东北大学学报(自然科学版) ›› 2025, Vol. 46 ›› Issue (3): 20-27.DOI: 10.12068/j.issn.1005-3026.2025.20239058

• 信息与控制 • 上一篇    下一篇

基于WP-TRP的滚动轴承故障诊断方法

王娜1,2(), 崔月磊1, 罗亮1, 王子从1   

  1. 1.天津工业大学 控制科学与工程学院,天津 300387
    2.天津工业大学 天津市电气装备智能控制重点实验室,天津 300387
  • 收稿日期:2023-10-11 出版日期:2025-03-15 发布日期:2025-05-29
  • 通讯作者: 王娜
  • 作者简介:王 娜(1977―),女,河北衡水人,天津工业大学讲师,博士.
  • 基金资助:
    天津市重点研发计划项目(19YFHBQY00040)

Fault Diagnosis Method for Rolling Bearings Based on WP-TRP

Na WANG1,2(), Yue-lei CUI1, Liang LUO1, Zi-cong WANG1   

  1. 1.School of Control Science and Engineering,Tiangong University,Tianjin 300387,China
    2.Tianjin Key Laboratory of Intelligent Control of Electrical Equipment,Tiangong University,Tianjin 300387,.
  • Received:2023-10-11 Online:2025-03-15 Published:2025-05-29
  • Contact: Na WANG
  • About author:WANG Na E-mail: wangna@tiangong.edu.cn

摘要:

针对故障诊断中传统时频域法提取特征时易受主观因素影响而导致冗余,且深度学习算法受训练数据影响导致计算复杂性较高的缺点,将时域和频域结合,提出一种基于小波包-无阈值递归图(WP-TRP)的滚动轴承故障诊断方法.首先,提出递减信息熵准则,以克服小波包分解的主观性,获取更准确的时频域特征;在此基础上,引入无阈值递归图思想,充分提取数据初始时域特征,并利用奇异值分解进一步降低冗余特征,提高计算效率.然后,引入海洋捕食者算法来获得支持向量机最优参数,实现故障诊断的准确分类.最后,通过标准滚动轴承数据集仿真验证了所提方法的有效性.

关键词: 故障诊断, 小波包分解, 信息熵, 无阈值递归图, 奇异值分解, 海洋捕食者算法

Abstract:

In fault diagnosis, the traditional time-frequency domain methods are easily affected by subjective factors while being used for feature extraction, so that the redundancy emerges. Deep learning algorithm is highly dependent on training data and has computation complexity. Fault diagnosis method for rolling bearings based on wavelet packet-thresholdless recurrence plot (WP-TRP)is proposed by combining time with frequency domains. Firstly, the decreasing information entropy criterion is developed to overcome the subjectivity of wavelet packet decomposition for acquisition of more accurate time-frequency feature. On this basis, the idea of thresholdless recurrence plot is introduced to extract the initial time domain feature. Moreover, by adopting the singular value decomposition to decrease the redundant feature, the computational efficiency can be increased. Secondly, the marine predator algorithm is introduced to obtain the optimal parameters of supporting vector machine, by which the more accurate classification can be realized. Finally, the effectiveness of the presented method is verified by using the simulation on the benchmark rolling bearing datasets.

Key words: fault diagnosis, wavelet packet decomposition, information entropy, thresholdless recurrence plot(TRP), singular value decomposition, marine predator algorithm

中图分类号: