Solar Power Forecasting Using Deep Learning Techniques PDF

Received February 17, 2022, accepted March 11, 2022, date of publication March 17, 2022, date of current version March 25, 2022. Digital Object Identifier 10.1109/ACCESS.2022.3160484 Solar Power Forecasting Using Deep Learning Techniques MEFTAH ELSARAITI , (Member, IEEE), AND ADEL MERABET , (Senior...

Received February 17, 2022, accepted March 11, 2022, date of publication March 17, 2022, date of current version March 25, 2022. Digital Object Identifier 10.1109/ACCESS.2022.3160484 Solar Power Forecasting Using Deep Learning Techniques MEFTAH ELSARAITI , (Member, IEEE), AND ADEL MERABET , (Senior Member, IEEE) Division of Engineering, Saint Mary’s University, Halifax, NS B3H 3C3, Canada Corresponding author: Meftah Elsaraiti ([email protected]) ABSTRACT The recent rapid and sudden growth of solar photovoltaic (PV) technology presents a future challenge for the electricity sector agents responsible for the coordination and distribution of electricity given the direct dependence of this type of technology on climatic and meteorological conditions. Therefore, the development of models that allow reliable future prediction, in the short term, of solar PV generation will be of paramount importance, in order to maintain a balanced and comprehensive operation. This article discusses a method for predicting the generated power, in the short term, of photovoltaic power plants, by means of deep learning techniques. To fulfill the above, a deep learning technique based on the Long Short Term Memory (LSTM) algorithm is evaluated with respect to its ability to forecast solar power data. An evaluation of the performance of the LSTM network has been conducted and compared it with the Multi- layer Perceptron (MLP) network using: Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), Root Mean Squared Error (RMSE) and Coefficient of Determination (R2 ). The prediction result shows that the LSTM network gives the best results for each category of days. Thus, it provides reliable information that enables more efficient operation of photovoltaic power plants in the future. The binomial formed by the concepts of deep learning and energy efficiency seems to have a promising future, especially regarding promoting energy sustainability, decarburization, and the digitization of the electricity sector. INDEX TERMS Prediction, deep learning, solar power, time series. I. INTRODUCTION nonlinear behaviors. In recent decades, the application of Renewable energy, especially solar PV, will gain prominence artificial neural networks (ANNs) in time series prediction as a major source of energy in the future. But as their share has grown due to the ideal characteristics offered by ANNs of the energy mix grows, ensuring the safety, reliability, for working with nonlinear models. Likewise, the develop- and profitability of power generation assets will be a top ment of applications that facilitate work when carrying out priority. Therefore, the successful integration of solar energy simulations with ANN continues to increase. In , artificial into the electrical grid requires an accurate prediction of neural networks are highlighted as one of the prediction the power generated by photovoltaic panels. Speaking of methods for time series thanks to their great adaptability and solar energy in particular, its unexpected behavior brings capacity to solve nonlinear and complex problems. In recent with it a series of problems when generating energy, such years, as a result of the research on artificial intelligence, as voltage variations, power factor details, and stability. For deep learning based on ANN has come to light to become this reason, these new tools are constantly being created very popular due to its capability to accelerate the solution that contribute to the prediction of future events, with the of some difficult computer problems. While multi-layer aim of reducing errors in predictions. Auto-Regressive perceptron (MLP)-type ANNs can be used to model complex Integrated Moving Averages (ARIMA) models have proven relationships, they are incapable of assimilating the long- and to be extremely useful for the short-term prediction of high- short-term dependencies present in historical data. These frequency time series. In contrast to ARIMA models and sta- dependencies refer to the ability of an ANN to identify and tistical methods, artificial neural networks are more powerful, remember behavior patterns from the distant past and the near especially in representing complex relationships that exhibit past, respectively. An ANN is needed to make predictions of sequential data behavior. As an attempt to address The associate editor coordinating the review of this manuscript and this