Abstract: Based on the wavelet decomposition theory and conditions of the support vector kernel function,a multivariable support vector kernel function is proposed,i.e.Littlewood-Paley wavelet kernel function for SVM(Support Vector Machine).This function is a kind of orthonormal function,and it can approximate almost any curve in quadratic continuous integral space,thus it enhances the generalization ability of the SVM.Using Littlewood-Paley wavelet function as the support vector kernel function,the Least Square Littlewood-Paley Wavelet Support Vector Machine(LS-LPWSVM) is proposed.Experiment results show that,compared with least square support vector machine under the same conditions,the learning precision is improved by LS-LPWSVM.So,it will be more suitable for learning complicated functions.