Robust Design Optimization for a Hairpin Winding Permanent Magnet Motor Considering Permanent Magnet Inconsistency and Manufacturing Uncertainty

This paper presents an effort to develop an effective method for the robust design optimization of a hairpin winding permanent magnet motor. The comprehensive influence of variable uncertainties due to PM property inconsistency and manufacturing uncertainties is considered. During the implementation of robust design optimization of the hairpin winding permanent magnet motor by using an evolution-based multi-objective optimization algorithm, one major obstacle is the huge computation cost, especially for the robust analysis process. Therefore, a two-level solution strategy, an adaptive least square support vector regression model, and a novel dominant judging condition are skillfully employed to realize the tradeoff between the computation cost and accuracy of the robust design optimization algorithm. To verify the effectiveness of the proposed method, the fluctuation and distribution of electromagnetic performance of the robust motor design are analyzed and compared with those of a deterministic motor design. Finally, a prototype of the hairpin winding permanent magnet motor is manufactured and tested to further confirm the validity of the proposed method.