A, Meng J, Pineda RL. Forecasting power
Paredes D. Hybridizing the fifth generation mesoscale model with artificial neural output of solar photovoltaic system using wavelet transform and artificial
networks for short-term wind speed prediction. Renew Energy intelligence techniques. Procedia Comput Sci 2012;12:332–7.
2009;34(6):1451–7. [274] Kalogirou SA. Optimization of solar systems using artificial neural-networks and
[240] Liu D, Niu D, Wang H, Fan L. Short-term wind speed forecasting using wavelet genetic algorithms. Appl Energy 2004;77(4):383–405.
transform and support vector machines optimized by genetic algorithm. Renew [275] Mellit A, Kalogirou SA. ANFIS-based modelling for photovoltaic power supply
Energy 2014;62:592–7. system: a case study. Renew Energy 2011;36(1):250–8.
[241] Kong X, Liu X, Shi R, Lee KY. Wind speed prediction using reduced support vector [276] Zarzalejo LF, Ramirez L, Polo J. Artificial intelligence techniques applied to hourly
machines with feature selection. Neurocomputing 2015;169(2):449–56. global irradiance estimation from satellite-derived cloud index. Energy
[242] Rahmani R, Yusof R, Seyedmahmoudian M, Mekhilef S. Hybrid technique of ant 2005;30(9):1685–97.
colony and particle swarm optimization for short term wind energy forecasting. J [277] Mellit A, Kalogirou SA, Shaari S, Salhi H, Arab AH. Methodology for predicting
Wind Eng Ind Aerodyn 2013;123:163–70. sequences of mean monthly clearness index and daily solar radiation data in
[243] Pousinho HMI, Mendes VMF, Catalão JPS. A risk-averse optimization model for remote areas: application for sizing a stand-alone PV system. Renew Energy
trading wind energy in a market environment under uncertainty. Energy 2008;33(7):1570–90.
2011;36(8):4935–42. [278] Mellit A, Kalogirou SA. Neuro-Fuzzy based modeling for photovoltaic power
[244] Dounis AI, Caraiscos C. Advanced control systems engineering for energy and supply system. In: Proceedings IEEE PES Putra; 2006, p. 1–6.
comfort management in a building environment—a review. Renew Sust Energy [279] Mellit A, Arab AH, Khorissi N, Salhi H. An ANFIS-based forecasting for solar
Rev 2009;13(6):1246–61. radiation data from sunshine duration and ambient temperature. In Proceedings
[245] Mellit A. Artificial Intelligence technique for modelling and forecasting of solar IEEE PES Tampa; 2007, p. 1–6.
radiation data: a review. Inter J Artif Intell Soft Comput 2008;1(1):52–76. [280] Amirkhani S, Nasirivatan S, Kasaeian AB, Hajinezhad A. ANN and ANFIS models
[246] Mellit A, Pavan AM. A 24-h forecast of solar irradiance using artificial neural to predict the performance of solar chimney power plants. Renew Energy
network: application for performance prediction of a grid-connected PV plant at 2015;83:579–607.
Trieste, Italy. Sol Energy 2010;84(5):807–21. [281] Caputo D, Grimaccia F, Mussetta M, Zich RE. Photovoltaic plants predictive
[247] Rehman S, Mohandes M. Artificial neural network estimation of global solar model by means of ANN trained by a hybrid evolutionary algorithm. In:
radiation using air temperature and relative humidity. Energy Policy Proceedings IEEE JCNN Barcelona; 2010, p.1–6.
2008;36(2):571–6. [282] Ji W, Chee KC. Prediction of hourly solar radiation using a novel hybrid model of
[248] Kalogirou SA, Panteliou S, Dentsoras A. Artificial neural networks used for the ARMA and TDNN. Sol Energy 2011;85(5):808–17.
performance prediction of a thermosiphon solar water heater. Renew Energy [283] Ogliari E, Grimaccia F, Leva S, Mussetta M. Hybrid predictive models for accurate
1999;18(1):87–99. forecasting in PV systems. Energies 2013;6(4):1918–29.
