Taylor ant lion optimization‐based generative adversarial networks for forecasting electricity consumption

Abstract

Summary This article proposes a Taylor ant lion optimization‐based generative adversarial method (TaylorALO‐based GAN) for predicting renewable energy. The proposed renewable energy prediction mechanism involves four different modules, namely, data transformation, extraction of technical indicator, feature selection, and the prediction. At first, the time‐series data is presented to data transformation module where the process is performed using Yeo–Johnson transformation. The transformed data is subjected to the technical indicator extraction module, where the technical indicators are efficiently extracted for further processing. After that, features are extracted based on wrapper selection model. Finally, renewable energy prediction is completed using GAN, which is trained using developed TaylorALO, which is the amalgamation of Taylor series and ant lion optimization algorithm (ALOA). The proposed technique achieved the better performance with the minimal mean square error (MSE) of 8.536 and minimal root mean square error of 2.921 based on per capita consumption.