A novel Fast Entrogram and its applications in rolling bearing fault diagnosis

Abstract Effectively identifying the health status of rolling bearings can reduce the maintenance costs of rotating mechanical components. With the development and improvement of various signal processing theories, the mode of extracting fault information from the frequency domain has gradually replaced the mode from the time domain. As a traditional spectrum segmentation analysis method, Fast Kurtogram can adaptively extract frequency bands that may contain fault information to diagnose faults. However, the frame of the center frequency and bandwidth obtained by the 1/3 binary tree filter bank segmentation method adopted by the Fast Kurtogram is fixed. This paper proposed a new method of segmenting the spectrum and accurately filtering fault information from the frequency domain----Fast Entrogram. The fluctuation state of the Fourier spectrum is of key importance in distinguishing the distribution of different components in the signal at each frequency. After the Fourier transform of the spectrum is intercepted and reconstructed, the minimum points of the new sequence can separate different components in the signal. Subsequently, the frequency slice function is used to extract each frequency band to obtain better filtering effects than the finite impulse response filter. Finally, the proposed novel correlation spectral negentropy is sensitive to periodic pulses and can be used to screen the component that contains the most fault information. The simulation results show that the proposed Fast Entrogram can effectively extract periodic pulses. It is verified by experimental signals that the method can be applied to fault diagnosis of bearing inner and outer rings.