Periodicity-based kurtogram for random impulse resistance

Abstract

The kurtogram developed from spectral kurtosis has been proven as an efficient tool for extracting the fault impulses in the diagnosis of rolling element bearings and gearboxes. Although the optimal narrowband chosen for demodulation by kurtogram is accurate and effective in experimental environment, this approach is very sensitive to large random impulses that are frequently encountered in industrial applications. The narrowband with maximum kurtosis is always associated with large interferential impulses, rather than the bearing fault. To overcome this limitation, the periodic component to aperiodic component ratio (PAR) is utilized in this article to differentiate the two types of impulses. The novel method named the PAR-based kurtogram focuses on finding the significant frequency band with periodic impulses. The effectiveness of the proposed method is verified by simulations, a test rig of locomotive rolling element bearings, and bearing data from the Case Western Reserve University. The results show that the PAR-based kurtogram improves the robustness to interference from aperiodic impulses significantly, which is very useful for bearing faults diagnosis.