A short term load forecasting based on bagging-ELM algorithm

Abstract

Focused on Load forecasting for electric power plan, a novel prediction model, which was based on machine learning, was established. We propose Bagging algorithm optimized Extreme Learning Machine (ELM) prediction model with the fast learning ability of ELM and weight altering of Bagging to increase the prediction accuracy. Finally, it is applied on short term load forecasting problem verified by the EUNITE load forecasting datasets. Compared with winning algorithm of EUNITE competition, Bagging-ELM prediction model has a better performance on prediction accuracy. © 2013 Springer-Verlag.