Short-term power load forecasting with least squares support vector machines and wavelet transform

Abstract

Based on least squares support vector machines (LS-SVM) and Wavelet Transform theory, a novel approach for short-term power load forecasting is presented. The historical time series is decomposed by wavelet, so the approximate part and several detail parts are obtained. Then the results of Wavelet Transform are predicted by a separate LS-SVM predictor. The new forecast model combines the advantage of WT with LS-SVM. Compared with other predictors, this forecast model has greater generalizing ability and higher accuracy.