AN OVERVIEW OF ARTIFICIAL INTELLIGENCE TECHNIQUES
FOR EFFICIENT LOAD FORECASTING
Anamika Singh and Manish Kumar Srivastava
Department of Electrical Engineering, Sam Higginbottom University of Agriculture, Technology and Sciences, India E-Mail:[email protected]
ABSTRACT
The need for electricity demand forecasting has become a significant aspect as well as core part in facility expansion and planning periodical operations specifically in the electricity sector. In addition, demand pattern is significantly complex in nature as the energy market is relatively deregulated. Hence, analysing a significant forecasting model that would perform a specific task is highly complex and critical in nature. It has been analysed that several forecasting methods have been developed in past and presented in the literature; however, none of them can be used for a generalized model for all demand patterns. Furthermore, several factors influence electric load data including economic factors, time factors, weather factors, etc which increases the need for Artificial Intelligence techniques in the field of load forecasting. In addition, it is essential to analyse the irrelevant factors associated with electric load forecasting which needs to be eliminated. This, in turn, increased the need for Artificial Intelligence techniques for minimizing estimation error and enhances the accuracy of forecasting. Artificial Neural Network is one of the most important contributions in the field of load forecasting as a core technique of Artificial Intelligence. In this paper, the review of different techniques based on classical as well as AI techniques would be analysed for electric load forecasting. In addition, the explanation of load forecasting has been also provided in a comprehensive way.
Keywords: artificial intelligence, electric load forecasting, neural network, power system planning.
INTRODUCTION
Load forecasting is one of the most significant areas in the field of electrical engineering. In addition, it is one of the most traditional forecasting models. This research paper intends to integrate artificial intelligence techniques for load forecasting. With the advent of technology, Artificial Intelligence techniques and neural networks have gained significant attention as well as there are several research papers focused on successful experiments based on such implementation [1]. The aim of this study is to offer a new dimension in the field of electrical engineering by integrating Artificial Intelligence techniques for load forecasting. The rationale of the study is to analyse load forecasting techniques as well as its planning and applications with its significant types in order to maintain accuracy. The purpose of the study is to analyse the success rate of load forecasting using AI techniques. In addition, several research studies stated that the advantage of using AI techniques for load forecasting has not been proved yet and remain sceptical about it. This research study would fill such a gap in literature offering the significance of AI techniques in load forecasting with justification and evidence.
Based on such significance of AI techniques in load forecasting, this paper highlights each essential element that would fulfil the criteria. The second part depicts an explanation of load forecasting with its applications and types. In addition, measures of predicting accuracy for load forecasting with formula have been also presented along with other essential aspects. The third part highlights the classification of load forecasting techniques based on n classical and AI-based models. The fourth part depicts the significance of the ANN approach. The fifth part highlights the various networks with diagrams. The sixth part provides a description of the future scope based
on this paper. Lastly, the conclusion and summary have been provided.
DISCUSSIONS
Load forecasting
Load forecasting is a significant part of electric energy generation, markets, transmission as well as distribution [2]. Load forecasting has a significant importance in the planning of the electricity industry as well as the operation of the electric power system. Furthermore, appropriate forecasts can significantly lead to savings in maintenance cost as well as operational cost [3]. It also enhances the reliability of power supply and delivery system which lead to appropriate decisions for development in future decisions. As there are several factors that impact electric load data, this, in turn, enhances the need for machine learning methods such as AI techniques for load forecasting [4]. In addition, the use of AI techniques for minimizing errors in load forecasting has gained significant importance in present days. Load forecasting is an essential component of the power system as well as energy management system [5].
Importance of load forecasting in power system planning
forecast depicts future and current trends in the power system, the appropriateness and accuracy of the forecast are significant towards every electric utility. This, in turn, increases the lead of accurate load forecasting in the planning of a utility company [8]. Hence, the need for load forecasting in power system planning is vital as it reduces the number of errors. Furthermore, load forecasting predicts future load with the assistance of available historical load data which is significant for the control, operation as well as planning of power system [9].
