基于双树复小波包自适应Teager能量谱的滚动轴承早期故障诊断

摘要

In view of that the incipient fault feature informations of rolling bearings are difficult to identify,and the parameters setting of band-pass fiher depends on the user's experience,which makes the resonance frequency band can't be effectively determined and extracted,the concept of amplitude entropy of frequency band was introduced.Combining the techniques of wavelet transform and Teager energy spectrum,a rolling bearing incipient fault feature extraction method was proposed and named as the method of dual-tree complex wavelet packet transform adaptive Teager energy spectrum.Original fault signals were decomposed into several different frequency components through wavelet packet transform,and the amplitude entropy of each frequency sub-band was calculated.Then the entropies were arranged in ascending order and in turn were adopted as a threshold to extract the sub bands with the entropy values greater than the threshold values.Based on the kurtosis index,the optimal threshold and the best dual tree complex wavelet packet decomposition levels were determined,thus,the resonance band was extracted adaptively and effectively.Finally,the Teager energy spectrum analysis of the resonance band was performed to identify the frequency of the bearing fault.Through the signal simulation and engineering experiment data analysis,the effectiveness of the proposed method was verified.%针对滚动轴承早期故障特征信息难以识别以及带通滤波器参数设置依赖使用者经验等造成共振带不能有效确定并自适应提取的问题,提出了频带幅值熵的概念.在此基础上,将双树复小波包变换和Teager能量谱结合,提出了基于双树复小波包变换自适应Teager能量谱的早期故障诊断方法.该方法首先利用双树复小波包将采集到的振动信号进行分解,并计算各子带的频带幅值熵.然后将熵值按升序排列后依次作为阈值,提取频带幅值熵大于或等于阈值的子带,依据峭度指标确定最佳熵阈值和双树复小波包最佳分解层数,从而自适应并有效地提取出共振带.最后对共振带进行Teager能量谱分析,即可从中准确地识别出轴承的故障特征频率.通过信号仿真与工程实验数据分析验证了该方法的有效性与优越性.