[249] Alam S, Kaushik SC, Garg SN. Computation of beam solar radiation at normal [284] Bouzerdoum M, Mellit A, Pavan AM. A hybrid model (SARIMA–SVM) for short-
incidence using artificial neural network. Renew Energy 2006;31(10):1483–91. term power forecasting of a small-scale grid-connected photovoltaic plant. Sol
[250] Dombayci OA, Gölcü M. Daily means ambient temperature prediction using Energy 2013;98:226–35.
artificial neural network method: a case study of Turkey. Renew Energy [285] Olatomiwa L, Mekhilef S, Shamshirband S, Mohammadi K, Petkovic´ D, Sudheer
2009;34(4):1158–61. C. A support vector machine–firefly algorithm-based model for global solar
[251] Bosch JL, López G, Batlles FJ. Daily solar irradiation estimation over a radiation prediction. Sol Energy 2015;115(9):632–44.
mountainous area using artificial neural networks. Renew Energy [286] Satrape JV Potential impacts of artificial intelligence expert systems on geother-
2008;33(7):1622–8. mal well drilling costs. Technical Report; 87: AC03-86SF16299.
[252] Almonacid F, Fernández EF, Rodrigo P, Pérez-Higueras PJ, Rus-Casas C. [287] Pruess K. Modeling of geothermal reservoirs: fundamental processes, computer
Estimating the maximum power of a high concentrator photovoltaic (HCPV) simulation and field applications. Geothermics 1990;19(1):3–15.
module using an artificial neural network. Energy 2013;53(1):165–72. [288] Sanyal SK, Butler SJ, Swenson D, Hardeman B. Review of the state-of-the art of
[253] Mubiru J, Banda EJKB. Estimation of monthly average daily global solar numerical simulation of enhanced geothermal systems. In: Proceedings WGC
irradiation using artificial neural networks. Sol Energy 2008;82(2):181–7. Japan; 2000, p. 3853–8.
[254] Kalogirou SA. Long-term performance prediction of forced circulation solar [289] O’Sullivan MJ, Pruess K, Lippmann MJ. State of the art of geothermal reservoir
domestic water heating systems using artificial neural networks. Appl Energy simulation. Geothermics 2001;30:395–429.
2000;66(1):63–74. [290] O’Sullivan MJ, Yeh A, Mannington WI. A history of numerical modelling of the
[255] Fadare DA. Modelling of solar energy potential in Nigeria using an artificial neural Wairakei geothermal field. Geothermics 2009;38:155–68.
network model. Appl Energy 2009;86(9):1410–22. [291] Esen H, Inalli M. Modelling of a vertical ground coupled heat pump system by
[256] Kalogirou SA, Bojic M. Artificial neural networks for the prediction of the energy using artificial neural networks. Expert Syst Appl 2009;36:10229–38.
consumption of a passive solar building. Energy 2000;25(5):479–91. [292] Bassam A, Santoyo E, Andaverde J, Herna´ndez JA, Espinoza-Ojeda OM.
[257] Ekici BB, Aksoy UT. Prediction of building energy consumption by using artificial Estimation of static formation temperatures in geothermal wells by using an
neural networks. Avd Eng Softw 2009;40(5):356–62. artificial neural network approach. Comput Geosci 2010;36:1191–9.
[258] Tasadduq I, Rehman S, Bubshait K. Application of neural networks for the [293] Arslan O. Power generation from medium temperature geothermal resources:
prediction of hourly mean surface temperatures in Saudi Arabia. Renew Energy ann-based optimization of Kalina cycle system-34. Energy 2011;36:2528–34.
2002;25(4):545–54. [294] Arslan O, Yetik O. ANN based optimization of supercritical ORC-Binary
[259] Alam S, Kaushik SC, Garg SN. Assessment of diffuse solar energy under general geothermal power plant: simav case study. Appl Therm Eng 2011;31:3922–8.
sky condition using artificial neural network. Appl Energy 2009;86(4):554–64. [295] Kalogirou SA, Florides GA, Pouloupatis PD, Panayides I, Joseph-Stylianou J,
[260] Tymvios FS, Jacovides CP, Michaelides SC, Scouteli C. Comparative study of Zomeni Z. Artificial neural networks for the generation of geothermal maps of
Ångström's and artificial neural networks' methodologies in estimating global ground temperature at various depths by considering land configuration. Energy
solar radiation. Sol Energy 2005;78(6):752–62. 2012;48:233–40.
