基于SVD降噪和谱峭度的滚动轴承故障诊断  

Fault Diagnosis for Rolling Bearings Based on SVD De-Noising and Spectral Kurtosis

作　　者：孟智慧[1] 王昌[1] 

机构地区：[1]内蒙古科技大学机械工程学院,内蒙古包头014010

出　　处：《轴承》2013年第10期52-55,共4页Bearing

基　　金：内蒙古自然科学基金资助项目(2010MS0714)

摘　　要：提出将奇异值分解(SVD)和谱峭度结合的滚动轴承故障诊断方法。首先,将原始振动信号构造为Hankel矩阵,进行奇异值分解,并利用奇异值差分谱进行有效的消噪;然后,利用谱峭度所得的峭度图选择最佳的带宽和中心频率对降噪后的信号进行带通滤波;最后,对滤波后的信号进行平方包络和Fourier变换得到包络解调谱,即可实现故障特征的提取。对滚动轴承故障试验信号的分析表明,该方法可以有效地提取故障特征频率,实现故障诊断。A fault diagnosis method for rolling bearings is proposed based on singular value decomposition （SVD） and spectral kurtosis. Firstly, the original vibration signal is constructed to Hankel matrix and decomposed by SVD, and then the noise is effectively eliminated by singular value difference spectrum. Secondly, the best bandwidth and center frequency are selected by kurtosis diagram of spectral kurtosis, and the band - pass filter is able to be done for signal after de - noising. Finally, the envelope demodulation spectrum of filter signal is able to be obtained by square enve- lope and Fourier transform, and then the fault feature is extracted. The analysis of the experiment fault signal of rolling bearings shows that the fault feature frequency is able to be effectively extracted and the fault diagnosis is realized.

关 键 词：滚动轴承 故障诊断 奇异值分解 HANKEL矩阵 差分谱 谱峭度 

分 类 号：TH133.3[机械工程—机械制造及自动化] TH165