Design optimization of permanent magnet motors using response surface methodology and genetic algorithms

This paper describes work done on optimization of design of permanent magnet motors using response surface modeling and genetic algorithms (GAs). Finite element computations have been used for numerical experiments on geometrical design variables in order to evaluate the coefficients of a second-order model for the response surfaces representing machine parameters (/spl lambda//sub m/, L/sub d/, and L/sub q/). GAs were used to optimize the torque and speed of the machine in terms of these variables.

I. Introduction

FINITE element method (FEM) is one of the most valuable tools in electric machine design since it accurately takes into account nonlinear materials and complex geometries. However, it is only an analysis tool; the design procedure is based on trial and error which is time-consuming and uncertain. This main drawback can be eliminated by combining FEM with the response surface method (RSM). RSM is a statistical tool to build an empirical model of a response with respect to some input variables. It is particularly suitable when the underlying phenomenon is not well known or too complex to be modeled mathematically [1]. The model is derived by regression from a small number of observations of the response provided by FEM. Optimization tools like genetic algorithm (GA) can then work directly on this model to optimize the device with a saving in computational costs. The combination of FEM, RSM, and GA results in a fast and accurate design procedure. We describe the methodology in this paper to optimize the constant power speed range (CPSR) of an interior permanent magnet synchronous motor (IPMSM) shown in Fig. 1.