problem, the first Recurrent Neural Networks (RNN) approving it for publication was Giambattista Gruosso. emerged in the 1980s, where the term ‘‘recurrent’’ refers to This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ 31692 VOLUME 10, 2022 M. Elsaraiti, A. Merabet: Solar Power Forecasting Using Deep Learning Techniques the characteristic of these networks to have internal feed- statistical methods. In the presence of noisy data, the back loops. However, these networks presented a great dis- LSTM model provides a more accurate prediction compared advantage, which is known in the literature as the vanishing to the ARIMA model. The use of deep learning techniques gradient problem. This problem refers to the difficulty of has allowed excellent results in classification and regression training these networks with methods based on gradients problems to be obtained. This is due to its automatic adjust- or backpropagation algorithms. This is due to the fact that ment of internal parameters by means of supervised learning when this type of method calculates the weight adjustment algorithms. Research has focused on describing vari- based on the chain rule, there are common multiples between ous artificial intelligence (AI) techniques used in renewable 0 and 1 that are multiplied by n times. Numbers less than energy for their prediction and optimization. Existing 1multiplying by n times, an exponential decrease in the error artificial intelligence tools have been used for the model- signal is generated, which significantly affected the training ing and simulation processes of solar energy systems. of these RNNs. Due to the above drawbacks, Long Short This study examines the importance and benefits of artificial Term Memory (LSTM)-type RNNs emerged in 1997 as a neural networks in predicting environmental variables and solution due to their memory units in the network cells. estimating some unconventional energy systems, as well as The prediction of solar radiation is a fundamental key to describing the main methodologies and neural models used increasing the supply of electrical energy generated from this to date. Recently, LSTM has become one of the most medium to the distribution networks. In the energy market, widely used artificial neural networks for short-term wind when an electric energy producer does not comply with the speed predictions ,. The LSTM has been proposed programmed offer, they are penalized with a proportional to make forecasts of wind speed on 24-hour wind farms. relationship between the energy actually produced and what This work compares the results obtained by LSTM with is stated in the contract. The integration of these renewable MLP, deep versions of MLP, and classical methods. LSTM energy sources intensifies the complexity of managing the showed better efficiency prediction results than MLP and grid and the ongoing balance between electricity consump- classical methods. Neural networks-based deep learning tion and production due to its unpredictable and intermittent are widely used in solar radiation applications. Evaluating nature. Several tools and methodologies have been developed the performance of the multilayer perceptron (MLP) and the for the prediction of solar energy at different horizons. In this empowered decision trees by combining them with linear study, the task of predicting the photovoltaic power for the regression for the estimation of solar energy; the results show coming days in short time intervals (30 minutes) from data that the MLP model had a better performance according to previously recorded during one year was considered using the coefficient of determination indicators (R2 ) which was Long Short Term Memory (LSTM) Recurrent Neural Net- 97.7% and the RMSE, which was 0.033. A long-term works (RNN). This is a type of (ANN) that estimates the next prediction approach was used with the MLP network and data from the data received in the past and the current input iterative strategy; however, their results show that the support data and compares it using a multi-layer perceptron (MLP) vector machine (SVM) model with a straightforward strategy neural network, which is a neural network that has limita- yields better results. LSTMs have the characteristic of tions in performance because it learns only using the one- being able to process large amounts of data and have the to-one relationship between input and output and does not ability to generalize; that is, they adapt to unknown data, consider the characteristics of time series data. The results which makes them get better results than SVM-based mod- of each scheme are evaluated using: Mean Absolute Error els. The Long-Term Memory Recurrent Neural Network (MAE), Mean Absolute Percentage Error (MAPE), Root (LSTM-RNN) has been proposed to accurately predict the Mean Squared Error (RMSE), and Coefficient of Determi- output power of solar PV systems using hourly data sets nation (r 2 ). A preliminary study of the data was performed for a year and has compared the proposed method results to find patterns and apply certain corrections, such as elim- with multiple linear regression (MLR), bagged regression inating missing data by averaging the previous 60 minutes’ trees (BRT), and neural networks (NNs) methods for pho- values (average 12 previous values). The remainder of the toelectric prediction. The LSTM networks prove less pre- paper is organized as follows: Section 2 provides a review of dictive error compared to other methods. Long short previous works on the topic, Section 3 provides an explana- term memory (LSTM) neural networks were used to predict tion of the various deep learning techniques based on LSTM solar power panels and wind power in the medium and long and MLP, Section 4 presents and discusses the results, and term. The errors obtained were lower than those of the per- Section 5 concludes the paper with a discussion of future sistence model and the SVM model. As an alternative suggestions. to the conventional approaches, ANNs have been success- fully applied to estimating solar irradiance. The multilayer II. LITERATURE REVIEW perceptron structure (MLP) is the most common neural net- Artificial Neural Networks (ANNs) are of great importance work. An MLP method has been proposed to predict in contemporary research on the mathematical modeling of solar radiation for the next 24 hours using current data of systems based on renewable energies. It reported higher mean solar radiation and air temperature for a region in solar forecast results than those obtained using traditional Italy. VOLUME 10, 2022 31693  M. Elsaraiti, A. Merabet: Solar Power Forecasting Using Deep Learning Techniques III. METHODOLOGY In this section, the sources of information and data processing methods are described, and the building of deep learning- based neural networks to predict solar radiation is presented. A. SOURCES OF INFORMATION In this study, the MATLAB software (R2019b) was used for the training process of the LSTM, which is an advanced architecture for RNN to predict the values of future time steps of a sequence. The sequence regression network was trained to the LSTM sequence, where responses are training sequences with changing values in one time step. That is, FIGURE 1. Shows the original photoelectric data. for each time step of the input sequence, the LSTM network learns to predict the value of the next time step. To evaluate the effectiveness of the proposed method, a case study was performed using a data set obtained from Nova Scotia Com- munity College in Halifax that includes one-year data (from January 1, 2017, to December 31, 2017) for Halifax, located in Nova Scotia, Canada. For each day, data was selected only during daylight hours, from 8 am to 5 pm. The original photo- electric data were collected at 5-minute intervals and included 365 × 120 = 43800 measurements. Each missing value is FIGURE 2. Schematic representation of neurons. processed by averaging the previous 60 min values. Data is summarized at 30-minute intervals as the task is to make a highly connected to each other, to work together to solve a forecast for each half-hour for the next day. Thus, we have problem. Deep learning techniques can be used in forecasting 20 values in one day and 20 × 365 × 1 = 7300 values in one and identifying characteristics of each one; among them: year. The data is then normalized to the interval [0-1]. Upon stages implemented for the approximation, the accuracy with completion of the forecasting process, the results obtained for which the power generation is approximated, the convergence the proposed models are compared with the actual value. The time and the uncertainty associated with the forecast. In the fact that the observation recording period is 5 minutes. Thus, case of neural networks, studies have demonstrated the ability within one hour, you will have 12 observations. In turn, 188 of this technique to accurately determine the time series of (12 × 24) values. The number of past values of the series data. that are considered to forecast the target vector is independent of the size of the target vector and will depend in each case C. MULTILAYER PERCEPTRONS (MLP) on the algorithm used and the nature of the problem. In this The Multilayer Perceptron (MLP) network is the most popu- strategy, a single execution of the algorithm, with the passed lar ANN architecture used in solving scientific problems , data that you need, results in the forecast with the necessary due to its demonstrated ability to approximate