[261] Jiang Y. Computation of monthly mean daily global solar radiation in China using [296] Kecebas A, Yabanova I. Thermal monitoring and optimization of geothermal
artificial neural networks and comparison with other empirical models. Energy district heating systems using artificial neural network: a case study. Energy Build
2009;34(9):1276–83. 2012;50:339–46.
[262] Zeng J, Qiao W. Short-term solar power prediction using a support vector [297] A´ lvarez del Castillo A, Santoyo E, Garcı´a-Valladares O. A new void fraction
machine. Renw Energy 2013;52:118–27. correlation inferred from artificial neural networks for modeling two-phase flows
[263] Li Z, Rahman SMM, Vega R, Dong B. A hierarchical approach using machine in geothermal wells. Comput Geosci 2012;41:25–39.
learning methods in solar photovoltaic energy production forecasting. Energies [298] Tota-Maharaj K, Scholz M. Artificial neural network simulation of combined
2016;9(1):2–12. permeable pavement and earth energy systems treating storm water. J Environ
[264] Sharma N, Sharma P, Irwin D, Shenoy P. Predicting solar generation from Eng 2012;138:499–509.
weather forecasts using machine learning. In: Proceedings IEEE Smart Grid [299] Yabanova I, Keçebas A. Development of ANN model for geothermal district
Communication; 2011, p. 528–33. heating system and a novel PID-based control strategy. Appl Therm Eng
[265] Mashohor S, Samsudin K, Noor AM, Rahman ARA. Evaluation of genetic 2013;51:908–16.
algorithm based solar tracking system for photovoltaic panels. In: Proceedings [300] Arslan O, Yetik O. ANN modeling of an ORC-binary geothermal power plant:
IEEE ICSET Singapore; 2008, p. 269–73. Simav case study. Energy Sources Part A 2014;36:418–28.
[266] Atia DM, Fahmy FH, Ahmed NM, Dorrah HT. Optimal sizing of a solar water [301] Yeo I, Yee J. A proposal for a site location planning model of environmentally
heating system based on a genetic algorithm for an aquaculture system. Math friendly urban energy supply plants using an environment and energy geogra-
Comput Model 2012;55(3):1436–49. phical information system (E-GIS) database (DB) and an artificial neural network
[267] Kumar P, Jain G, Palwalia DK. Genetic algorithm based maximum power tracking (ANN). Appl Energy 2014;119:99–117.
in solar power generation. In: IEEE ICPACE Bangalore; 2015, p. 1–6. [302] Kalogirou SA, Florides GA, Pouloupatis PD, Christodoulides P, Joseph-Stylianou
315
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
J. Artificial neural networks for the generation of a conductivity map of the [339] Malekmohamadi I, Ghiassi R, Yazdanpanah MJ. Wave hindcasting by coupling
ground. Renew Energy 2015;77:400–7. numerical model and artificial neural networks. Ocean Eng 2008;35:417–25.
[303] Bassam A, Álvarez del Castillo A, García-Valladares O, Santoyo E. Determination [340] De-Paz JF, Bajo J, González A, Rodríguez S, Corchado JM. Combining case-based
of pressure drops in flowing geothermal wells by using artificial neural networks reasoning systems and support vector regression to evaluate the atmosphere–
and wellbore simulation tools. Appl Therm Eng 2013;51:908–16. ocean interaction. Knowl Inf Syst 2012;30:155–77.
[304] Sayyaadi H, Amlashi EH, Amidpour M. Multi-objective optimization of a vertical [341] Shabani N, Akhtari S, Sowlati T. Value chain optimization of forest biomass for
ground source heat pump using evolutionary algorithm. Energy Convers Manag bioenergy production: a review. Renew Sust Energy Rev 2013;23:2992, [311].
2009;50:2035–46. [342] Yang H, Ring Z, Briker Y, McLean N, Friesen W, Fairbridge C. Neural network
[305] Beck M, de Paly M, Hecht-Méndez J. Evaluation of the performance of evolu- prediction of cetane number and density of diesel fuel from its chemical
tionary algorithms for optimization of low-enthalpy geothermal heating plants. In: composition determined by LC and GC–MS. Fuel 2002;81:65–74.
