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

发布时间：2018-10-18 14:12

【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。
[Abstract]:Aiming at the problem that the feature of early rolling fault is not obvious and the feature extraction is difficult, A new method to measure the complexity of time series, compound multi-scale entropy (CMSE), is applied to feature extraction of rolling bearing fault vibration signal. CMSE overcomes the shortage of coarse granulation in multi-scale entropy, and the entropy value is consistent and stable. At the same time, to solve the problem that it is easy to collect a large number of samples in intelligent diagnosis of mechanical faults, but it is difficult to label all samples, Laplace support vector machine (LapSVM) is applied to the intelligent diagnosis of rolling bearing faults. On this basis, a rolling bearing fault diagnosis method based on CMSE, sequence forward selection (SFS) feature selection and LapSVM is proposed. Finally, the proposed method is applied to the analysis of test data. The results show that CMSE can effectively extract the fault features of rolling bearings, and when the number of labeled samples is small, it is compared with the support vector machine which only uses labeled samples to learn. The LapSVM method combined with SFS feature selection can improve the accuracy of fault diagnosis by using a large number of unlabeled samples to assist in learning.
【作者单位】：安徽工业大学机械工程学院;
【基金】：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)
【分类号】：TH133.33

【参考文献】