non-linear rela- horizon. Figure 1 presents the original photoelectric data tionships. They are a type of ANN capable of modeling during daylight hours, from 8 am to 5 pm with an interval complex functions. They are apt to ignore noise and irrelevant of 5 minutes. inputs, and they can easily adapt their weights. They are also easy to use. The learning process of an MLP can be divided B. PREDICTION TECHNIQUES FOR SOLAR POWER into two main phases: the input of data through the inputs FORECASTING of the MLP and the correction of the errors, during which Artificial Neural Networks (ANN) are part of the area of the errors are calculated by comparing the real data against knowledge of Artificial Intelligence (AI) and Deep Learning, the answer that it delivers to the model through a technique simulated through computer programs, to mimic the human known as Backpropagation. This iteration is repeated multi- capacity to learn, memorize, and find relationships. ANNs, ple times to reduce the error, using an algorithm to obtain a in particular, attempt to reproduce the behavior of biological result with better aptitude, being the Bayesian Regularization, neural networks in an extremely simple way. The ability one of the most common algorithms. A multilayer per- to learn nonlinear relationships and their ability to model ceptron is made up of an input layer, an output layer, and one complex systems have made them a useful tool in different or more hidden layers; although it has been shown that for scientific fields ,. The basic unit of ANNs is the arti- most problems, a single hidden layer will suffice ,. ficial neuron, which is a simplified mathematical abstraction In figure 2, a neuron j is represented, where xi are the inputs, of the behavior of a biological neuron. ANNs are made up wij are the weights that relate each input i with the neuron j, of a large number of artificial neurons grouped in layers and and yj is the output. 31694 VOLUME 10, 2022 M. Elsaraiti, A. Merabet: Solar Power Forecasting Using Deep Learning Techniques The neuron performs two types of operations: first, the fluctuations, and random noise. LSTM structures are dis- propagation rule; and later, the activation function. The prop- tinguished by their ability to model and recognize temporal agation rule (Z ) is defined by the inputs and the synaptic patterns from data sequences, with the presence of memory weights. The most commonly used is the sum of the product cells and the way information is transferred between their of the inputs xi by the weights wij that join them to the neuron units. In this way, they are able to process the sequences and j. This operation represents a linear function that passes by records of the available operational data, thus extracting tem- way of the origin. To remove this limitation, a parameter porary information that allows reducing forecast errors. called threshold bj is added. This can be considered one more In general, past events influence future events, and based on input, with a fixed value of 1, whose weight bj must also be this idea, the recurrent neural network has a structure where updated. The propagation rule is calculated as information from the previous step is passed to the next step Xn and used for estimation. Accordingly, the recurrent neural Zj = (xi.wij + bj ) (1) network has achieved great results in estimating sequential i=1 information, that is, time series data. However, as the length where, Zj is the result of the propagation rule applied to of the past data, required for data estimation, increases, the neuron j, xi is the input vector i, wij is the weight that joins vanishing gradient problem occurs in the existing recurrent input i with neuron j, and bj is the threshold associated with neural network. This problem can be solved by using the gates neuron j. of the LSTM algorithm as proposed in ,. The networks The activation function (A) is responsible for evaluating the of Long Short-Term Memory (LSTM) are very similar to the activation of the neuron and obtaining the output of neuron j. MLP structure. They have input layers, hidden layers, and It is determined based on the result of the propagation rule an output layer. However, LSTM in its hidden layer has a such as memory unit. LSTMs are an ANN type that is classified as Recurrent Neural Networks, which are characterized by
