A Fault Diagnosis Method based on Singular Spectrum Decomposition and Envelope Autocorrelation for Rolling Bearing

Aiming at the difficulty in extracting periodic impulse fault feature of rolling bearing under strong background noise and harmonic component interference, a fault diagnosis method based on singular spectrum decomposition and envelope autocorrelation for rolling bearing is proposed. First, a novel signal decomposition method, singular spectrum decomposition (SSD), is used to decompose the vibration signal into a number of singular spectrum components (SSC). According to the kurtosis of the singular spectrum component and the correlation coefficient with the original signal, the singular spectrum component that contains the bearing fault information is selected. The Hilbert envelope is made to the selected singular spectrum component, and then the autocorrelation is performed on the envelope signal to further extract periodic impulse component in the signal. Finally, the spectral characteristics of the autocorrelation function are analyzed and the fault type of bearing can be accurately identified by the prominent frequency components. The simulated analysis and experimental analysis results prove the validity of the proposed method.