跳到主要导航 跳到搜索 跳到主要内容

数字孪生模型驱动的永磁同步电机偏心故障定量诊断方法

  • Naipeng Li
  • , Tao Zhang
  • , Yaguo Lei
  • , Xiang Li
  • , Bin Yang
  • , Jinze Jiang
  • Xi'an Jiaotong University

科研成果: 期刊稿件 › 文章 › 同行评审

摘要

Eccentricity fault is a common type of failure in permanent magnet synchronous motors, which causes fluctuations in the internal magnetic field and current signals. Based on an improved winding function approach, a high-fidelity simulation-enhanced model considering actual air-gap distribution is established to reveal the electromechanical coupling mechanism. To address the lack of fault samples under multiple eccentricity levels, a quantitative diagnosis method driven by high-fidelity simulation and transfer learning is proposed. Specifically, fault samples are generated using the simulation model, transfer learning aligns features between the source domain (simulation data) and target domain (experimental data), and margin-aware regularization enhances fault recognition under imbalanced conditions. Finally, experiments on a dedicated test bench demonstrate that when fault samples account for only 4.8% of the total, the proposed method achieves an average diagnostic accuracy of 92.5%, significantly improving accuracy and robustness.

投稿的翻译标题Digital Twin Model-Driven Eccentricity Fault Quantitative Diagnosis of Permanent Magnet Synchronous Motors
源语言繁体中文
页(从-至)59-71
页数13
期刊Jixie Gongcheng Xuebao/Journal of Mechanical Engineering
卷62
期14
DOI
出版状态已出版 - 7月 2026

关键词

  • digital twin
  • fault diagnosis
  • permanent magnet synchronous motor
  • transfer learning

学术指纹

探究 '数字孪生模型驱动的永磁同步电机偏心故障定量诊断方法' 的科研主题。它们共同构成独一无二的学术指纹。

引用此