Optimal Design of an In-Wheel Permanent Magnet Synchronous Motor Using a Design of Experiment and Kriging Model

Abstract

Recently, the in-wheel PMSM is commonly being used for electric vehicles and robots. In order to have a better torque characteristic, cogging torque, one of the main factors of torque ripple [1], has to be minimized. Therefore, the in-wheel PMSM has a specially designed permanent magnet. In the optimization of a PM motor designed, the use of a response surface method (RSM) is now well established. In RSM, a 2nd polynomial regression (PR) model is generally used for fitting the data [2]. However, the model is not adequate to approximate a nonlinear function due to the low-order PR. Even if the high-order PR is used, it may create an unstable approximation function [3]. In order to reduce these problems, the Kriging model using stochastic processes is combined for global optimization. In this paper, the Kriging model based on DOE is proposed to reduce the cogging torque for in-wheel PMSM considering a root mean square (RMS) value of backemf.