Fault Frequency Analysis and Feature Extraction Based on CMWT-FH of Elliptically Shaped Bearing

Elliptically shaped bearing (ESB) is the most easily damaged core component of harmonic drive under cyclic load of alternating stress. In order to diagnose the fault on raceway surface of ESB, the formulas of fault feature frequency is derived by using the kinematics equations of ESB. Considering the effect of characteristic excitation of vibration produced by alternating between long axis and short axis, the fault features of ESB are easily concealed. A method called CMWT-FH which is based on continuous Morlet wavelet transform and FFT-based Hilbert spectrum analysis is proposed for fault feature extraction. Compared with the conventional FFT-based Hilbert method, the proposed CMWT-FH method can identify fault feature frequency and its sideband of ESB more easily and accurately. It is shown that, the fault feature frequency of the ESB outer-race and its harmonics are surrounded by sidebands with a spacing equal to the modulating frequency of rotation frequency. The fault feature frequency of the ESB inner-race and its harmonics are surrounded by sidebands with a spacing equal to the modulating frequency of second-order rotation frequency.