【目的】为进一步提高车用永磁同步电机(PMSM)的性能，针对电机传统优化方法的低效性，提出一种基于改进广义回归神经网络(GRNN)和改进樽海鞘群优化算法的车用PMSM的多目标优化设计方法。【方法】首先搭建参数化电机模型，通过有限元模型对电机进行参数化扫描，从而获取电机结构参数和对应性能的样本数据，并通过GRNN进行模型搭建;其次以齿槽转矩峰峰值最小、额定平均转矩最大和转矩脉动最小为优化目标，采用樽海鞘群算法对电机的结构参数进行优化。【结果】优化后的电机平均转矩提高了2%，转矩脉动降低了16%，齿槽转矩降低了60.58%。【结论】本文所提出的方法能够充分利用算力资源进行并行计算，相比传统的优化方法，更快速有效地实现电机的多目标优化设计。

[Objective] To further improve the performance of automotive permanent magnet synchronous motors (PMSM) and address the inefficiency of traditional optimization methods for motors, a multi-objective optimization design method for automotive permanent magnet synchronous motors is proposed based on improved generalized regression neural network (GRNN) and improved salp swarm optimization algorithm. [Methods] Firstly, a parameterized motor model was constructed, and the motor was scanned through the finite element model to obtain sample data of the motor's structural parameters and corresponding performance. Then the model was built using GRNN. With the objectives of minimizing the peak-to-peak value of cogging torque, maximizing the rated average torque, and minimizing torque ripple, the structural parameters of the motor were optimized using the salp swarm algorithm. [Results] The optimized motor’s average torque increased by 2%, the torque ripple decreased by 16%, and the cogging torque decreased by 60.58%. This demonstrates the efficiency and accuracy of the method. [Conclusion] The proposed method can fully utilize computing power resources for parallel computing. Compared to traditional optimization methods, it can achieve multi-objective optimization design of motors more quickly and effectively.

国家自然科学基金(42176194)