中国机械工程

• 智能制造 • 上一篇    下一篇

基于复合多尺度熵与拉普拉斯支持向量机的滚动轴承故障诊断方法

代俊习;郑近德;潘海洋;潘紫微   

  1. 安徽工业大学机械工程学院,马鞍山,243032
  • 出版日期:2017-06-10 发布日期:2017-06-13
  • 基金资助:
    国家自然科学基金资助项目(51505002);
    安徽省高校自然科学研究重点资助项目(KJ2015A080);
    安徽工业大学研究生创新研究基金资助项目(2016061)

Rolling Bearing Fault Diagnosis Method Based on Composite Multiscale Entropy and Laplacian SVM

DAI Junxi;ZHENG Jinde;PAN Haiyang;PAN Ziwei   

  1. School of Mechanical Engineering, Anhui University of Technology,Ma'anshan,Anhui,243032
  • Online:2017-06-10 Published:2017-06-13

摘要: 针对早期滚动故障特征不明显和特征提取难等问题,将一种新的衡量时间序列复杂性的方法——复合多尺度熵(CMSE)应用于滚动轴承故障振动信号的特征提取。CMSE克服了多尺度熵中粗粒化方式的不足,得到的熵值一致性和稳定性好。同时,针对机械故障智能诊断中收集大量的样本比较容易而要对所有的样本进行类别标记却较为困难这一问题,将拉普拉斯支持向量机(LapSVM)应用于滚动轴承故障的智能诊断中。在此基础上,提出了一种基于CMSE,序列前向选择(SFS)特征选择和LapSVM的滚动轴承故障诊断方法。最后,将提出的方法应用于试验数据分析,结果表明:CMSE能够有效地提取滚动轴承的故障特征;当有标记样本的数量较少时,与仅使用有标记样本进行学习的支持向量机相比,结合SFS特征选择的LapSVM方法利用大量的无标记样本进行辅助学习,可以显著提高故障诊断的正确率。

关键词: 多尺度熵, 复合多尺度熵, 支持向量机, 拉普拉斯支持向量机, 故障诊断

Abstract: Since the unclear of early fault of rolling bearings and it was difficult to extract the features from the mechanical systems, a new judging time series complexity testing method called composite multiscale entropy (CMSE) was applied to extract the fault features from the vibration signals of rolling bearings. CMSE overcome the defects of coarse-graining in MSE and was an effective method for measuring the complexity of time series with better consistency and stability. Besides, as it was easy to collect a large number of samples, but difficult to label them in mechanical fault intelligent diagnosis, the LapSVM was applied to the intelligent fault diagnosis of rolling bearings. Then a new fault diagnosis method for rolling bearings was proposed based on the CMSE, sequential forward selection and LapSVM. Finally, the experimental data were analyzed based on the proposed method. The results show that the fault features of rolling bearings are extracted effectively by CMSE, compared with SVM that may only be trained by the labeled samples, the LapSVM combining with sequential forward selection for feature selection and studying from a large number of unlabeled samples may significantly improve the accuracy of fault diagnosis for fewer number of labeled samples.

Key words: multiscale entropy(MSE), composite multiscale entropy, support vector machine(SVM), Laplacian support vector machine(LapSVM), fault diagnosis

中图分类号: