Scholarlink

To overcome limitations in IIoT-based bearing monitoring, this paper presents an edge-cloud collaborative system integrating real-time perception with intelligent diagnostics. The architecture leverages an RK3568 edge platform and VTall-S203L triaxial sensor (Modbus protocol) for low-latency feature extraction (RMS, peak values) and threshold-based early warning, with data transmitted via Wi-Fi to Supabase cloud. Cloud analysis combines FFT/envelope spectrum processing with an EMD-CNN-Transformer model for high-precision fault classification, while a Next.js/ECharts frontend and Flask backend enable interactive visualization. The system innovatively incorporates DeepSeek LLM for automated report generation. PHM testbed validation using CWRU data confirms significant improvements in real-time performance and diagnostic accuracy (95.98%).

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

Smart Bearing Health Monitoring System Based on End-Side Perception and Closed-Loop Diagnosis

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • TianLe Liu,
  • JunHao Li,
  • YaLin Xiong,
  • XuDong Zhang,
  • GuiSheng Luo,
  • ShanYu Wei

摘要

To overcome limitations in IIoT-based bearing monitoring, this paper presents an edge-cloud collaborative system integrating real-time perception with intelligent diagnostics. The architecture leverages an RK3568 edge platform and VTall-S203L triaxial sensor (Modbus protocol) for low-latency feature extraction (RMS, peak values) and threshold-based early warning, with data transmitted via Wi-Fi to Supabase cloud. Cloud analysis combines FFT/envelope spectrum processing with an EMD-CNN-Transformer model for high-precision fault classification, while a Next.js/ECharts frontend and Flask backend enable interactive visualization. The system innovatively incorporates DeepSeek LLM for automated report generation. PHM testbed validation using CWRU data confirms significant improvements in real-time performance and diagnostic accuracy (95.98%).