Estimating Electric Energy Consumption in Turkey Using Artificial Neural Networks Optimized with Jaya Algorithm

Abstract

This study’s primary objective was to use a Jaya algorithm to train an artificial neural network (ANN) model such that it can estimate Turkey’s future electric energy consumption (EEC). Employing gross domestic product (GDP), population, import, and export data as independent variables, the proposed method was examined. In order to show accuracy of proposed method, ANN-Jaya was compared with ANN models trained with two other high-performing optimization methods, namely the artificial bee colony (ABC) method and the teaching–learning-based optimization (TLBO) method. The ANN-Jaya model converges to a lower error on the test dataset than the ANN-ABC and ANN-TLBO models. Thus, using the ANN-Jaya model, Turkey's EEC was projected out to 2023 under two different scenarios. The results were compared with projections by TEIAS (Turkish Electricity Transmission Corporation) and other related studies in the literature. The results show that EEC can be accurately modeled using ANN-Jaya, and that this optimization method is advantageous for predicting future electricity consumption

Keywords

• jaya algorithm
• electric energy consumption/demand
• artificial neural networks
• turkey