Proceedings GECCO USA; 2012, p. 1047–54. [343] Strik D, Domnanovich AM, Zani L, Braun R, Holubar P. Prediction of trace
[306] Farghally HM, Atia DM, El-madany HT, Fahmy FH. Control methodologies based compounds in biogas from anaerobic digestion using the MATLAB neural network
on geothermal recirculating aquaculture system. Energy 2014;78:826–33. toolbox. Environ Model Softw 2005;20:803–10.
[307] Farghally HM, Atia DM, El-madany HT, Fahmy FH. Fuzzy logic controller based [344] Ramadhas AS, Jayaraj S, Muraleedharan C, Padmakumari K. Artificial neural
on geothermal recirculating aquaculture system. Egypt J Aquat Res networks used for the prediction of the cetane number of biodiesel. Renew Energy
2014;40:103–9. 2006;31:2524–33.
[308] Esen H, Inalli M. ANN and ANFIS models for performance evaluation of a vertical [345] Ozkaya B, Demir A, Bilgili MS. Neural network prediction model for the methane
ground source heat pump system. Expert Syst Appl 2010;37:8134–47. fraction in biogas from field-scale landfill bioreactors. Environ Modell Softw
[309] Şahin AS, Yazıcı H. Thermodynamic evaluation of the Afyon geothermal district 2007;22:518–22.
heating system by using neural network and neuro-fuzzy. J Volcanol Geotherm [346] Balabin RM, Lomakina EI, Safieva RZ. Neural network (ANN) approach to
Res 2012;233–234:65–71. biodiesel analysis: analysis of biodiesel density, kinematic viscosity, methanol and
[310] Porkhial S, Salehpour M, Ashraf H, Jamalic A. Modeling and prediction of water contents using near infrared (NIR) spectroscopy. Fuel 2011;90:2007–15.
geothermal reservoir temperature behavior using evolutionary design of neural [347] Kumar S, Pai PS, Rao BRS. Radial-basis-function-network-based prediction of
networks. Geothermics 2015;53:320–7. performance and emission characteristics in a bio diesel engine run on WCO ester.
[311] Kishor N, Saini RP, Singh SP. A review on hydropower plant models and control. Adv Artif Intell 2012;2012:1–7.
Renew Sust Energy Rev 2007;11:776–96. [348] Balabin RM, Safieva RZ. Biodiesel classification by base stock type (vegetable oil)
[312] Nourani V, Baghanam AH, Adamowski J, Kisi O. Applications of hybrid wavelet– using near infrared spectroscopy data. Anal Chim Acta 2011;689:190–7.
Artificial Intelligence models in hydrology: a review. J Hydrol [349] Izquierdo J, Minciardi R, Montalvo I, Robba M, Tavera M. Particle swarm
2014;514(6):358–77. optimization for the biomass supply chain strategic planning. In: Proceedings
[313] Liang RH, Hsu YY. Scheduling of hydroelectric generations using artificial neural IEMSS; 2008, p. 1272–80.
networks. IEE Proc Gener Transm Distrib 1994;141(5):452–8. [350] Ghugare SB, Tiwary S, Elangovan V, Tambe SS. Prediction of higher heating value
[314] Smith J, EIi RN. Neural network models of rainfall-runoff process. J Water of solid biomass fuels using artificial intelligence formalisms. Bioenergy Res
Resour Plan Manag 1995;121:499–508. 2014;7(2):681–92.
[315] Dolling OR, Varas EA. Artificial neural networks for streamflow prediction. J [351] Koutroumanidis T, Ioannou K, Arabatzis G. Predicting fuelwood prices in Greece
Hydraul Res 2002;40(5):547–54. with the use of ARIMA models artificial neural networks and a hybrid ARIMA–
[316] Kişi Ö. River flow modeling using artificial neural networks. J Hydrol Eng ANN model. Energy Policy 2009;37:3627–34.
2004;9(1):60–3. [352] Romeo LM, Gareta R. Fouling control in biomass boilers. Biomass- Bioenergy
[317] Estoperez N, Nagasaka K. A month ahead micro-hydro power generation 2009;33:854–61.
scheduling using artificial neural network. In: Proceedings IEEE PES; 2005, p. [353] Qdais HA, Hani KB, Shatnawi N. Modeling and optimization of biogas production
28–34. from a waste digester using artificial neural network and genetic algorithm.