Load forecasting applications
The applications of load forecasting integrate infrastructure development, energy purchasing, and generation, contract evaluation as well as load switching [10]. Furthermore, it assists in accurate prediction of the load which is one of the primary indicators of power intelligence. In addition, it assists to take inputs based on loads, data focusing on weather data, type of the day as well as the time of the day. This, in turn, assists to attain a systematic approach of electric distribution management making forecasting area up-to-date [11]. There has been always a chance of uncertain load jumps which critically affects the electricity system. Based on such assumption, the load forecasting can assist to minimize such uncertain incidents with significant analysis of past trends and data [12]. Furthermore, it assists electricity companies to retain elevated efficiency of energy as well as reliable operation. Load forecasting also assists to analyse historical data and minimize the error which enhances the efficiency of the system producing a significant increase in prediction quality [13]. Load forecasting can be done through the use of several techniques including expert systems, similar day approach, artificial neural networks, statistical methods, regression models, time series as well as fuzzy logic. On the other hand the application of artificial neural networks has significantly enhanced the forecasting areas by mitigating the limitations that have been analysed in the others forecasting methods. In addition, through the use of such non-linear circuits known as neural networks the capability to formulate non-linear curve fitting have been significantly integrated. Hence, through this review paper the effectiveness of artificial neural network has been analysed that would assist to formulate significant scope in future studies.
Load forecasting types
The types of load forecasting are significant in order to attain reliable operation as well as energy efficiency [14]. Long-term electric load forecasting assists electric utility company management to supply forecasts of future needs. In addition, it assists to analyse staff hiring, expansion and equipment purchases for successful electric management. Based on long-term load forecasting, the duration for the prediction of an electric load is 3 year to 50 years [15]. On the other hand, medium-term electric load forecasting assists the electric utility company to schedule unit maintenance and fuel supplies efficiently. The duration for such schedule prediction ranges from one
forecasting, its assistance to the electric utility company is relatively to provide significant information for the system management [16]. This in turn assists to conduct operations on regular basis as well as unit commitment. Short-term load forecasting is used for day to day operations and unit commitment. It assists in security functions as well as online scheduling of an energy management system. Lastly, very short-term electric load forecasting integrated a few minutes to a maximum of an hour for forecasting of electric loads [17].
Factors affecting load forecasting
The time factor is one of the significant factors that affect load forecasting critically as its impact on the customer’s load is relatively high [18]. Furthermore, the increased use of electricity by people in daily life has been critically affected by an economic factor as well. Specifically, in long-term and short-term load forecasting, the impact of an economic factor is relatively high as it depends on the degree of industrialization, the price of electricity as well as management of load [19]. Weather factor has also a significant impact on the load forecasting is an independent variable. In addition, the impact of weather is critical on agricultural and domestic consumers. Models for minimizing the cost of load forecasting are generally used for such a situation. Backdoor fronts, sea breeze, afternoon thunderstorms decrease the temperature creating overestimated load forecast [20].
The weather factor integrated temperature, occasional or random spikes and humidity. Alternation of the transmission lines is impacted due to temperature. Hence, the temperature can critically impact on the capability of transmission lines [21]. Furthermore, a higher temperature can not only amplify the resistance of the transmission lines but also alter the reactance of the line as the expansion of the length of transmission lines occur during temperature change. Humidity intensifies the severity of hot climate that critically affects the daily load of domestic consumers. However, it does not have any impact on real temperature [22].
Ideal load forecasting model
Ideal forecasting model requires strategic concept in smart electricity grids. In addition, the analysis of active demand is significant to forecast demand response scenario. The analysis of load behaviour is significant for an ideal load forecasting model based on algorithms as well as considering the effect of active demand [21]. The impact of external factors including population, weather and time must be reduced at a minimum level in order to predict future demand at ease. Furthermore, integration of AI techniques in load forecasting model has been significantly important in order to forecast electric load accurately without any error.