Aj = f Zj (2) not being strictly fed forward, such as MLPs. This is because where, Aj is the activation of neuron j and f is the activation LSTMs use inputs from previous iterations for future output function. calculations, thus providing feedback. This type of ANN is The most common activation functions are step, linear, potentially more powerful than MLP, and they are character- and sigmoidal functions, among which the logistic functions ized by showing a temporary behavior. The memory unit and hyperbolic tangents stand out. Depending on the type of consists of three gates: input gate, forget gate, and output gate problem to be solved, one type of response or another will be (Input gate (it ), Forget gate (ft ), and Output gate (ot )) and required. For classification problems, where binary outputs a recurring connection as shown in Figure 3. The drive has are desired, activation functions of the sigmoid type are often one input xt and two drive previous state feedbacks which are used. These types of functions have a range of small values the previous state output st−1 and the state variable ct−1. The with saturation at the extremes. The most used functions are gates use a sigmoid activation function g, while states use a the logistic function, with a working range on the ordinate tanh function. The memory unit of the LSTM can be defined from 0 to 1, and the hyperbolic tangent, with a working by a set of equations, where w are the parameters and b are range on the ordinate between −1 and 1. On the other hand, the biases. to solve regression problems, they usually use functions of a In general, the LSTM structure consists of three different linear type, since more variability in the response is needed. layers: forget, input and output. In the LSTM architecture, A neural network is organized into layers connected to each first, xt and st−1 information are used as inputs, and it is other creating a neural network. A layer is understood to decided which information to delete. These operations are be the set of neurons that are located at the same level in done in the forget layer ft determined by the network and that process information at the same time.
In addition to the activation function of each neuron, the ft = g wxf xt + whf st−1 + bf (3) behavior of the neural network depends on the topology and the training carried out to establish the value of the weights. where, g is the activation function, which is the sigmoid in this work. D. LONG SHORT TERM MEMORY (LSTM) In the second step, the input layer it , where new informa- Due to the rapid growth of the field of deep learning, which tion will be determined, is given by has manifested itself in the development of new technologies and architectures as well as greater integration in many areas it = g (wxi xt + whi st−1 + bi ) (4) of research, structures such as long-term memory (LSTM) have been recently used in the development of new solar fore- Then, the candidate information, that will form the new infor- casting techniques. LSTM networks are currently one mation, is expressed by of the most popular models in deep learning applied to time series prediction. This kind of prediction is a difficult problem ct = ft.ct−1 + it.it t (5) due to the presence of long-term trends, seasonal and cyclical it t = tanh (wxc xt + whc st−1 + bi ti ) (6) VOLUME 10, 2022 31695  M. Elsaraiti, A. Merabet: Solar Power Forecasting Using Deep Learning Techniques FIGURE 3. Diagram of operation of a cell of the long short-term memory networks. Finally, the output data is obtained by using the following expressions in the output layer. FIGURE 4. Shows one day 30 minutes ahead (December 31). ot = g (wxo xt + who st−1 + b0 ) (7) TABLE 1. Performance evaluation for the winter day. st = ot.tanh (ct ) (8) The process described above continues iteratively. Weight parameters (w) and bias parameters (b) are learned by the model in a way that minimizes the difference between actual TABLE 2. Performance evaluation for the summer day. training values and LSTM output values ,. E. PERFORMANCE EVALUATION Four different statistical evaluation criteria were used to evaluate the prediction performance of the proposed LSTM model. These criteria are: IV. RESULTS AND DISCUSSIONS 1. Mean absolute error (MAE) In order to implement the proposed model, 43,800 records It expresses the mean absolute deviation of the difference were taken during daylight hours with a 5-minute solar inter- between the predicted values and the actual values and is val in Halifax, corresponding to the days from January 1, calculated by 2017 to December 31, 2017 available at the Nova Scotia Community College (NSCC) and to find the missing values. 1 XN A prediction was made every 30 minutes for a winter day MAE = |oi − pi | (9) N i=1 (December 31), and a summer day (June 30) for the total of 2. Root mean square error (RMSE) 7300 values, which were divided into 7280 training data, and It represents the standard deviation of the estimation errors the last 20 for the prediction test. To evaluate the effectiveness and is calculated by of the proposed method, MATLAB R2019b was used for s PN a LSTM training process with an initial learning rate of 2 i=1 (oi − pi ) 0.01 and a maximum epoch of 1000. A case study was car- RMSE = (10) N ried out using original photovoltaic data for one year during daylight hours, from 8 am to 5 pm at 5-minute intervals. 