Scholarlink

<p>In recent years, with the increasing speed of high-speed trains, higher safety and stability are needed. Due to the long-term impact of complex excitation on the gearbox, the gear often fails in the early stage, and the signal fluctuation is strongly disturbed. Therefore, a new feature extraction and fault diagnosis algorithm under strong interference is proposed in this paper. The first step of this research is to establish the electromechanical coupling model of the integrated electric drive system. The second step is to explore the response characteristics of stator current signal electromagnetic torque signal to gear fault based on the mechanism of electromechanical coupling. In the third step, a feature extraction method combining variational mode decomposition, harmonic spectrum kurtosis and variational mode extraction is proposed, and the feature vector is formed based on 22 eigenvalues. In the fourth step, a fault diagnosis algorithm combining particle swarm optimization and backpropagation neural network is proposed. Finally, the method is verified with 99.5% accuracy using 1200 eigenvector sets of stator current and electromagnetic torque. Compared with traditional BP neural network algorithm and mainstream prediction algorithm, this method has higher accuracy and superiority.</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

Non-intrusion Gear Fault Diagnosis of 400 km/h+ High-Speed Train Based on Signal Fusion of Stator Current and Electromagnetic Torque

  • 期刊论文

  • 发表日期：

    2025年5月21日

查看全文
  • Feng Wang,
  • Hao Gong,
  • Yi-qing Ni

摘要

In recent years, with the increasing speed of high-speed trains, higher safety and stability are needed. Due to the long-term impact of complex excitation on the gearbox, the gear often fails in the early stage, and the signal fluctuation is strongly disturbed. Therefore, a new feature extraction and fault diagnosis algorithm under strong interference is proposed in this paper. The first step of this research is to establish the electromechanical coupling model of the integrated electric drive system. The second step is to explore the response characteristics of stator current signal electromagnetic torque signal to gear fault based on the mechanism of electromechanical coupling. In the third step, a feature extraction method combining variational mode decomposition, harmonic spectrum kurtosis and variational mode extraction is proposed, and the feature vector is formed based on 22 eigenvalues. In the fourth step, a fault diagnosis algorithm combining particle swarm optimization and backpropagation neural network is proposed. Finally, the method is verified with 99.5% accuracy using 1200 eigenvector sets of stator current and electromagnetic torque. Compared with traditional BP neural network algorithm and mainstream prediction algorithm, this method has higher accuracy and superiority.