Scholarlink

Early issue identification is more crucial than ever because wind energy projects must reduce costs, which increases the value of minimizing downtime. The strategy of condition monitoring and problem identification lowers downtime, boosts productivity, and eventually lowers maintenance costs. The data gathered by the SCADA system can be utilized to identify and diagnose issues using machine learning techniques. Hyper-parameter tuning is used to accomplish the goal of building a framework for fault prediction with fewer features than necessary. The model achieves f1-scores ranging from 58 to 94% and accuracy ranging from 73 to 98% by using SVM and hyper-parameter adjustment to enhance classification performance. Five fault modes were generated and evaluated using data from a SCADA system for a wind turbine to test the method.

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

Efficient Wind Turbine Fault Diagnosis Using Machine Learning Technique and Hyper-Parameter Tuning

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Soumya Ranjan Mohapatro,
  • Mayur Mulchandani,
  • Nitish Mohanty,
  • Manas Ranjan Sethi,
  • Sudarsan Sahoo,
  • Abdus Samad

摘要

Early issue identification is more crucial than ever because wind energy projects must reduce costs, which increases the value of minimizing downtime. The strategy of condition monitoring and problem identification lowers downtime, boosts productivity, and eventually lowers maintenance costs. The data gathered by the SCADA system can be utilized to identify and diagnose issues using machine learning techniques. Hyper-parameter tuning is used to accomplish the goal of building a framework for fault prediction with fewer features than necessary. The model achieves f1-scores ranging from 58 to 94% and accuracy ranging from 73 to 98% by using SVM and hyper-parameter adjustment to enhance classification performance. Five fault modes were generated and evaluated using data from a SCADA system for a wind turbine to test the method.