Modulation Signal Bispectrum Optimization for Fault Diagnosis of Intershaft Bearings

Abstract

Modulation signal bispectrum (MSB) has attracted much attention due to its sensitivity to nonlinear characteristics related to rolling bearings’ faults. However, there are still two challenges in applying MSB to the diagnosis of aero-engines’ intershaft bearing: 1) the variable working conditions make the vibration signals show strong non-stationarity; 2) unlike the ordinary rolling bearings, the fault features in vibration of intershaft bearings are extremely weak and difficult to extract. Thus, an optimized MSB (OMSB) is proposed for diagnostics of intershaft bearings under variable speed conditions. This paper mainly discusses two core issues: 1) an enhanced filter is presented with the idea of whitening and deconvolution, so as to highlight weak fault features from vibration signal with strong interference; 2) an MSB algorithm in rotation speed difference (RSD) domain is established to deal with the non-stationarity caused by various operating speeds, and a lifted modulation spectrum is constructed to exhibit modulation orders in the signal, i.e., the fault feature orders can be clearly extracted. Experimental signals are analyzed by OMSB, and the results prove that the method can suppress the interference of irrelevant components and effectively extract the intershaft bearing’s weak fault features. Comparisons also show the superiority of the proposed OMSB.