3. Mean absolute percentage error (MAPE) In this work, data is summarized at 30-minute intervals to It measures the prediction accuracy of the models as a make a half-hour forecast for the next day. In Figures 4 and 5, percentage and is calculated by the actual graphs of daily power generation are shown in 1 XN oi − pi comparison with the graphs expected in the forecast models. MAPE = (11) N i=1 oi Figure 4 shows the results for one day in December, and figure 5 shows the results for one day in June. The results 4. Coefficient of determination (R2 ) show that the forecast is close to the actual data, especially It represents the strength of the linear relationship between during the winter period when the system did not generate the predicted values of the models and the actual values and much power, and in the summer, the LSTM model has a is calculated by significant advantage over the MLP model. PN (oi − pi )2 Tables 1 and 2 show the performance criteria for the two 2 R = 1 − Pi=1 N (12) models, for the winter day and the summer day, respectively. i=1 (oi − õ) 2 In general, when the results are examined, it has been shown where, N is the number of samples used for statistical evalu- that the LSTM model gives the best results in all performance ation criteria, oi is the actual value of the observation, and pi criteria on both days. is the forecasted value of the observation and õ is the average The forecasted values of the proposed LSTM model used of the actual observation values. to predict the solar power and the time graph showing the 31696 VOLUME 10, 2022 M. Elsaraiti, A. Merabet: Solar Power Forecasting Using Deep Learning Techniques factors of changes in the atmosphere, such as water vapor, clouds, or pollution. The results show the importance of the time series used to train the models. The quality of the data has a significant impact on the performance obtained by the forecasting model, especially the outliers. A specific time series may perform better with one model, while the same model may perform poorly with another time series model. Consequently, it is convenient to consider multiple models to search for suitable predictions. However, there are several techniques for time series forecasting that can be implemented according to the methodology proposed in this study. FIGURE 5. Shows one day 30 minutes ahead (June 30). V. CONCLUSION The LSTM model, which is based on a deep learning approach, is proposed to predict daily solar power values. The prediction was evaluated using solar power data from Halifax, Nova Scotia, Canada over the course of a year (January 1, 2017, to December 31, 2017). This data was split into two categories: training and test sets. While only the training data was used in the learning process of the FIGURE 6. Prediction result of June (30 minutes interval). model, the test data was not used in the learning process. The task of predicting photovoltaic power for the coming days at 30-minute intervals was considered. The results of the suggested model were compared using the Multi-Layer Perceptron (MLP) algorithm, which is the most extensively used technique in the literature, in order to evaluate and exam- ine their correctness and performance. When compared to the MLP algorithm, the prediction performance of the suggested LSTM model offered more effective values in all performance parameters MAE, MAPE, RMSE, and R2 for each category of days, according to the findings of simulation tests and the results presented in Tables 1 and 2. The proposed model gives trustworthy data that will allow photovoltaic power plants to operate more efficiently in the future. The combination formed by the concepts of artificial intelligence and energy efficiency appears to have a bright future, particularly in terms of boosting energy sustainability, decarburization, and electrical sector digitization. The importance of data process- FIGURE 7. Forecast results in June compared to actual values and RMSE. ing (time series) utilized to train models was highlighted in this study. A specific time series may perform better with one actual values are shown in Figure 6. The actual 30 minute model, while the same model may have poorer performance ahead values for the June data test and the LSTM forecast with another time series model. This reason suggests focusing value are used in the chart. When the forecasted values of on research and development in multiple models in order to the LSTM model are compared with the actual values, the arrive at predictions with high suitability. In