Scholarlink

<p>The traction systems of high-speed trains operate under variable load conditions, which induce significant variations in system data characteristics. Traditional methods usually use a unified global model to describe the traction system, but this approach easily ignores the local features of the data, resulting in an increase in false alarm rate. Therefore, this paper proposes a new data-driven FDD method based on a conditional variational autoencoder (CVAE) to address this challenge. The key advantages of the proposed method include: 1) The proposed method significantly improves the sensitivity and reliability of fault detection under variable loads. 2) The proposed FDD framework does not require precise physical models or system-specific parameters, making it highly adaptable. 3) The proposed method can be readily extended to other nonlinear industrial systems. The effectiveness of the proposed method is validated on a traction system of a high-speed train simulation platform.</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

A Variable Load Fault Detection and Diagnosis Method for Traction Systems of High-speed Trains

  • 期刊论文

  • 发表日期：

    2025年9月9日

查看全文
  • Xuedong Li,
  • Hongzhi Wang,
  • Zhiwei Wan,
  • Chao Cheng

摘要

The traction systems of high-speed trains operate under variable load conditions, which induce significant variations in system data characteristics. Traditional methods usually use a unified global model to describe the traction system, but this approach easily ignores the local features of the data, resulting in an increase in false alarm rate. Therefore, this paper proposes a new data-driven FDD method based on a conditional variational autoencoder (CVAE) to address this challenge. The key advantages of the proposed method include: 1) The proposed method significantly improves the sensitivity and reliability of fault detection under variable loads. 2) The proposed FDD framework does not require precise physical models or system-specific parameters, making it highly adaptable. 3) The proposed method can be readily extended to other nonlinear industrial systems. The effectiveness of the proposed method is validated on a traction system of a high-speed train simulation platform.