Improved genetic optimization algorithm with subdomain model for multi-objective optimal design of SPMSM

For an optimal design of a surface-mounted permanent magnet synchronous motor (SPMSM), many objective functions should be considered. The classical optimization methods, which have been habitually designed based on magnetic circuit law or finite element analysis (FEA), have inaccuracy or calculation time problems when solving the multi-objective problems. To address these problems, the multi-independent -population genetic algorithm (MGA) combined with subdomain (SD) model are proposed to improve the performance of SPMSM such as magnetic field distribution, cost and efficiency. In order to analyze the flux density harmonics accurately, the accurate SD model is first established. Then, the MGA with time-saving SD model are employed to search for solutions which belong to the Pareto optimal set. Finally, for the purpose of validation, the electromagnetic performance of the new design motor are investigated by FEA, comparing with the initial design and conventional GA optimal design to demonstrate the advantage of MGA optimization method.