Title: A particle swarm optimised support vector regression for short-term load forecasting

Authors: Su Wutyi Hnin; Chawalit Jeenanunta

Addresses: School of Management Technology, Sirindhorn International Institute of Technology, Thammasat University, Pathum Thani 12000, Thailand ' School of Management Technology, Sirindhorn International Institute of Technology, Thammasat University, Pathum Thani 12000, Thailand

Abstract: The aim of this paper is to present a forecasting model for daily electricity demand. Support vector regression (SVR) has the ability that can perform well in nonlinear forecasting problems. In this paper, the parameter optimisation for SVR is proposed by using particle swarm optimisation (PSO). The data for testing the proposed method is obtained from the Electricity Generating Authority of Thailand (EGAT). The data have been recorded in every 30 minutes. The data from 2012 to 2013 is used for training to forecast daily electricity load demand in 2013. The performance of the model is measured by the mean absolute percentage error (MAPE). The results of SVR and SVR-PSO are compared. Optimising hyperparameters with PSO outperforms the SVR.

Keywords: particle swarm optimisation; PSO; support vector regression; SVR; optimising hyperparameters; short-term load forecasting.

DOI: 10.1504/IJETP.2020.107965

International Journal of Energy Technology and Policy, 2020 Vol.16 No.4, pp.399 - 412

Accepted: 11 Feb 2019
Published online: 01 Jul 2020 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article