Scholarlink

<p>To solve the problems of multi-optimization objective conflict and fuzzy optimization range of design parameters caused by nonlinearity in the doubly salient electro-magnetic machine (DSEM), this paper presents an intelligent model-based motor optimization framework. The training dataset of machine learning model is systematically constructed by univariate analysis, research on the effects of key factors, and reasonable design of experiment (DOE). This strategy significantly reduces the number of samples while ensuring data representativeness, thus effectively reducing data requirements and time costs in the modeling process. The framework fits the relationship between optimization parameters and optimization objectives of the DSEM through the genetic algorithm-optimized extreme learning machine (GA-ELM). Multi-objective particle swarm optimization (MOPSO) algorithm is used to obtain the optimal structure of DSEM for multi-objective optimization. The results show that after optimization, the cogging torque of the motor decreases by 49.76%, the three-phase asymmetry (TPA) of the motor decreases by 57.5%, and the electromotive force (EMF) of the motor increases by 0.49%, which verifies the effectiveness of the optimization method.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Multi-objective optimization design of doubly salient electro-magnetic machine based on hybrid machine learning and particle swarm techniques

  • 期刊论文

  • 发表日期：

    2026年2月16日

查看全文
  • Siyuan Jiang,
  • Zhenhao Zhang,
  • Mengzhen Gao,
  • Xiaohang Yang,
  • Yunji Zhao,
  • Manman Yuan,
  • Xiaozhuo Xu

摘要

To solve the problems of multi-optimization objective conflict and fuzzy optimization range of design parameters caused by nonlinearity in the doubly salient electro-magnetic machine (DSEM), this paper presents an intelligent model-based motor optimization framework. The training dataset of machine learning model is systematically constructed by univariate analysis, research on the effects of key factors, and reasonable design of experiment (DOE). This strategy significantly reduces the number of samples while ensuring data representativeness, thus effectively reducing data requirements and time costs in the modeling process. The framework fits the relationship between optimization parameters and optimization objectives of the DSEM through the genetic algorithm-optimized extreme learning machine (GA-ELM). Multi-objective particle swarm optimization (MOPSO) algorithm is used to obtain the optimal structure of DSEM for multi-objective optimization. The results show that after optimization, the cogging torque of the motor decreases by 49.76%, the three-phase asymmetry (TPA) of the motor decreases by 57.5%, and the electromotive force (EMF) of the motor increases by 0.49%, which verifies the effectiveness of the optimization method.