Short-Term Load Forecasting Based on Gaussian Wavelet SVM

Abstract

Gaussian wavelet support vector machine (SVM) is constructed for short-term load forecasting (STLF). It is proved that the pth -derivative Gaussian wavelet is an admissible translation-invariant kernel function of SVM when p is an even number. The Gaussian wavelet SVM is constructed with wavelet kernel function, and improved stochastic focusing search (SFS) algorithm is used to optimize the parameters of SVM and its kernel function. The experiments of STLF are conducted using the proposed SVM, the conventional Gaussian SVM and Morlet wavelet SVM respectively. The comparison shows that the proposed method is efficient and superior, and has some application value in STLF.