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

Finite element method (FEM) reveals to be one of the most valuable tools in electric machines design as it accurately takes into account non-linear materials and complex geometry. However, this is by nature an analysis tool, which makes the design procedure very iterative, hence time consuming. Furthermore, the final result is quite uncertain. This main drawback can be eliminated by combining FEM with response surface method (RSM). RSM is a statistical tool to derive experimentally an analytical relationship between a response and some input variables. It is particularly suitable when the underlying phenomenon linking them is not well known or too complex to be modeled mathematically. The relationship is derived by regression from a small number of observations of the response that can be provided by FEM. Genetic algorithm (GA) can then worked directly on this model to optimize the device with a decrease in computational cost. This paper presents the combination of FEM, RSM and GA to optimize the rotor geometry of an interior permanent magnet synchronous motor for fast and accurate design procedure.