Chaos modeling and real-time online prediction of permanent magnet synchronous motor based on multiple kernel least squares support vector machine

Abstract

A multiple kernel least squares support vector machine (MK-LSSVM) modeling method is proposed for the chaos of permanent magnet synchronous motor (PMSM). An equivalent kernel is built by linear-weighted combination of multi kernels to reduce the dependence of modeling accuracy on kernel function and parameters. The solutions of regression parameters and MK-LSSVM output are given in theory. C-C method is employed for the phase space reconstruction of PMSM chaos, then one-step and multi-step real-time online prediction of reconstructed chaotic series are investigated based on moving window learning method. The effect of different measurement noises on the proposed method is discussed. Simulations show that the proposed method can enhance the modeling accuracy and have strong anti-noise capability.