[318] Carneiro AAFM, Leite PT, Filho DS, Carvalho ACPLF. Genetic algorithms applied Resour Conserv Recycl 2010;54:359–63.
to hydrothermal system scheduling. In: Proceedings IEEE ICPST China; 2005, p. [354] Kana EBG, Oloke JK, Lateef A, Adesiyan MO. Modeling and optimization of
547–51. biogas production on saw dust and other co-substrates using artificial neural
[319] Gil E, Bustos J, Rudnick H. Short-term hydrothermal generation scheduling network and genetic algorithm. Renew Energy 2012;46:276–81.
model using a genetic algorithm. IEEE Trans Power Syst 2003;18(4):1256–64. [355] Petrone R, Zheng Z, Hissel D, Pera MC, Pianese C, Sorrentino M, Becherif M,
[320] Yuan X, Zhang Y, Yuan Y. Improved self-adaptive chaotic genetic algorithm for Yousfi-Steiner N. A review on model-based diagnosis methodologies for PEMFCs.
hydrogeneration scheduling. J Water Resour Plan Manag 2008;134:319–25. Int J Hydrog Energy 2013;38:7077–91.
[321] Adhikary P, Roy PK, Mazumdar A. Selection of penstock material for small hydro [356] Zheng Z, Petrone R, Pera MC, Hissel D, Becherif M, Pianese C, Steiner NS,
power project – a fuzzy logic approach. Int J Adv Sci Technol Res Sorrentino M. A review on non-model based diagnosis. Int J Hydrog Energy
2012;2(6):521–8. 2013;38:8914–26.
[322] Chang L, Chang F. Intelligent control for modelling of real-time reservoir [357] Garg A, Vijayaraghavan A, Mahapatra SS, Tai K, Wonga CH. Performance
operation. Hydrol Process 2001;15:1621–34. evaluation of microbial fuel cell by artificial intelligence methods. Expert Syst Appl
[323] Firat M, Gungor M. River flow estimation using adaptive neuro fuzzy inference 2014;41:1389–99.
system. Math Comput Simul 2007;75:87–96. [358] Liu Y, Liu Y, Liu D, Cao T, Han S, Xu G. Design of CO2 hydrogenation catalyst by
[324] Molina JM, Isasi P, Berlanga A, Sanchis A. Hydroelectric power plant manage- an artificial neural network. Comput Chem Eng 2001;25:1711–4.
ment relying on neural networks and expert system integration. Eng Appl Artif [359] Ho T, Karri V, Lim D, Barret D. An investigation of engine performance
Intell 2000;13:357–69. parameters and artificial intelligent emission prediction of hydrogen powered car.
[325] Sinha SK, Patel RN, Prasad R. Application of GA and PSO tuned fuzzy controller Int J Hydrog Energy 2008;33:3837–46.
for AGC of three area thermal-thermal-hydropower system. Int J Comput Theory [360] Hatti M, Tioursi M. Dynamic neural network controller model of PEM fuel cell
Eng 2010;2(2):238–44. system. Int J Hydrog Energy 2009;34:5015–21.
[326] Toro CHF, Meire SG, Gálvez JF, Fdez-Riverola F. A hybrid artificial intelligence [361] Chavez-Ramirez AU, Munoz-Guerrero R, Duron-Torres SM, Ferraro M,
model for river flow forecasting. Appl Soft Comput 2013;13:3449–58. Brunaccini G, Sergi F, Antonucci V, Arriaga LB. High power fuel cell simulator
[327] Uzlu E, Akpinar A, Özturk HT, Nacar S, Kankal M. Estimates of hydroelectric based on artificial neural network. Int J Hydrog Energy 2010;35:12125–33.
generation using neural networks with the artificial bee colony algorithm for [362] Ho T, Karri V. Basic tuning of hydrogen powered car and artificial intelligent
Turkey. Energy 2014;69:638–47. prediction of hydrogen engine characteristics. Int J Hydrog Energy
[328] Adam JA. Probing beneath the sea: sending vessels into environments too harsh 2010;35:10004–12.
for humans poses challenges in communications, artificial intelligence, and [363] Yap WK, Ho T, Karri V. Exhaust emissions control and engine parameters
power-supply technology. IEEE Spectr 1985;22(4):55–64. optimization using artificial neural network virtual sensors for a hydrogen-
[329] Aartrijk ML, Tagliola CP, Adriaans PW. AI on the ocean: the RoboSail project. powered vehicle. Int J Hydrog Energy 2012;37:8704–15.
