双树复小波和双谱在轴承故障诊断中的应用

Applications of Dual-tree Complex Wavelet Transform and Bi-spectrum in Roller Bearing Fault Diagnosis

针对非线性、非平稳且包含强烈的噪声的轴承故障振动信号难以有效提取故障特征信息进行故障识别的问题,提出基于双树复小波变换和双谱的故障诊断方法。首先利用双树复小波变换将故障轴承振动信号分解为若干个不同频带的分量,选择出包含故障特征的分量;然后对该分量进行希尔伯特包络解调;最后对包络信号求其双谱图,从而有效地提取出故障信号的特征频率,准确地进行故障识别。滚动轴承故障实验和工程应用表明,该方法能有效地提取故障轴承的故障特征频率,并且几乎可以完全抑制噪声,验证了方法的可行性和有效性。

It is difficult to extract fault characteristic signals from the nonlinear, non- stationary and strong noiseincluded fault vibration signals of roller bearings for fault identification. In this paper, a new fault diagnosis method isproposed based on dual- tree complex wavelet transform (DT-CWT) and bi- spectrum. Firstly, the fault vibration signal isdecomposed into several components with different frequency-band through DT-CWT, and the components which containthe fault feature information are selected. Then, the Hilbert envelope demodulation is used to the components. Finally, the bispectrumdiagram of the envelop signals is acquired and the fault signal frequency feature can be effectively extracted toidentify the fault accurately. The results of the roller bearing fault tests and engineering application show that the faultfeature frequency of the fault roller bearing in operation can be extracted accurately by this method and the noise can bealmost completely suppressed. Thus, the feasibility and effectiveness of this method are verified.