Measures of prediction accuracy for load forecasting with formula
analysing both components, the accuracy for predicting load forecasting can be attained successfully.
MAPE = 1 𝑛∑
⃒𝐴𝑐𝑡𝑢𝑎𝑙 𝐿𝑜𝑎𝑑−𝐹𝑜𝑟𝑒𝑐𝑎𝑠𝑡𝑒𝑑 𝐿𝑜𝑎𝑑⃒
𝐴𝑐𝑡𝑢𝑎𝑙 𝐿𝑜𝑎𝑑 ∗ 100 (1)
In the above equation (1), ‘n’ is the total number of predicted loads to calculate error. Mean Absolute Error is the standard metric that can be used for forecasting load with accuracy [22].
CLASSIFICATION OF LOAD FORECASTING TECHNIQUES
CLASSICAL MODELS
Multiple regressions
This model was first proposed by Pearson in the year 1908 in order to learn about the relationship between predictor variables or several independent variables.It is a technique that evaluates the dependent variable to the specified independent. The dependent variable is the demand of the electric as it depends on production and the independent variables are considered first [23]. It is significantly affected by meteorological effects, economic growth, per capita growth, prices of electricity in order to forecast the load. In addition, the developed model also provides a day-ahead forecast that analysed sensitive load components and weather insensitive components.
Least square
This method was developed by Carl Friedrich Gauss in the year 1809. It is used to analyse model order and parameters. It uses an operator that handles the variable at a time and analyses optimal starting point. In order to analyse the load dynamics, the use of the autocorrelation function is utilized [24].
Stochastic time series
This model was first proposed by Markov Chain in the year 1906. It is a time series method that is based on the data of internal structure including seasonal variation and autocorrelation. It first analysed the matching data available and then uses it to attain the forecasted value based on time [25].
Exponential smoothing model
This model was first proposed by Robert Goodell Brown in 1956 for enhancing the technique to analyse the future load based on the past or historical data. At first the load is modelled in relation to the previous data and later it uses the model for future prediction of load [26].
Moving average models
This model was first proposed in the year 1975. It depicts the present value of the time series linearly in relation to its values at previous periods. In addition, a recursive scheme is utilized for analysing the parameters. It is a time-temperature approach for load forecasting [27].
End-user model
This model was first proposed in the year 1971. It is used for long-term load forecasting. It offers accurate and specific results based on electric load [28]. The general concept of this model is based on the demand of electricity relies upon its use. By analysing historical data of electric consumption in a particular geographic location, this model can analyse the future demand efficiently. Its major advantage is it depicts the exact flow of electricity with significant information and formulates exact electricity demands.
Econometric model
This model was first proposed by Ragnar Frisch in the year 1910. It integrates the economy as it is a driver of long-term electricity demand. In addition, load forecasting requires economy consideration based on people or population [29]. In addition, it uses statistics, economics and mathematics to predict electricity demand. It is a combination of end-use and trend analysis however it does not offer any assumption based on predicting future demand in relation to the historical data. It enables variations and works bets at state, regional or national level.
Time series model
This model was first proposed by G. U Yule and J. Walker in 1920s and 1930s. It works with the modelling patterns in a time series plot and then using the patterns for future use in order to attain reliable predictions [30]. It is a kind of short term load forecasting based on the concept that appropriate predictions can be attained through modelling patterns in the time series plot. Later, those patters would be extrapolated for future. It uses historical information as its input and highly efficient model in relation to seasonality and trend. However, it is complex in nature as it requires huge amount of historical data.
AI-BASED MODELS
Artificial neural network (ANN)
This model was first proposed by Warren McCulloch and Walter Pitts in the year 1943. ANN is a soft technique that can be utilized for performing several optimization processes. It conducts both non-linear adaptation and modelling. In addition, it does not require any weather variables in advance [31]. It is also presently a powerful tool in solving power system problems incorporating topological observance, security assessment, fault diagnosis, harmonic load identification as well as alarm processing.
