Forecasting of Iran's renewable energy consumption using a hybrid method of backpropagation neural network and Grey Wolf optimization algorithm

Abstract

With the ever-increasing population growth, limited fossil fuels and increasing environmental pollution, many countries in the world are trying to use renewable energy sources for their energy supply by investing in new technologies. In order to provide renewable energy, high accuracy models should be used to predict energy consumption. It is important to know the factors affecting renewable energy consumption for forecasting. In this research, the factors affecting renewable energy consumption have been identified and extracted by reviewing previous researches. Then the criteria and sub-criteria have been prioritized using the Analytical Hierarchy process. The sub-criteria with higher priority include environmental, economic, technical and demographic sub-criteria which were considered as input parameters to the artificial neural network method. Then, the amount of renewable energy consumption in Iran was predicted using the hybrid method of Backpropagation Neural Network and Grey Wolf Optimization Algorithm (GWOA-BNN) and the results were compared with two methods of Backpropagation Neural Network (BNN) and the hybrid method of Improved Whale Optimization Algorithm, Simulated Annealing and BIdirectional Long Short-Term Memory network (IWOA-SA-BILSTM) method. Finally, due to the superiority of the proposed combined GWOA-BNN method, Iran's renewable energy consumption was predicted based on the three scenarios determined for the next three years using the introduced combined method.