addition, the observed consistency in all training epochs is very close to preprocessing approaches used to reduce noise, eliminate the real data; however, there are no very sharp fluctuations. outliers, and reduce errors from prediction models should be It is also reported that the forecasted and actual values show taken into account. When used to predict solar energy, artifi- the same trend. This means the LSTM algorithm rarely under- cial intelligence algorithms have demonstrated their aptitude goes gradient exploding or a vanishing gradient. and superiority in obtaining favorable outcomes. However, Figure 7 shows the results of the LSTM solar forecast obtaining such results would necessitate a significant amount models for the month of June 2017. The graph represents of hyperparameter adjustment. Furthermore, the quality of the daily average solar energy during daylight hours from the data, particularly the outliers, has a substantial impact 8 am to 5 pm. Although the graph shows that the prediction on the prediction model’s performance. Several time series is close to reality, solar energy is not only affected by the prediction approaches can be implemented using the method- parameters of this model. However, it differs due to various ology provided in this work. VOLUME 10, 2022 31697  M. Elsaraiti, A. Merabet: Solar Power Forecasting Using Deep Learning Techniques REFERENCES M. Hagan, M. Beale, and H. Demuth, Neural Network Design. Boston, D. Anderson and M. Leach, ‘‘Harvesting and redistributing renewable MA, USA: PWS Publishing Company, 1996. energy: On the role of gas and electricity grids to overcome intermittency A. K. Yadav and S. S. Chandel, ‘‘Solar radiation prediction using artifi- through the generation and storage of hydrogen,’’ Energy Policy, vol. 32, cial neural network techniques: A review,’’ Renew. Sustain. Energy Rev., no. 14, pp. 1603–1614, Sep. 2004. vol. 33, pp. 772–781, May 2014. C. Paoli, C. Voyant, M. Muselli, and M.-L. Nivet, ‘‘Forecasting of pre- T. Rojat, R. Puget, D. Filliat, J. Del Ser, R. Gelin, and N. Díaz-Rodríguez, processed daily solar radiation time series using neural networks,’’ Sol. ‘‘Explainable artificial intelligence (XAI) on TimeSeries data: A survey,’’ Energy, vol. 84, no. 12, pp. 2146–2160, Dec. 2010. 2021, arXiv:2104.00950. F. S. Tymvios, S. C. Michaelides, and C. S. Skouteli, ‘‘Estimation of G. Briganti and O. Le Moine, ‘‘Artificial intelligence in medicine: Today surface solar radiation with artificial neural networks,’’ in Modeling and tomorrow,’’ Frontiers Med., vol. 7, p. 27, Feb. 2020. Solar Radiation at the Earth’s Surface. Berlin, Germany: Springer, 2008, J. Bedi and D. Toshniwal, ‘‘Deep learning framework to forecast electricity pp. 221–256. demand,’’ Appl. Energy, vol. 238, pp. 1312–1326, Mar. 2019. R. Reed and R. J. Marks, Neural Smithing: Supervised Learning in Feed- L. Alzubaidi, J. Zhang, A. J. Humaidi, A. Al-Dujaili, Y. Duan, forward Artificial Neural Networks. Cambridge, MA, USA: MIT Press, O. Al-Shamma, J. Santamaría, M. A. Fadhel, M. Al-Amidie, and Feb. 1999. L. Farhan, ‘‘Review of deep learning: Concepts, CNN architectures, R. Velo, P. López, and F. Maseda, ‘‘Wind speed estimation using multilayer challenges, applications, future directions,’’ J. Big Data, vol. 8, no. 1, perceptron,’’ Energy Convers. Manage., vol. 81, pp. 1–9, May 2014. pp. 1–74, Dec. 2021. K.-I. Funahashi, ‘‘On the approximate realization of continuous mappings S. Hochreiter, ‘‘The vanishing gradient problem during learning recurrent by neural networks,’’ Neural Netw., vol. 2, no. 3, pp. 183–192, Jan. 1989. neural nets and problem solutions,’’ Int. J. Uncertainty, Fuzziness Knowl.- J. H. Byrne, ‘‘Introduction to neurons and neuronal networks,’’ in Text- Based Syst., vol. 6, no. 2, pp. 107–116, Apr. 1998. book for the Neurosciences, 2013. [Online]. Available: https://nba.uth. S. Hochreiter and J. Schmidhuber, ‘‘Long short-term memory,’’ Neu- tmc.edu/neuroscience/s1/introduction ral Comput., vol. 9, no. 8, pp. 1735–1780, 1997. [Online]. Available: S. Srivastava and S. Lessmann, ‘‘A comparative study of LSTM neural net- https://deeplearning.cs.cmu.edu/pdfs/Hochreiter97-lstm.pdf works in forecasting day-ahead global horizontal irradiance with satellite S. A. Kalogirou, ‘‘Artificial neural networks in renewable energy sys- data,’’ Sol. Energy, vol. 162, pp. 232–247, Mar. 2018. tems applications: A review,’’ Renew. Sustain. Energy Rev., vol. 5, no. 4, S. Muzaffar and A. Afshari, ‘‘Short-term load forecasts using LSTM pp. 373–401, Dec. 2001. networks,’’ Energy Proc., vol. 158, pp. 2922–2927, Feb. 2019. M. Elsaraiti and A. Merabet, ‘‘A comparative analysis of the ARIMA F. A. Gers, N. N. Schraudolph, and J. Schmidhuber, ‘‘Learning precise and LSTM predictive models and their effectiveness for predicting timing with LSTM recurrent networks,’’ J. Mach. Learn. Res., vol. 3, no. 1, wind speed,’’ Energies, vol. 14, no. 20, p. 6782, Oct. 2021, doi: pp. 115–143, 2003. 