Scholarlink

<p>Gearboxes are the most likely subsystems of rotating machinery that are liable for equipment breakdown. Most studies address the presence of a fault and evaluate its accuracy, but fault severity analysis is vital for decision-making in predictive maintenance. The present study focused on the fault diagnosis of gears using tree-based machine learning techniques and fault severity analysis using weighted aggregation approach. Features are extracted from the collected vibration data and ranked via the decision tree ranking technique. The highest testing accuracy of 97% was obtained with the top nine ranked features and the extreme gradient boosting classifier. The proposed study uses a novel weighted aggregation technique and assigns feature-specific weights on the basis of their diagnostic relevance to compute a composite severity score, effectively ranking fault conditions from severe to mild. The findings underscore the importance of feature extraction and aggregation techniques for enhanced fault analysis and prioritize predictive maintenance strategies.</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

Weighted Aggregation Approach and Tree-Based Ensemble Technique for Anomaly Detection and Fault Severity Analysis in Gear Fault Diagnosis

  • 期刊论文

  • 发表日期：

    2025年4月11日

查看全文
  • Ramesh V. Bhandare,
  • Vikas M. Phalle,
  • Vishwadeep C. Handikherkar,
  • Sangram S. Patil

摘要

Gearboxes are the most likely subsystems of rotating machinery that are liable for equipment breakdown. Most studies address the presence of a fault and evaluate its accuracy, but fault severity analysis is vital for decision-making in predictive maintenance. The present study focused on the fault diagnosis of gears using tree-based machine learning techniques and fault severity analysis using weighted aggregation approach. Features are extracted from the collected vibration data and ranked via the decision tree ranking technique. The highest testing accuracy of 97% was obtained with the top nine ranked features and the extreme gradient boosting classifier. The proposed study uses a novel weighted aggregation technique and assigns feature-specific weights on the basis of their diagnostic relevance to compute a composite severity score, effectively ranking fault conditions from severe to mild. The findings underscore the importance of feature extraction and aggregation techniques for enhanced fault analysis and prioritize predictive maintenance strategies.