Fuzzy logic system (FLS)
The input value is in the form of ‘0’ and ‘1’. It assists to compare qualities based on inputs.
SIGNIFICANCE OF ANN APPROACH
Artificial neural networks as proposed by proposed by Warren McCulloch and Walter Pitts in the year 1943 have significant application in load forecasting as their ability to learn is immense. Artificial neural networks provide the potency to overcome the reliability through a functional form [33]. The types of neural networks are self-organizing and multilayer perceptron network. In addition, there are several other hidden layers in the network. The main advantage of ANN is the forecasting methods which do not require any load model. Furthermore, it represents the load as the output of the dynamic system considering environmental variables including weather and time [34]. It also can be used in hourly based load forecasting for predicting the further demand of the load. In addition, the algorithm also integrates supervised learning concept and historical relationship between temperature and load for the provided season, an hour of the day as well as day type. Hence, the significance of the ANN is immense specifically in load forecasting for electric company utility [35].
INTRODUCTION TO VARIOUS NEURAL NETWORKS WITH DIAGRAM
Artificial neural network
This model was first proposed by proposed by Warren McCulloch and Walter Pitts in the year 1943 and later developed by Minsky and Papert in the year 1969. ANN is a significant technique that is used for optimization processes. This method is highly reliable to conduct both adaptation and non-linear modelling. In addition, it must not need assumption of a functional relationship between weather variables as well as load in advance. It can be also exposed to new data. It is also presently a powerful tool in solving power system problems incorporating topological observance, security assessment, fault diagnosis, harmonic load identification as well as alarm processing [36]. Furthermore, it has the potential to overcome functional form in the forecasting model. It connects hidden inputs as well as connects inputs to the output units as linear functions [37].
Figure-1. ANN Topology.
Biological neurons
In general, a biological neuron in load testing receives inputs from other sources. It combines each of the inputs together by performing a nonlinear operation on the output. A biological neuron model is a mathematical representation of the characteristics of certain cells within the nervous system. It generates electrical potentials throughout the cell membrane. Artificial Neural Network assists to stimulate multiple layers through the simple processing elements [38]. In ANN, there are several neurons that are hidden in multiple layers. The inputs are multiplied by weights and added to a threshold in order to formulate an inner product number known as net function [39].
Figure-2. Biological neurons.
Multilayer perceptron neural network
The multilayer perceptron integrates multi layers that are sigmoid processing elements and simple that interacts by utilizing weighted connections. It is one of the trained using several techniques [40]. The training method that has been used for several years is back propagation and its various derivatives. It offers an advanced level of performance and in comprehending the way neural networks operate. It consists of three-layer integrating the input layer, a hidden layer and an output layer [41].
Figure-3. Multiple layers of perceptron neural network.
Types of neural network architecture
machine, mean field theory machine, and multilayer perceptron network. Each hidden layers of the network have several neurons. Each of the neural networks has been developed and analysed significantly [42].
Figure-4. Types of neural network architecture.
Learning types of neural network
Learning of a neural network is a significant aspect of developing significant network architectures [43]. In addition, the learning techniques incorporating back propagation, output weight optimization hidden weight optimization, as well as the full conjugate gradient method, are significant for mandating the balance of neural networks [44]. Learning techniques of the neural network are divided into two types a) Supervised learning and b) unsupervised learning. The supervised learning method is significant for multilayer feed-forward to inculcate large deep learning networks. It requires expected and known value based on its input for calculating loss function gradient. It has significant applications in the practical field specifically in the area of AI. On the other hand, the unsupervised learning technique is a new artificial neural model by considering a two-class pattern recognition problem. Furthermore, it iterates the weights in a way that would shift the decision boundary to a low pattern density.
Error function
Error functions are critical in load testing and a separate hidden unit is utilized for it [45]. In addition, with multiple sets of linear equations assist to analyse the weights associated with the hidden units. When the surface of the error is extremely curved, the estimation would not be valid. These cases are restored by historical calculated networks for reducing the learning factor [46].