ECAI; 2002. [364] Vijayaraghavan V, Garg A, Wong CH, Tai K, Bhalerao Y. Predicting the mechanical
[330] Jain P, Deo MC. Neural networks in ocean engineering. Ships Offshore Struct characteristics of hydrogen functionalized graphene sheets using artificial neural
2006;1:25–35. network approach. J Nanostruct Chem 2013;3:1–5.
[331] Iglesias G, Carballo R. Wave resource in El Hierrodan Island towards energy self- [365] Marra D, Sorrentino M, Pianese C, Iwanschitz B. A neural network estimator of
sufficiency. Renew Energy 2011;36:689–98. solid oxide fuel cell performance for on-field diagnostics and prognostics
[332] Makarynskyy O, Makarynska D, Kuhn M, Featherstone WE. Predicting sea level applications. J Power Sources 2013;241(1):320–9.
variations with artificial neural networks at Hillarys Boat Harbour, Western [366] Tardast A, Rahimnejad M, Najafpour G, Ghoreyshi A, Premier GC, Bakeri G, Oh S.
Australia. Estuar Coast Shelf Sci. 2004;61:351–60. Use of artificial neural network for the prediction of bioelectricity production in a
[333] Londhe SN, Panchang V. One-day wave forecasts based on artificial neural membrane less microbial fuel cell. Fuel 2014;117:697–703.
networks. J Atmos Ocean Technol 2006;23:1593–603. [367] Ho T, Karri V. Fuzzy expert system to estimate ignition timing for hydrogen car.
[334] Makarynskyy O, Makarynska D. Wave prediction and data supplementation with LNCS 2008;5264:570–9.
artificial neural networks. J Costa Res 2007;23(4):951–60. [368] Flemming A, Adamy J. Modeling solid oxide fuel cells using continuous-time
[335] Toprak ZF, Cigizoglu HK. Predicting longitudinal dispersion coefficient in natural recurrent fuzzy systems. Eng Appl Artif Intell 2008;21:1289–300.
streams by artificial intelligence methods. Hydrol Process 2008;22:4106–29. [369] Caux S, Hankache W, Fadel M, Hissel D. On-line fuzzy energy management for
[336] Chen C, Lin J, Lee W, Chen C. Fuzzy control for an oceanic structure: a case study hybrid fuel cell systems. Int J Hydrog Energy 2010;35:2134–43.
in time-delay TLP system. J Vib Control 2010;16(1):147–60. [370] Caux S, Wanderley-Honda D, Hissel D, Fadel M. On-line energy management for
[337] Ghorbani MA, Makarynskyy O, Shiri J, Makarynska D. Genetic programming for HEV based on particle swarm optimization. In: Proceedings IEEE VPP FR; 2010,
sea level predictions in an island environment. Int J Ocean Clim Syst p. 1–7.
2010;1(1):27–36. [371] Nath K, Das D. Modeling and optimization of fermentative hydrogen production.
[338] Karimi S, Kisi O, Shiri J, Makarynskyy O. Neuro-fuzzy and neural network Bioresour Technol 2011;102:8569–81.
techniques for forecasting sea level in Darwin Harbor, Australia. Comput Geosci [372] Askarzadeh A, Rezazadeh A. A new heuristic optimization algorithm for modeling
2013;52:50–9. of proton exchange membrane fuel cell: bird mating optimizer. Int J Energy Res
316
S.K. Jha et al. Renewable and Sustainable Energy Reviews 77 (2017) 297–317
2013;37(10):1196–204. [394] Hong Y, Lian R. Optimal sizing of hybrid wind/PV/diesel generation in a stand-
[373] Entchev E, Yang L. Application of adaptive neuro-fuzzy inference system alone power system using Markov-based genetic algorithm. IEEE Trans Power Del
techniques and artificial neural networks to predict solid oxide fuel cell perfor- 2012;27(2):640–7.
mance in residential microgeneration installation. J Power Sources [395] Maleki A, Askarzadeh A. Comparative study of artificial intelligence techniques for
2007;170:122–9. sizing of a hydrogen-based stand-alone photovoltaic/wind hybrid system. Int J
[374] Ho T, Karri V, Madsen O. Artificial neural networks and neuro-fuzzy inference Hydrog Energy 2014;39:9973–84.
