Short-Term Load Forecasting Using Ensemble Empirical Mode Decomposition and Harmony Search Optimized Support Vector Regression

Abstract

Short-term load forecasting (STLF) is very significant for the planning and reliable operation of power grids. This paper proposes an STLF model by hybridizing ensemble empirical mode decomposition (EEMD) and harmony search (HS) optimized support vector regression (SVR), namely EEMD-HS-SVR model. Firstly, the original load time series are decomposed into a set of intrinsic mode functions (IMFs) and a residue by EEMD. Secondly, both IMF components and residue are forecasted by HS optimized SVR. Finally, the prediction values of the original load time series are calculated by the sum of the forecasting values of every sub-series. The proposed model is verified using two data sets, and results are compared against different alternative models. The forecasting results indicate that the proposed EEMD-HS-SVR model effectively generates more accurate predictive results.