Optimal Design of SPMSM Using a Subdivided Meta-Model Assisted Multi-Objective Optimization Algorithm

A subdivided meta-model assisted multi-objective optimization (SMAMOO) algorithm is proposed for the optimal design of a surface-mounted permanent magnet synchronous motor (SPMSM) to supply traction for an electric wheelchair. The SMAMOO algorithm remarkably reduces computation cost by directly adding solutions to the objective function area using a metamodel to solve the problem of the number of function calls of existing algorithms. However, because a meta-model may be different from the actual model due to the use of interpolation, a sample is added near where the most change occurs in the meta-model, increasing the accuracy of the meta-model. The sample is added considering all objective functions. In addition, the grids of the meta-model in the design variable area are subdivided to improve algorithm performance. The process is applied only to the latter part of the algorithm because a small grid requires considerable computation time. The superiority of the SMAMOO algorithm is demonstrated by comparing its performance to that of the non-dominated sorting genetic algorithm-II and multi-objective particle swarm optimization. Finally, the proposed algorithm is applied to the optimal design of a SPMSM, and a prototype is manufactured to verify its validity.