Research of PMSM Modeling Based on Least Square Support Vector Machines

Abstract

A modeling approach based on least square support vector machine (LSSVM) had been applied for permanent magnet synchronous motor (PMSM) and inverter with PMSM, which was multi-variable, nonlinear and coupled system. The modeling parameters with RBF kernel function was optimized by using cross validation method. The simulation result show that the tow modeling method is very effective. The maximum root mean square error (RMSE) of modeling of PMSM is 0.3196 and the maximum relative error is 0.2341%. And the maximum RMSE of modeling of inverter with PMSM is 0.4421 and the maximum relative error is 2.4121%. Using LSSVM for modeling of PMSM performs better forecast accuracy and successful modeling of PMSM.