Application of lifting wavelet fractal strategy in feature extraction of motor bearing electrolytic corrosion fault

Electrolytic corrosion is a common physical phenomenon in service process of induction motor bearings. The occurrence of electrolytic corrosion makes motor bearing surface easier to have local damage, such as, pitting and spalling. Here, a lifting wavelet fractal strategy was proposed to extract impact fault features of bearing faults from multi-component coupled vibration signals. By using the proposed strategy to post-process adjacent subspace generated using the classical redundant lifting wavelet packet decomposition, a new hidden wavelet packet scale was generated to effectively make up for shortages of the dyadic wavelet packet analysis method in extracting features of transition zone. The central polarization multiresolution analysis of vibration signals was realized with the central nested set of hidden wavelet packet subspace. This method could also inherit many excellent characteristics of redundant lifting wavelet packet decomposition, such as, accurate linear phase and translation invariance to further enhance the analysis and extraction ability of non-stationary impact features in vibration signals. The proposed method was applied in vibration signal analysis of induction motor bearing of a certain type of skin pass mill. Periodic impact features characterizing bearing mechanical fault were extracted from a certain hidden wavelet packet subspace generated using the lifting wavelet fractal strategy. After shutdown and maintenance, it was verified that the fault feature is caused by the existing electrolytic corrosion fault in bearing. By comparing the feature extraction effect of the proposed method with those of the mainstream analysis methods based on dual-tree complex wavelet, it was verified that the lifting wavelet fractal strategy has a more comprehensive bearing fault feature analysis capability. © 2022 Chinese Vibration Engineering Society. All rights reserved.