Fault Feature Extraction of Elliptically Shaped Bearing Raceway

The elliptically shaped bearing (ESB) with a rigid, elliptical inner race and a flexible, thin-walled outer race is the most easily damaged core component of harmonic drive. The ESB rotates under cycle load of alternating stress due to its special elliptic structure. Hence, the fault features of ESB such as fatigue spalling and pitting are apt to be concealed by the excitation of impulses caused by alternating between major axis and minor axis. In order to diagnose the fault on raceway surfaces of ESB, a new method of CMWT-FH based on Continuous Morlet Wavelet Transform (CMWT) and FFT-based Hilbert (FH) spectrum analysis is proposed to extract the fault feature. First, the formulas of feature frequency is deduced based on the geometry and kinematics characteristics of ESB; then the CMWT method is employed to decompose the fault signal of ESB; finally, the FH spectrum analysis is performed to extract the feature frequency of faulty ESB from the decomposition signal with the maximum kurtosis in the first several layers. Compared with the traditional FH method, the feature frequency and its sideband and whether the fault location is on the outer or inner ring of faulty ESB can be identified by the proposed CMWT-FH method more accurately.