基于小波包能量矩阵的轴承信号特征提取

Feature extraction of bearing vibration signals based on wavelet energy matrix

为通过振动信号识别轴承的工作状态,结合小波包变换和矩阵特征值理论,提出了一种新的轴承信号特征提取方法。引入了能量值方法,对小波包分解信号进行分层分段能量计算,组成能量特征矩阵,求得矩阵特征值;定义基于特征值的振动信号特征参数,并探讨了特征参数与轴承运行状态间的联系。最后在特征提取基础上,提出了故障早期模式识别的对应系数相乘方法。结果表明:最大值特征参数能够敏感的反映轴承工作性能的变化,可作为轴承状态监测特征量;对应系数相乘法可以作为故障部位诊断的有效方法。

In order to recognize the operating states of bearings, a new approach for extracting features of vibration signals was proposed using the theory of wavelet transformation and matrix eigenvalues. The method of energy was introduced and calculated using segmented datas, then a energy matrix was built and the matrix eigenvalues were gained. Characteristic parameters of bearing vibration signals were defined based on the matrix eigenvalues. The relationship between the characteristic parameters and the operating states of bearing were discussed. At last, a method of early-age fault diagnosis multiplying the wavelet coefficients was proposed on the basis of feature extraction. It was shown that the max characteristic parameters can sensitively reflect the variation of working performance of bearings, therefore, they can be taken as parameters to inspect the conditions of bearings ; multiplying wavelet coefficients can be an efficient method of bearing fault diagnosis.