systems as virtual sensors for hydrogen safety prediction. Int J Hydrog Energy [396] Rajkumar RK, Ramachandaramurthy VK, Yong BL, Chia DB. Techno-economical
2008;33:2857–67. optimization of hybrid pv/wind/battery system using Neuro-Fuzzy. Energy
[375] Karri V, Ho TN. Predictive models for emission of hydrogen powered car using 2011;36:5148–53.
various artificial intelligent tools. Neural Comput Appl 2009;18:469–76. [397] Natsheh EM, Albarbar A. Hybrid power systems energy controller based on neural
[376] Becker S, Karri V. Predictive models for PEM-electrolyzer performance using network and fuzzy logic. Smart Grid Renew Energy 2013;4:187–97.
adaptive neuro-fuzzy inference systems. Int J Hydrog Energy 2010;35:9963–72. [398] Liu J, Wang X, Lu Y. A novel hybrid methodology for short-term wind power
[377] Amirinejad M, Tavajohi-Hasankiadeh N, Madaeni SS, Navarra MA, Rafiee E, forecasting based on adaptive neuro-fuzzy inference system. Renew Energy
Scrosati B. Adaptive neuro-fuzzy inference system and artificial neural network 2017;103:620–9.
modeling of proton exchange membrane fuel cells based on nanocomposite and [399] Naderloo L, Javadikia H, Mostafaei M. Modeling the energy ratio and productivity
recast Nafion membranes. Int J Energy Res 2013;37(4):347–57. of biodiesel with different reactor dimensions and ultrasonic power using ANFIS.
[378] Erdinc O, Vural B, Uzunoglu M. A wavelet-fuzzy logic based energy management Renew Sust Energy Rev 2017;70:56–64.
strategy for a fuel cell/battery/ultra-capacitor hybrid vehicular power system. J [400] Zou L, Wang L, Xia L, Lin A, Hu B, Zhu H. Prediction and comparison of solar
Power Sources 2009;194:369–80. radiation using improved empirical models and adaptive neuro-fuzzy inference
[379] Minqiang P, Dehuai Z, Gang X. Temperature prediction of hydrogen producing systems. Renew Energy 2017;106:343–53.
reactor using SVM regression with PSO. J Comput 2010;5(3):388–93. [401] Dong Q, Sun Y, Li P. A novel forecasting model based on a hybrid processing
[380] Prakasham RS, Sathish T, Brahmaiah P. Imperative role of neural networks strategy and an optimized local linear fuzzy neural network to make wind power
coupled genetic algorithm on optimization of biohydrogen yield. Int J Hydrog forecasting: a case study of wind farms in China. Renew Energy
Energy 2011;36:4332–9. 2017;102:241–57.
[381] Bozorgmehri S, Hamedi M. Modeling and optimization of anode-supported solid [402] Kavousi-Fard A. A hybrid accurate model for tidal current prediction. IEEE Trans
oxide fuel cells on cell parameters via artificial neural network and genetic Geosci Remote Sens 2017;55(1):112–8.
algorithm. Fuel Cells 2012;12(1):11–23. [403] Monjoly S, André M, Calif R, Soubdhan T. Hourly forecasting of global solar
[382] Zhang W, Wang N, Yang S. Hybrid artificial bee colony algorithm for parameter radiation based on multiscale decomposition methods: a hybrid approach. Energy
estimation of proton exchange membrane fuel cell. Int J Hydrog Energy 2017;119:288–98.
2013;38:5796–806. [404] Sepasi S, Reihani E, Howlader AM, Roose LR, Matsuura MM. Very short term
[383] Luna-Rubio R, Trejo-Perea M, Vargas-Vazquez D, Rios-Moreno GJ. Optimal load forecasting of a distribution system with high PV penetration. Renew Energy
sizing of renewable hybrids energy systems: a review of methodologies. Sol Energy 2017;106:142–8.
2012;86:1077–88. [405] Chang GW, Lu HJ, Chang YR, Lee YD. An improved neural network-based
[384] Zhou W, Lou C, Li Z, Lu L, Yang H. Current status of research on optimum sizing approach for short-term wind speed and power forecast. Renew Energy
of stand-alone hybrid solar–wind power generation systems. Appl Energy 2017;105:301–11.