Figure-5.Mapping error.
An overview of back propagation ANN model
Back propagation in the ANN model is an algorithm that assists for neural networks learning significantly. In addition, the graphical approach is considered in which the algorithm minimizes the labelling problem of the graph [47]. It manages the case of particular network topologies. However, it is easier to follow. Furthermore, the algorithm can be integrated into eth computing system efficiently with the use of local information [48].
Figure-6. Sigmoid in back propagation of ANN model.
ANN-based optimization techniques with Fuzzy Fuzzy optimization in the ANN model is based on the Boolean logic that is used for the design of the digital circuit. The input value is in the form of 0 and 1 based on the comparison in relation to qualities [49]. In addition, in the training stages, it uses historical data for generating patterns database as a forecaster. After training, it would be linked to load change in online platform [50].
Expert system
The expert system is used for updating function. In addition it significant approaches to unsupervised learning [51]. It assists to modify the forecast to accommodate changes in temperature and weather. In addition, is assist to analyse hourly load foresting as well [52]. It is a new system that has emerged in the field of AI. It is a computer program that has its own knowledge, reason and explains as new information is available to it [53].
Genetic algorithm
It is used to analyse the autoregressive moving average with the integration of the ARMAX model for forecasting load demands [54]. It is an evolutionary programming approach having the capability of converging on the global extreme of the surface of a complex error [55]. It enhances the fitting accuracy of the model by offering a globally optimal solution [56].
Particle swarm optimization mode
FUTURE SCOPE
For future scope, a wide area has been established that needs to be formulated for better scope and research in the fields of load forecasting. The requirement of AI in the field of load forecasting is immense and requires significant research for mitigating uncertain conditions including weather data, economy, time factors, end-use factors as well as customer factors [59]. For future scope, these areas need to be established in order to yield the opportunity offered by AI techniques towards load forecasting [60].
CONCLUSIONS
Hence, the demand forecasting represents the primary task as it requires significant planning for the production of electricity which can be critically delivered by load forecasting. It is the most crucial way of planning for electric networks and companies. This research would forecast the significant of the Artificial intelligence techniques in load forecasting which have been developed in the posted literature in order to analyse its effectiveness. The impact of weather data, economy, time factors, end-use factors, as well as customer factors, are critical towards load forecasting [61]. Hence, it states the criticality of predicting future loads due to such influencing factors. Hence, it is significant to enhance new methods for load forecasting with the integration of AI techniques for eliminating the uncertainty predictions [62]. The significant details of the types of load forecasting and specifically ANN model has been illustrated in this paper that highlights the significance of integration of AI techniques of attaining desired outputs as well as to calculate load accurately during uncertain situations.
ACKNOWLEDGMENT
I would like to express my appreciation to my advisor Dr. Manish Kumar Srivastava, Associate Professor, Department of Electrical Engineering for this work. Without his valuable assistance, this work would not have been completed.
REFERENCES
[1] Ahmad A.S., Hassan M.Y., Abdullah M.P., Rahman H.A., Hussin F., Abdullah H. and Saidur R. 2014. A review on applications of ANN and SVM for building electrical energy consumption forecasting. Renewable and Sustainable Energy Reviews. 33: 102-109.
[2] Anbazhagan S. and Kumarappan N. 2013. Day-ahead deregulated electricity market price forecasting using recurrent neural network. IEEE Systems Journal. 7(4): 866-872.
[3] Anctil F. and Tape D.G. 2004. An exploration of artificial neural network rainfall-runoff forecasting combined with wavelet decomposition. Journal of
Environmental Engineering and Science. 3(S1): S121-S128.
[4] Awan S.M., Aslam M., Khan Z.A. and Saeed H. 2014. An efficient model based on artificial bee colony optimization algorithm with Neural Networks for electric load forecasting. Neural Computing and Applications. 25(7-8): 1967-1978.
