Fault Feature Enhancement Method for Rolling Bearing Fault Diagnosis Based on Wavelet Packet Energy Spectrum and Principal Component Analysis

The acquired vibration signal is usually unstable once rolling bearing damage occurs, which results in inaccurately detecting the fault features of rolling bearing by time-domain or frequency-domain analysis. A fault diagnosis method which uses the wavelet packet energy spectrum and principal component analysis (PCA) to diagnose the faults of rolling bearing is presented. The wavelet packet decomposition algorithm is used to decompose and refine the vibration signals in different frequency ranges. The energy spectra in the focused frequency ranges are calculated after the vibration signal is decomposed by wavelet packet decomposition. PCA is performed to decrease the dimension of the energy spectrum and reduce the noise interference, thus enhancing the extracted fault feature without the noise interference. And then the different fault types of rolling bearing are classified by two types of clustering algorithms, i.e., hierarchical clustering analysis (HCA) and fuzzy c-means (FCM). The results show that the fault types can be correctly identified by both cluster algorithms. The example verification indicates that the proposed method can be used to effectively extract the useful fault features in the vibration signal and identify the fault types exactly. This provides a feasible method for diagnosing a machine with some similar faults. © 2019, Editorial Board of Acta Armamentarii. All right reserved.