2010;87:380–9. [406] Parvizimosaed M, Farmani F, Monsef H, Rahimi-Kian A. A multi-stage smart
[385] Fadaee M, Radzi MAM. Multi-objective optimization of a stand-alone hybrid energy management system under multiple uncertainties: a data mining ap-
renewable energy system by using evolutionary algorithms: a review. Renew Sust proach. Renew Energy 2017;102:178–89.
Energy Rev 2012;16:3364–9. [407] Kermadi M, Berkouk EM. Artificial intelligence-based maximum power point
[386] Al-Alawi A, Al-Alawi SM, Islam SM. Predictive control of an integrated PV-diesel tracking controllers for Photovoltaic systems: comparative study. Renew Sust
water and power supply system using an artificial neural network. Renew Energy Energy Rev 2017;69:369–86.
2007;32:1426–39. [408] Voyant C, Notton G, Kalogirou S, Nivet ML, Paoli C, Motte F, Fouilloy A. Machine
[387] Chavez-Ramirez AU, Vallejo-Becerra V, Cruz JC, Ornelas R, Orozco G, Munoz- learning methods for solar radiation forecasting: a review. Renew Energy
Guerrero R, Arriaga LG. A hybrid power plant (solar-wind-hydrogen) model based 2017;105:569–82.
in artificial intelligence for a remote-housing application in Mexico. Int J Hydrog [409] Wasilewski J, Baczynski D. Short-term electric energy production forecasting at
Energy 2013;38:2641–55. wind power plants in pareto-optimality context. Renew Sust Energy Rev
[388] Berrazouane S, Mohammedi K. Parameter optimization via cuckoo optimization 2017;69:177–87.
algorithm of fuzzy controller for energy management of a hybrid power system. [410] Kumar A, Sah B, Singh AR, Deng Y, He X, Kumar P, Bansal RC. A review of multi
Energ Convers Manag 2014;78:652–60. criteria decision making (MCDM) towards sustainable renewable energy devel-
[389] Hakimi SM, Moghaddas-Tafreshi SM. Optimal sizing of a stand-alone hybrid opment. Renew Sust Energy Rev 2017;69:596–609.
power system via particle swarm optimization for Kahnouj area in south-east of [411] Messalti S, Harrag A, Loukriz A. A new variable step size neural networks MPPT
Iran. Renew Energy 2009;34:1855–62. controller: review, simulation and hardware implementation. Renew Sust Energy
[390] Zeng J, Li M, Liu JF, Wu J, Ngan HW. Operational optimization of a stand-alone Rev 2017;68:221–33.
hybrid renewable energy generation system based on an improved genetic [412] Ramli MA, Twaha S, Ishaque K, Al-Turki YA. A review on maximum power point
algorithm. In: Proceedings IEEE PES MN; 2010, p. 1–6. tracking for photovoltaic systems with and without shading conditions. Renew
[391] Tudu B, Majumder S, Mandal KK, Chakraborty N. Optimal unit sizing of stand- Sust Energy Rev 2017;67:144–59.
alone renewable hybrid energy system using bees algorithm. In: Proceedings IEEE [413] Dileep G, Singh SN. Application of soft computing techniques for maximum power
ICEAS Odisha; 2011, p. 1–6. point tracking of SPV system. Sol Energy 2017;141:182–202.
[392] Khatib T, Mohamed A, Sopian K. Optimization of a PV/wind micro-grid for rural [414] Molina-Solana M, Ros M, Ruiz MD, Gómez-Romero J, Martin-Bautista MJ. Data
housing electrification using a hybrid iterative/genetic algorithm: case study of science for building energy management: a review. Renew Sust Energy Rev
Kuala Terengganu, Malaysia. Energy Build 2012;47:321–31. 2017;70:598–609.
[393] Nasiraghdam H, Jadid S. Optimal hybrid PV/WT/FC sizing and distribution [415] Zhang J, Zhao L, Deng S, Xu W, Zhang Y. A critical review of the models used to
system reconfiguration using multi-objective artificial bee colony (MOABC) estimate solar radiation. Renew Sust Energy Rev 2017;2017(70):314–29.
algorithm. Sol Energy 2012;86:3057–71.
317