汽车工程 ›› 2025, Vol. 47 ›› Issue (10): 1942-1952.doi: 10.19562/j.chinasae.qcgc.2025.10.010

• • 上一篇    

基于小样本WPRBF-MEVO车用电机转矩脉动优化抑制研究

陈龙1,3(),李飞鸿1,陈晓斌1,卢传浩1,2,赵肖楠1,刘巧斌4   

  1. 1.太原理工大学机械工程学院,太原 030024
    2.太原理工大学航空航天学院,太原 030606
    3.中国煤炭科工集团太原研究院有限公司,太原 100048
    4.华南理工大学机械与汽车工程学院,广州 510641
  • 收稿日期:2025-03-07 修回日期:2025-05-06 出版日期:2025-10-25 发布日期:2025-10-20
  • 通讯作者: 陈龙 E-mail:chenlong@tyut.edu.cn
  • 基金资助:
    国家重点研发计划项目(2024YFB4711001);中国博士后科学基金(2021M753542);山西省基础研究计划(20210302124119);山西省基础研究计划(202203021212259);国家工程实验室开放项目(GCZX-2023-02);山西省重点研发计划项目(2022ZDYF019)

Research on Optimization and Suppression of Torque Ripple of Vehicle Motor Based on Small Sample WPRBF-MEVO

Long Chen1,3(),Feihong Li1,Xiaobin Chen1,Chuanhao Lu1,2,Xiaonan Zhao1,Qiaobin Liu4   

  1. 1.School of Mechanical Engineering,Taiyuan University of Technology,Taiyuan 030024
    2.School of Aeronautics and Astronautics,Taiyuan University of Technology,Taiyuan 030606
    3.China Coal Technology Engineering Group Taiyuan Research Institute Co. ,Ltd. ,Taiyuan 100048
    4.School of Mechanical and Automotive Engineering,South China University of Technology,Guangzhou 510641
  • Received:2025-03-07 Revised:2025-05-06 Online:2025-10-25 Published:2025-10-20
  • Contact: Long Chen E-mail:chenlong@tyut.edu.cn

摘要:

驱动电机转矩脉动直接影响整车NVH性能,针对其优化抑制难题,提出融合多项式增强平均权重径向基函数(weighted average and polynomial augmented radial basis function, WPRBF)代理模型与多目标能量谷优化算法(multi-objective energy valley optimizer, MEVO)的协同优化框架。首先建立电机参数化有限元模型并进行台架实验验证;其次采用拉丁超立方抽样开展实验设计获取样本,提出WPRBF方法构建高精度代理模型;最后采用MEVO算法开展多目标优化设计,并引入熵权-模糊集理论综合决策机制获取Pareto前沿最优解。结果表明:(1)WPRBF模型在相同建模精度的前提下,较传统KRG代理模型减少约40%的样本需求;(2)优化后电机转矩输出均值提升6.83%,转矩脉动系数降低20.00%,齿槽转矩峰值削减23.80%。验证了本文方法的有效性。

关键词: 电驱动系统, 多目标优化, 代理模型, 有限元分析

Abstract:

The torque ripple of the drive motor directly affects the NVH performance of the entire vehicle. To address the optimization and suppression challenge of this issue, a collaborative optimization framework integrating the Weighted Average and Polynomial Augmented Radial Basis Function (WPRBF) surrogate model and the Multi- Objective Energy Valley Optimizer (MEVO) algorithm is proposed. Firstly, a parametric finite element model of the motor is established and verified through bench tests. Secondly, Latin hypercube sampling is employed for experimental design to obtain samples, and the WPRBF method is proposed to construct a high-precision surrogate model. Finally, the MEVO algorithm is used for multi-objective optimization design, and the entropy weight-fuzzy set theory comprehensive decision-making mechanism is introduced to obtain the Pareto front optimal solution. The results show that: (1) Under the same modeling accuracy, the WPRBF model requires approximately 40% fewer samples than the traditional KRG surrogate model; (2) After optimization, the mean value of the motor torque output increases by 6.83%, with the torque ripple coefficient decreasing by 20.00%, and the peak value of cogging torque reduced by 23.80%. This verifies the effectiveness of the method proposed in this paper.

Key words: electric drive system, multi-objective optimization, surrogate model, FEA