Scholarlink

The significance of power equipment anomaly detection has become increasingly evident. However, existing methods exhibit notable deficiencies in data preprocessing and the application of unsupervised algorithms. To address these issues, this study introduces the RobustSTL decomposition algorithm and the Anomaly Transformer algorithm. The experimental results demonstrate that, on some datasets, compared to the direct detection method without decomposition, the detection accuracy is significantly enhanced by 18% and f1 has increased by 20% when using RobustSTL decomposition.

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

Enhancing Power Equipment Anomaly Detection Performance with RobustSTL and Anomaly Transformer

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Zuo Lu,
  • Yixiang Deng,
  • Peiran Xing,
  • Xiaoyang Wang

摘要

The significance of power equipment anomaly detection has become increasingly evident. However, existing methods exhibit notable deficiencies in data preprocessing and the application of unsupervised algorithms. To address these issues, this study introduces the RobustSTL decomposition algorithm and the Anomaly Transformer algorithm. The experimental results demonstrate that, on some datasets, compared to the direct detection method without decomposition, the detection accuracy is significantly enhanced by 18% and f1 has increased by 20% when using RobustSTL decomposition.