[5] Azad H.B., Mekhilef S. and Ganapathy V.G. 2014. Long-term wind speed forecasting and general pattern recognition using neural networks. IEEE Transactions on Sustainable Energy. 5(2): 546-553.
[6] Borges C.E., Penya Y.K. and Fernandez I. 2013. Evaluating combined load forecasting in for construction of energy demand forecasting knowledge-based expert systems. Knowledge-Based Systems. 39, 194-206.
[7] Ceperic E., Ceperic V. and Baric A. 2013. A strategy for short-term load forecasting by support vector regression machines. IEEE Transactions on Power Systems, 28(4): 4356-4364.
[8] Chang W.Y. 2013. Short-term wind power forecasting using the enhanced particle swarm optimization based hybrid method. Energies. 6(9): 4879-4896.
[9] Chaturvedi D.K., Premdayal S.A. and Chandiok A. 2013. Short Term Load Forecasting using Neuro-fuzzy-Wavelet Approach. Int. J. Comput. Acad. Res. 2: 36-48.
[10]ChenS.X., Gooi H.B. and Wang M.Q. 2013. Solar radiation forecast based on fuzzy logic and neural networks. Renewable Energy. 60: 195-201.
[11]Cheng M.Y. and Cao M.T. 2014. Accurately predicting building energy performance using evolutionary multivariate adaptive regression splines. Applied Soft Computing. 22: 178-188.
[12]Chitsaz H., Shaker H., Zareipour H., Wood D. and Amjady N. 2015. Short-term electricity load forecasting of buildings in microgrids. Energy and Buildings. 99: 50-60.
[14]Chou J.S. and Pham A.D. 2015. Smart artificial firefly colony algorithm‐based support vector regression for enhanced forecasting in civil engineering. Computer‐Aided Civil and Infrastructure Engineering. 30(9): 715-732.
[15]Chu Y., Pedro H.T. and Coimbra C.F. 2013. Hybrid intra-hour DNI forecasts with sky image processing enhanced by stochastic learning. Solar Energy. 98: 592-603.
[16]Coelho V.N., Coelho I.M., Coelho B.N., Reis A.J., Enayatifar R., Souza M.J. and Guimarães F.G. 2016. A self-adaptive evolutionary fuzzy model for load forecasting problems on smart grid environment. Applied Energy. 169: 567-584.
[17]Diagne M., David M., Lauret P., Boland J. and Schmutz N. 2013. Review of solar irradiance forecasting methods and a proposition for small-scale insular grids. Renewable and Sustainable Energy Reviews. 27: 65-76.
[18]Fard A.K. and Akbari-Zadeh M.R. 2014. A hybrid method based on wavelet, ANN and ARIMA model for short-term load forecasting. Journal of Experimental & Theoretical Artificial Intelligence. 26(2): 167-182.
[19]Ghanbari A., Kazemi S.M., Mehmanpazir F. and Nakhostin M.M. 2013. A cooperative ant colony optimization-genetic algorithm approach for construction of energy demand forecasting knowledge-based expert systems. Knowledge-Based Systems. 39: 194-206.
[20]Hernandez L., Baladrón C., Aguiar J.M., Carro B., Sanchez-Esguevillas A.J. and Lloret J. 2013. Short-term load forecasting for microgrids based on artificial neural networks. Energies. 6(3): 1385-1408.
[21]Hong T. and Fan S. 2016. Probabilistic electric load forecasting: A tutorial review. International Journal of Forecasting. 32(3): 914-938.
[22]Hong T. and Wang P. 2014. Fuzzy interaction regression for short term load forecasting. Fuzzy optimization and decision making. 13(1): 91-103.
[23]Hong T., Pinson P., Fan S., Zareipour H.,Troccoli A. and Hyndman R.J. 2016. Probabilistic energy forecasting: Global energy forecasting competition 2014 and beyond.
