Scholarlink

Interturn short circuit (ITSC) faults are the most common faults in switched reluctance machines (SRMs), and minor ITSC faults are even prone to triggering other types of winding short circuit faults to expand the scope of faults. In this paper, a combination of online and offline ITSC fault diagnosis method is proposed. Firstly, the ITSC fault motor model is constructed and its phase torque signal is analyzed harmonically, and the second harmonic after coordinate transformation is extracted online as the fault eigenvalue. Secondly, the optimal intrinsic mode function (IMF) components are obtained by using the variational mode decomposition (VMD) optimized by the osprey-cauchy-sparrow search algorithm (OCSSA), and 19 time-frequency domain features are extracted for the detection of ITSC faults and their severity by using a convolutional neural network-bidirectional long short-term memory (CNN-BiLSTM) network model. Finally, the validity and superiority of the method proposed in this paper is verified by example comparison.

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

Diagnosis of Interturn Short Circuit Faults in Switched Reluctance Machines Based on Parameter Optimized VMD and CNN-BiLSTM

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Lefei Ge,
  • Jiahe Song,
  • Yanwen Zheng,
  • Ronghui Zhou,
  • Qianrao Fu

摘要

Interturn short circuit (ITSC) faults are the most common faults in switched reluctance machines (SRMs), and minor ITSC faults are even prone to triggering other types of winding short circuit faults to expand the scope of faults. In this paper, a combination of online and offline ITSC fault diagnosis method is proposed. Firstly, the ITSC fault motor model is constructed and its phase torque signal is analyzed harmonically, and the second harmonic after coordinate transformation is extracted online as the fault eigenvalue. Secondly, the optimal intrinsic mode function (IMF) components are obtained by using the variational mode decomposition (VMD) optimized by the osprey-cauchy-sparrow search algorithm (OCSSA), and 19 time-frequency domain features are extracted for the detection of ITSC faults and their severity by using a convolutional neural network-bidirectional long short-term memory (CNN-BiLSTM) network model. Finally, the validity and superiority of the method proposed in this paper is verified by example comparison.