关键词: 永磁电机, 有限元法, 网格重构, 响应面模型, 多目标优化

Abstract: In order to solve the problem of time consuming of traditional optimization methods which uses finite element method with traditional grid reconstruction in the iterations, a method of solving the motor multi-objective optimization problem based on the improved finite element method was presented. The overlapping re-mesh free finite element method was used. This method improves the mesh of traditional finite element method, the mesh will not be reconstructed entirely when the design area changes greatly. To maintain the requirements of structure designing of permanent magnet motor, torque and temperature rise were set as the objective functions. The objective functions were simulated by the moving least squares algorithm and improved response surface model. With multi-objective genetic algorithm, the structure of the different types of motor were optimized. The optimization results indicated that, compared with the original design, the proposed algorithm can obviously reduce the motor temperature rise, improve the torque density, and can greatly reduce the calculation load and calculation time.

Key words: permanent magnet motor, finite element method (FEM), mesh reconstruction, response surface model (RSM), multi-objective optimization