Scholarlink

<p>The planetary transmission mechanism is a critical component in heavy-duty off-road vehicle transmissions, where bearings and gears are prone to damage, often leading to operational failures. Intelligent fault diagnosis systems are essential for mitigating these issues. However, traditional centralized computing approaches face significant challenges, such as high latency and excessive data transmission demands. Edge computing has emerged as a promising solution, yet it must strike a balance between high diagnostic accuracy, real-time performance, and efficient utilization of limited computational resources. This study proposes a lightweight model specifically designed for edge computing, termed the spectral hierarchical feature convolution network (SHFCN). Vibration signals obtained from fault injection tests are enhanced to highlight fault-related features, creating a robust dataset for training the SHFCN model. An edge-based intelligent fault diagnosis system (EC-SHFCN) is developed, integrating SHFCN to offload diagnostic tasks to edge devices, thereby enhancing computational efficiency and reducing response times. The system is implemented on an industrial three-proof terminal and validated using a planetary transmission mechanism test bench. Experimental results demonstrate an average classification accuracy of 99.31%, meeting the real-time requirements for online fault diagnosis. This system significantly improves the reliability and efficiency of fault diagnosis, setting a new benchmark for edge computing applications in industrial diagnostics.</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

Lightweight edge fault diagnosis model for planetary transmission mechanism using enhanced signal feature processing

  • 期刊论文

  • 发表日期：

    2025年7月23日

查看全文
  • Ran Gong,
  • Wei Jiang,
  • Jinxiao Li,
  • Jinle Zhang,
  • Jiamin He

摘要

The planetary transmission mechanism is a critical component in heavy-duty off-road vehicle transmissions, where bearings and gears are prone to damage, often leading to operational failures. Intelligent fault diagnosis systems are essential for mitigating these issues. However, traditional centralized computing approaches face significant challenges, such as high latency and excessive data transmission demands. Edge computing has emerged as a promising solution, yet it must strike a balance between high diagnostic accuracy, real-time performance, and efficient utilization of limited computational resources. This study proposes a lightweight model specifically designed for edge computing, termed the spectral hierarchical feature convolution network (SHFCN). Vibration signals obtained from fault injection tests are enhanced to highlight fault-related features, creating a robust dataset for training the SHFCN model. An edge-based intelligent fault diagnosis system (EC-SHFCN) is developed, integrating SHFCN to offload diagnostic tasks to edge devices, thereby enhancing computational efficiency and reducing response times. The system is implemented on an industrial three-proof terminal and validated using a planetary transmission mechanism test bench. Experimental results demonstrate an average classification accuracy of 99.31%, meeting the real-time requirements for online fault diagnosis. This system significantly improves the reliability and efficiency of fault diagnosis, setting a new benchmark for edge computing applications in industrial diagnostics.