Weak fault feature extraction of rolling bearing based on secondary clustering segmentation and Teager energy spectrum

The key to identify local faults of rolling bearing is to accurately extract weak periodic fault features from noisy vibration signals. Aiming at this problem, a method to extract weak fault features of rolling bearing based on secondary clustering segmentation and Teager energy spectrum was proposed here. Firstly, the frequency spectrum of a fault signal was obtained with Fourier transform, and the clustering segmentation was done for this spectrum using the fuzzy C-means algorithm. Then, the inverse Fourier transform was done for each frequency segment to calculate kurtosis values of time domain signals in different frequency bands, and the time domain signal corresponding to the frequency band with the maximum kurtosis was selected as the signal to be filtered. The secondary clustering segmentation and the inverse Fourier transform were done for the filtered signal, the frequency band with the maximum kurtosis was taken as the passband filter one to further eliminate effects of noise and natural periodic components. Finally, Teager energy operator was used to demodulate the obtained time domain fault signal to acquire feature frequencies of bearing weak faults. Simulation analysis and test verification results showed that the proposed method can accurately and effectively extract weak fault features of rolling bearing. © 2020, Editorial Office of Journal of Vibration and Shock. All right reserved.