A study of hybrid data selection method for a wavelet SVR mid-term load forecasting model

Mid-term load forecasting (MTLF) is used to predict the loads for the durations from a week up to a year. Many methods have been used for selecting the best input data which is a critical issue in load forecasting. Recently, two separate approaches based on fuzzy logic system and support vector machine have shown better results compared to statistical techniques. The main purpose of this paper is to employ a novel hybrid approach based on wavelet support vector machines (WSVM) and chaos theory for MTLF. First, kernel-based fuzzy clustering technique and two-step correlation analysis are separately used for selecting training samples. Moreover, chaos theory is used to find the optimum time delay constant and embedding dimension of the load time series. Furthermore, genetic algorithm is employed to optimize the parameters of the WSVM model. EUNITE competition data and Iran power system data are selected to test the proposed method. The results show the efficiency of the suggested method compared with the other methods.