According to the constructional conditions of Support Vector Machine kernel functions and reproducing kernel theory in Hilbert space,a new kernel function is constructed and combined with the improved Least Square Support Vector Machine (LS-SVM),and thus LS SVM with reproducing kernel is proposed.The numerical simulation of the piezoelectric response of the piezoelectric smart composite laminated plates under the low velocity impact load is performed by using the first order shear deformation theory and finite element method,and the features are extracted.Then,based on the features of piezoelectric sensors’ responsive signals,LS-SVM with reproducing kernel is applied to detect the impact locations for the piezoelectric smart composite laminated plates,and compared with LS-SVM with RBF kernel.The results show that,LS-SVM with reproducing kernel possesses the higher accuracy of impact locating,and the better generalization ability than LS SVM with RBF kernel.In addition,the simulation method can give a degree of guidance for the practical structural impact damage locations self diagnosing.