报告详情
  1. 活动首页
  2. 报告清单
  3. 报告详情
A Hierarchical Framework for Three-Phase Motor Fault Diagnosis Using Health Baseline Intervals and Discrete Wavelet Transform
编号：184 访问权限：仅限参会人 更新：2026-10-19 18:08:41 浏览：10次 张贴报告

报告开始：暂无开始时间（Asia/Shanghai）

报告时间：暂无持续时间

所在会场：[暂无会议] [暂无会议段]

暂无文件

摘要
Diagnosing incipient faults in three-phase motors under varying operating conditions is challenging because healthy current patterns change with operating conditions and weak fault signatures are easily obscured. This paper proposes a hierarchical diagnostic framework that connects anomaly detection with fine-grained fault classification. Frequency-domain cross-correlation aligns the three-phase currents, while condition-specific health baseline intervals and multidomain features support anomaly-probability estimation through a heterogeneous classifier ensemble. A db4 discrete wavelet transform and a dual-attention residual convolutional neural network then identify fault types from multiscale current representations. A class-adaptive soft gate combines the anomaly probability with the classifier output to maintain consistency between health-state assessment and fault identification. The framework achieves 99.44% accuracy and 99.44% macro-F1 on BJTU-RAO, and 99.92% accuracy and 99.92% macro-F1 on an in-house dataset. Disturbance and ablation experiments further demonstrate its robustness to signal variations and its applicability to datasets with different numbers of fault classes. These results support the proposed framework as a modular approach to multi-condition motor fault diagnosis.
关键词
Three-phase motor fault diagnosis, health baseline intervals, discrete wavelet transform, attention mechanism, soft gating, hierarchical diagnosis
报告人
Yang Dai
Master Candidate, Me Chongqing university

稿件作者
Yang Dai Chongqing university
Wenbin Huang Chongqing university
Weiming Ma Chongqing university
Yijie Shen Chongqing university
Yuchao Jin Chongqing university
Xuegang Li Chongqing university
发表评论
验证码 看不清楚，更换一张
全部评论
重要日期

  • 会议日期

    11月06日

    2026

    至

    11月08日

    2026

  • 10月15日 2026

    初稿截稿日期

主办单位
IEEE Instrumentation and Measurement Society
承办单位
Sichuan University
联系方式
登录查看完整联系方式
历届会议
移动端
在手机上打开
小程序
打开微信小程序
客服
扫码或点此咨询