[24]Hu Z., Bao Y. and Xiong T. 2013. Electricity load forecasting using support vector regression with memetic algorithms. The Scientific World Journal.
[25]Humeau S., Wijaya T.K., Vasirani M. and Aberer K. 2013, October. Electricity load forecasting for residential customers: Exploiting aggregation and correlation between households. In Sustainable Internet and ICT for Sustainability (SustainIT). 2013: 1-6.
[26]Ibe O.G. and Onyema A.I. 2013. Concepts of reactive power control and voltage stability methods in power system network. IOSR Journal of Computer Engineering. 11(2): 15-25.
[27]Jaramillo-Morán M.A., González-Romera E. and Carmona-Fernández D. 2013. Monthly electric demand forecasting with neural filters. International Journal of Electrical Power & Energy Systems. 49: 253-263.
[28]Jurado S., Peralta J., Nebot À, Mugica F. and Cortez P. 2013, July. Short-term electric load forecasting using computational intelligence methods. In Fuzzy Systems (FUZZ), IEEE International Conference: 1-8.
[29]Kavousi-Fard A. and Kavousi-Fard F. 2013. A new hybrid correction method for short-term load forecasting based on ARIMA, SVR and CSA. Journal of Experimental & Theoretical Artificial Intelligence. 25(4): 559-574.
[30]Kavousi-Fard A., Samet H. and Marzbani F. 2014. A new hybrid modified firefly algorithm and support vector regression model for accurate short term load forecasting. Expert systems with applications. 41(13): 6047-6056.
[31]Khan A.R., Mahmood A., Safdar A., Khan Z.A. and Khan N.A. 2016. Load forecasting, dynamic pricing and DSM in smart grid: A Review. Renewable and Sustainable Energy Reviews. 54: 1311-1322.
[32]Khatib T., Mohamed A. and Sopian K. 2013. A review of photovoltaic systems size optimization techniques. Renewable and Sustainable Energy Reviews. 22: 454-465.
[34]Kisi O., Shiri J. and Tombul M. 2013. Modeling rainfall-runoff process using soft computing techniques. Computers & Geosciences. 51: 108-117.
[35]Ko C.N. and Lee C.M. 2013. Short-term load forecasting using SVR (support vector regression)-based radial basis function neural network with dual extended Kalman filter. Energy. 49: 413-422.
[36]Kunwar N., Yash K. and Kumar R. 2013. Area-load based pricing in DSM through ANN and heuristic scheduling. IEEE Transactions on Smart Grid. 4(3): 1275-1281.
[37]Lafdani E.K., Nia A.M. and Ahmadi A. 2013. Daily suspended sediment load prediction using artificial neural networks and support vector machines. Journal of Hydrology. 478: 50-62.
[38]Laghari J.A., Mokhlis H., Bakar A.H.A. and Mohamad H. 2013. Application of computational intelligence techniques for load shedding in power systems: A review. Energy conversion and management. 75: 130-140.
[39]Li S., Wang P. and Goel L. 2015. Short-term load forecasting by wavelet transform and evolutionary extreme learning machine. Electric Power Systems Research. 122: 96-103.
[40]Liu J., Fang W., Zhang X. and Yang C. 2015. An improved photovoltaic power forecasting model with the assistance of aerosol index data. IEEE Transactions on Sustainable Energy. 6(2): 434-442.
[41]Liu Q.J., Shi Z.H., Fang N.F., Zhu H.D. and Ai L. 2013. Modeling the daily suspended sediment concentration in a hyper concentrated river on the Loess Plateau, China, using the Wavelet–ANN approach. Geomorphology. 186: 181-190.
[42]Lydia M., Selvakumar A.I., Kumar S.S. and Kumar G.E.P. 2013. Advanced algorithms for wind turbine power curve modeling. IEEE Transactions on sustainable energy. 4(3): 827-835.
