Affiliations:
School of Economic and Management, North China Electric Power University, Hui Longguan, Chang Ping District, Beijing, China
Correspondence:
[*]
Corresponding author. Zongyun Song, School of Economic and Management, North China Electric Power University, Beinong Street 2, Hui Longguan, Chang Ping District, Beijing 102206, China. Tel.: +86 152 10800270; Fax: +010 617 73311; E-mail: songzongyun@126.com.
Abstract: Accurate short-term load forecasting plays a crucial role in electricity industry and market. In this study, a novel forecasting method based on Support Vector Machine (SVM) and Firefly Algorithm (FA) has been created to realize accurate and reliable load prediction. The performance of SVM highly depends on the selection of parameters, and Gaussian disturbance Firefly Algorithm (GDFA) proposed in this study can satisfy the necessary. Ensemble Empirical Mode Decomposition (EEMD) is employed to decompose the load data into sub-series with different frequency. This paper extracts daily maximum temperature, daily minimum temperature, wind speed, rainfall, day type, and the load one week before the forecasting day as input variables. Two cases are taken to verify the effective performance of GDFA compared with FA, as well as the superiority of EEMD-GDFA-SVM over other forecasting techniques in short-term load forecasting.