10.3390/en14206782. A. Sagheer and M. Kotb, ‘‘Time series forecasting of petroleum produc- S. R. Young, D. C. Rose, T. P. Karnowski, S.-H. Lim, and R. M. Patton, tion using deep LSTM recurrent networks,’’ Neurocomputing, vol. 323, ‘‘Optimizing deep learning hyper-parameters through an evolutionary pp. 203–213, Jan. 2019. algorithm,’’ in Proc. Workshop Mach. Learn. High-Perform. Comput. Env- T. Fischer and C. Krauss, ‘‘Deep learning with long short-term memory iron., Nov. 2015, pp. 1–5. networks for financial market predictions,’’ Eur. J. Oper. Res., vol. 270, M. Q. Raza and A. Khosravi, ‘‘A review on artificial intelligence based no. 2, pp. 654–669, Oct. 2018. load demand forecasting techniques for smart grid and buildings,’’ Renew. Sustain. Energy Rev., vol. 50, pp. 1352–1372, Oct. 2015. S. A. Kalogirou, ‘‘Solar thermal power systems,’’ in Solar Energy Engi- neering, 2009. MEFTAH ELSARAITI (Member, IEEE) received A. Azadeh, R. Babazadeh, and S. M. Asadzadeh, ‘‘Optimum estimation the B.S. degree in physics from Tripoli Uni- and forecasting of renewable energy consumption by artificial neural versity, Tripoli, Libya, in 1984, and the M.S. networks,’’ Renew. Sustain. Energy Rev., vol. 27, pp. 605–612, Nov. 2013. degree in mechanical engineering from Bradford M. Elsaraiti and A. Merabet, ‘‘Application of long-short-term-memory University, Bradford, U.K., in 2005. He is cur- recurrent neural networks to forecast wind speed,’’ Appl. Sci., vol. 11, no. 5, rently pursuing the Ph.D. degree in applied science p. 2387, Mar. 2021. with Saint Mary’s University, Halifax, Canada. J. Bottieau, F. Vallee, Z. De Greve, and J.-F. Toubeau, ‘‘Leveraging provi- From 1985 to 1990, he was a Research Assistant sion of frequency regulation services from wind generation by improving with the Iron and Steel Training Center, Misurata, day-ahead predictions using LSTM neural networks,’’ in Proc. IEEE Int. Libya, and from 2005 to 2013, he was a Lecturer Energy Conf. (ENERGYCON), Limassol, Cyprus, Jun. 2018, pp. 1–6. with the Higher Institute for Technical Sciences Misurata, Libya. His cur- A. Rabehi, M. Guermoui, and D. Lalmi, ‘‘Hybrid models for global solar rent research interest includes developing models for predicting renewable radiation prediction: A case study,’’ Int. J. Ambient Energy, vol. 41, no. 1, energies. pp. 31–40, Jan. 2020, doi: 10.1080/01430750.2018.1443498. E. Pasero, G. Raimondo, and S. Ruffa, ‘‘MULP: A multi-layer perceptron application to long-term, out-of-sample time series prediction,’’ in Proc. Int. Symp. Neural Netw., Jun. 2010, pp. 566–575. ADEL MERABET (Senior Member, IEEE) Q. Xiaoyun, K. Xiaoning, Z. Chao, J. Shuai, and M. Xiuda, ‘‘Short- received the Ph.D. degree in engineering from term prediction of wind power based on deep long short-term memory,’’ the University of Québec at Chicoutimi, Canada, in Proc. IEEE PES Asia–Pacific Power Energy Eng. Conf., Oct. 2016, in 2007. He joined Saint Mary’s University, pp. 1148–1152. Halifax, Canada, in 2009. He is currently an Asso- M. Abdel-Nasser and K. Mahmoud, ‘‘Accurate photovoltaic power fore- ciate Professor with the Division of Engineering. casting models using deep LSTM-RNN,’’ Neural Comput. Appl., vol. 31, From 2016 to 2017, he was a Visiting Academic no. 7, pp. 2727–2740, Jul. 2019. with the Department of Sustainable and Renew- S. Han, Y.-H. Qiao, J. Yan, Y.-Q. Liu, L. Li, and Z. Wang, ‘‘Mid-to-long able Energy Engineering, University of Sharjah, term wind and photovoltaic power generation prediction based on copula Sharjah, United Arab Emirates. In 2019, he was a function and long short term memory network,’’ Appl. Energy, vol. 239, pp. 181–191, Apr. 2019. recipient of the Tan Chin Tuan Exchange Fellowship in Engineering with M. Diagne, M. David, P. Lauret, J. Boland, and N. Schmutz, ‘‘Review Nanyang Technological University in Singapore. In 2021, he was a Visiting of solar irradiance forecasting methods and a proposition for small-scale Research Scientist with the Advanced Power and Energy Center, EECS insular grids,’’ Renew. Sustain. Energy Rev., vol. 27, pp. 65–76, Nov. 2013. Department, Khalifa University, Abu Dhabi, United Arab Emirates. His A. Mellit and A. M. Pavan, ‘‘A 24-h forecast of solar irradiance using research interests include renewable (wind-solar) energy conversion systems, artificial neural network: Application for performance prediction of a energy management, advanced control, electric drives, artificial intelligence, grid-connected PV plant at Trieste, Italy,’’ Sol. Energy, vol. 84, no. 5, and smart grid. pp. 807–821, May 2010. 31698 VOLUME 10, 2022