[43]Monteiro C., Fernandez-Jimenez L.A., Ramirez-Rosado I.J., Muñoz-Jimenez A. and Lara-Santillan P.M. 2013. Short-term forecasting models for photovoltaic plants: Analytical versus soft-computing techniques. Mathematical problems in engineering.
[44]Moturi C.A. and Kioko F.K., 2013. Use of artificial neural networks for short-term electricity load
forecasting of Kenya national grid power system. International Journal of Computer Applications. 63(2).
[45]Nezzar M.R., Farah N. and Khadir T. 2013, May. Mid-long term Algerian electric load forecasting using regression approach. In Technological Advances in Electrical, Electronics and Computer Engineering (TAEECE), 2013 International Conference: 121-126.
[46]Nourani V., Baghanam A.H., Adamowski J. and Kisi O. 2014. Applications of hybrid wavelet–Artificial Intelligence models in hydrology: A review. Journal of Hydrology. 514, 358-377.
[47]Ogliari E., Grimaccia F., Leva S. and Mussetta M. 2013. Hybrid predictive models for accurate forecasting in PV systems. Energies. 6(4): 1918-1929.
[48]Rahmani R., Yusof R., Seyedmahmoudian M. and Mekhilef S. 2013. Hybrid technique of ant colony and particle swarm optimization for short term wind energy forecasting. Journal of Wind Engineering and Industrial Aerodynamics. 123: 163-170.
[49]Raza M.Q. and Khosravi A. 2015. A review on artificial intelligence based load demand forecasting techniques for smart grid and buildings. Renewable and Sustainable Energy Reviews. 50: 1352-1372.
[50]Roushangar K., Mehrabani F.V. and Shiri J. 2014. Modeling river total bed material load discharge using artificial intelligence approaches (based on conceptual inputs). Journal of hydrology. 514: 114-122.
[51]Sahay K.B. and Tripathi M.M. 2013, December. Day ahead hourly load and price forecast in ISO New England market using ANN. In India Conference (INDICON), 2013 Annual IEEE: 1-6.
[52]Saranya N. and Zoraida B. 2018. Predicting energy consumption of a house using neural network. International Journal of Computer Sciences and Engineering. 6(7): 275-277.
[53]Seo Y., Kim S., Kisi O. and Singh V.P. 2015. Daily water level forecasting using wavelet decomposition and artificial intelligence techniques. Journal of Hydrology. 520: 224-243.
Emerging Trends in Engineering and Development. (3): 174-184.
[55]Shaikh L., Sawlani K. 2017. A rainfall prediction model using artificial neural network. International Journal of Scientific Research in Network Security and Communication. 5(1): 24-28.
[56]Singh A.K., Ibraheem S.K., Muazzam M. and Chaturved, D.K. 2013. An overview of electricity demand forecasting techniques. Network and complex systems. 3(3): 38-48.
[57]Singh A.K., Khatoon S., Muazzam M. and Chaturvedi D.K. 2012, December. Load forecasting techniques and methodologies: A review. In Power, Control and Embedded Systems (ICPCES), IEEE 2nd International Conference: 1-10.
[58]Ssekulima E.B., Anwar M.B., Al Hinai A. and El Moursi M.S. 2016. Wind speed and solar irradiance forecasting techniques for enhanced renewable energy integration with the grid: a review. IET Renewable Power Generation. 10(7): 885-989.
[59]Sudheer G. and Suseelatha A. 2015. Short term load forecasting using wavelet transform combined with Holt-Winters and weighted nearest neighbor models. International Journal of Electrical Power & Energy Systems. 64: 340-346.
[60]Voyant C., Paoli C., Muselli M. and Nivet M.L. 2013. Multi-horizon solar radiation forecasting for Mediterranean locations using time series models. Renewable and Sustainable Energy Reviews. 28, 44-52.
[61]Xu Y., Dai Y., Dong Z.Y., Zhang R. and Meng K. 2013. Extreme learning machine-based predictor for real-time frequency stability assessment of electric power systems. Neural Computing and Applications. 22(3-4): 501-508.
