Scholarlink

As an integral part of mechatronics, MEMS technology plays a crucial role in fault detection in bearings and gears. It provides the essential sensing and actuation capabilities required for developing intelligent, efficient, compact systems for condition monitoring. Therefore, this research article is focused on developing a comprehensive MEMS-based framework for fault detection of gears and bearings using vibration signal analysis. The MEMS system includes a CPU (Raspberry Pi), STM32 MCU (NuceloF401), MEMS ADXL1002z sensor for vibration signal acquisition and a display screen. The framework leverages the synergy of mechanical, electronic, and computational components to achieve efficient fault detection and classification. Machine learning (ML) models employing Random Forest (RF), Multi-Class Support Vector Machine (MSVM), and Backpropagation Neural Network (BPNN) are used to analyse the extracted features from the vibration data. These three models are compared and analysed to determine their performance in classifying various types of faults and assessing the effectiveness of the MEMS-based system. The integration of MEMS technology with sophisticated ML techniques underscores the potential of mechatronics to enhance condition monitoring and fault diagnosis in rolling bearings and gears.

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

MEMS-Based Gearbox Fault Diagnosis Using Machine Learning Approaches

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Gagandeep Sharma,
  • Tejbir Kaur,
  • Sanjay Kumar Mangal

摘要

As an integral part of mechatronics, MEMS technology plays a crucial role in fault detection in bearings and gears. It provides the essential sensing and actuation capabilities required for developing intelligent, efficient, compact systems for condition monitoring. Therefore, this research article is focused on developing a comprehensive MEMS-based framework for fault detection of gears and bearings using vibration signal analysis. The MEMS system includes a CPU (Raspberry Pi), STM32 MCU (NuceloF401), MEMS ADXL1002z sensor for vibration signal acquisition and a display screen. The framework leverages the synergy of mechanical, electronic, and computational components to achieve efficient fault detection and classification. Machine learning (ML) models employing Random Forest (RF), Multi-Class Support Vector Machine (MSVM), and Backpropagation Neural Network (BPNN) are used to analyse the extracted features from the vibration data. These three models are compared and analysed to determine their performance in classifying various types of faults and assessing the effectiveness of the MEMS-based system. The integration of MEMS technology with sophisticated ML techniques underscores the potential of mechatronics to enhance condition monitoring and fault diagnosis in rolling bearings and gears.