Support Vector Regression Based on Scaling Reproducing Kernel for Black-Box System Identification

Abstract

A new least squares support vector regression model based on scaling reproducing kernel for black-box system identification is presented in this paper. The scaling reproducing kernel, which is a reproducing kernel in reproducing kernel Hilbert space (RKHS), is generated from the set of scaling basis function of some subspace of L 2(R). The support vector regression model incorporated the advantage of the support vector machines and the multi-resolution property of wavelet is discussed in detail. Experiments show that this method has better performance than other approaches