In practice， it is difficult to obtain analytical solution for two-point boundary value problem， the two-scale wavelet kernel LS-SVM method is proposed to solve the approximate solution of two-point boundary value problem. Firstly， the two-point boundary value problem is transformed into an objective optimization problem with two constraint conditions， and then an approximate solution satisfying the boundary conditions is constructed by using the combination of two-scale wavelet kernel functions. The first constrained condition is approximated by the first scale wavelet kernel function， and the second constraint condition is that the error function of the first approximation is approximated by the second scale wavelet kernel function， which can improve the approximation accuracy of the approximate solution. Finally， the objective optimization problem is transformed into the regression problem， and then the LS-SVM method is used to solve the regression coefficients. In the process of coefficients solving， the key is to transform the parametric regression problem into quadratic programming problem， which can avoid complex differential operation. Numerical experiments show that the proposed method has high accuracy， less calculation and good stability. Therefore， the two-scale wavelet kernel LS-SVM method is effective for solving the two-point boundary value problem， and has the characteristics of high precision， differentiability， simple expression and fixed form.