Wavelet Support Vector Machines and its application for nonlinear system identification

Abstract

Based on a novel multidimensional wavelet kernel constructed in Reproducing Kernel Hubert Space (RKHS), an identification scheme with the Wavelet Support Vector Machine (WSVM) estimator is proposed for nonlinear dynamic systems. The good reproducing property of wavelet kernel function enhances the generalization ability of the system identification scheme. Two cases are presented to validate the proposed method and show its feasibility. © Springer-Verlag Berlin Heidelberg 2005.