Multiobjective optimization of IPM synchronous motor using response surface methodology and filtered Monte Carlo approach

Abstract

Permanent Magnet Synchronous Motors offer high efficiency and power density besides low assembly effort and hence have been established in a wide market over the past years. Especially buried magnets enable a superior field weakening ability but require an exceptional design effort. Costly Finite Elements computations are inevitable for consideration of the occurring non-linearity and non-trivial magnet shapes. The Response Surface Methodology can reduce the number of FE runs significantly by introducing an acceptably exact second order regression model based on a few carefully chosen design samples. Instead of commonly used, but time-consuming Evolutionary Strategy methods, the Monte Carlo approach is applied for optimization. Using simple filter algorithms, distinctive Pareto frontiers can be determined quickly and related to their causative motor designs.