Feature Extraction of Weak Fault for Rolling Bearing based on Spectral Kurtosis and MOMEDA

Abstract

Aiming at the problem that the early periodic transient impulse of rolling bearings is not obvious and the spectral kurtosis is poorly analyzed under low signal-to-noise ratio， a method of extracting the weak fault features of rolling bearing based on the combination of multipoint optimal minimum entropy deconvolution adjusted （MOMEDA） and spectral kurtosis is proposed. Firstly， MOMEDA is used as the prefilter to reduce the noise of weak fault impulse signal with strong noise and highlight the periodic impulse component in the signal. Then， through spectral kurtosis analysis， the denoised signal is filtered under the optimal center frequency and bandwidth. Finally， the fault characteristic frequency of bearing signal can be accurately obtained by Hilbert envelope spectrum analysis of filtered signal. The simulation and experimental results show that the method can effectively enhance the periodic transient impulse characteristics of vibration signals and extract the early weak fault characteristics of rolling bearing.

Keywords

• Rolling bearing
• Multipoint optimal minimum entropy deconvolution adjusted
• Spectral Kurtosis
• Weak fault
• Feature extraction