辽宁石油化工大学学报 ›› 2026, Vol. 46 ›› Issue (2): 58-67.DOI: 10.12422/j.issn.1672-6952.2026.02.007

• 机械工程 • 上一篇    下一篇

基于物理注意力机制的KAN⁃Transformer旋转机械轴承故障诊断方法

张奎1(), 晏永飞1, 王新刚2(), 杨赟1   

  1. 1.辽宁石油化工大学 机械工程学院,辽宁 抚顺 113001
    2.广东石油化工学院 机械工程学院,广东 茂名 525000
  • 收稿日期:2025-04-03 修回日期:2025-04-16 出版日期:2026-04-25 发布日期:2026-04-21
  • 通讯作者: 王新刚
  • 作者简介:张奎(1998-),男,硕士研究生,从事数字孪生,深度学习相关的故障诊断预测等方面的研究;E⁃mail: Z1832609224@outlook.com
  • 基金资助:
    辽宁省科技厅应用基础研究计划资助项目(2023JH2/101300223)

KAN⁃Transformer Fault Diagnosis Method for Rotating Machinery Bearings Based on Physical Attention Mechanism

Kui ZHANG1(), Yongfei YAN1, Xingang WANG2(), Yun YANG1   

  1. 1.School of Mechanical Engineering,Liaoning Petrochemical University,Fushun Liaoning 113001,China
    2.School of Mechanical Engineering,Guangdong University of Petrochemical Technology,Maoming Guangdong 525000,China
  • Received:2025-04-03 Revised:2025-04-16 Published:2026-04-25 Online:2026-04-21
  • Contact: Xingang WANG

摘要:

针对旋转机械轴承故障诊断中振动信号噪声干扰强、特征提取依赖人工设计等导致的故障诊断不准确问题,提出了一种融合轴承动力学机理与Kolmogorov⁃Arnold网络(KAN)的物理注意力Transformer模型。首先,基于Hertz接触理论推导轴承内圈、外圈及滚动体故障特征频率方程,构建频域掩码引导注意力机制聚焦故障敏感频带;其次,设计KAN⁃Transformer架构,通过KAN的多尺度分解能力自适应解析振动信号的时频特征,结合Transformer的全局注意力实现长程依赖建模;最后,在凯斯西储大学(CWRU)轴承数据集上验证了模型的性能。结果表明,该模型的准确率达99.75%,显著优于传统模型。研究结果为旋转机械轴承故障诊断提供了一种高精度、高鲁棒性且物理可解释的解决方案。

关键词: KAN?Transformer, 注意力机制, Hertz理论, 滚动轴承, 故障诊断

Abstract:

Addressing the challenge of inaccurate fault diagnosis caused by the strong interference of vibration signal noise and the dependence of feature extraction on manual design in the fault diagnosis of rotating machinery bearings, this paper proposes a physics⁃informed attention Transformer model that integrates bearing dynamics mechanisms with the Kolmogorov⁃Arnold Network (KAN). First, based on Hertz contact theory, the characteristic fault frequency equations for the bearing inner ring, outer ring and rolling element are derived, and the frequency⁃domain mask⁃guided attention mechanism is constructed to focus on the fault⁃sensitive frequency band; Second, the Kan⁃Transformer architecture is designed to adaptively analyze the time⁃frequency characteristics of vibration signals through the multi⁃scale decomposition ability of Kan, and realize the long⁃range dependence modeling combined with the Transformer's global attention; Finally, the proposed model is evaluated using the Case Western Reserve University (CWRU) bearing data set. Experiments show that the accuracy of the model is 99.75%, which is significantly better than the traditional model. It provides a high⁃precision, robust and physically interpretable solution for bearing fault diagnosis of rotating machinery.

Key words: KAN?Transformer, Attention mechanism, Hertz theory, Rolling bearing, Fault diagnosis

中图分类号: 

引用本文

张奎, 晏永飞, 王新刚, 杨赟. 基于物理注意力机制的KAN⁃Transformer旋转机械轴承故障诊断方法[J]. 辽宁石油化工大学学报, 2026, 46(2): 58-67.

Kui ZHANG, Yongfei YAN, Xingang WANG, Yun YANG. KAN⁃Transformer Fault Diagnosis Method for Rotating Machinery Bearings Based on Physical Attention Mechanism[J]. Journal of Liaoning Petrochemical University, 2026, 46(2): 58-67.

使用本文

0
    /   /   推荐

导出引用管理器 EndNote|Ris|BibTeX

链接本文: https://journal.lnpu.edu.cn/CN/10.12422/j.issn.1672-6952.2026.02.007

               https://journal.lnpu.edu.cn/CN/Y2026/V46/I2/58