For the crack-rubbing coupled fault in rotor system

a fault diagnosis method was proposed. Firstly

the rubbing fault

crack fault and crack-rubbing coupled fault were simulated on rotor test rig. The acceleration signals of each fault were collected. The integration of information entropy distance was calculated for fault diagnosis using singular spectrum entropy in time-domain

power spectrum entropy in frequency domain

wavelet energy entropy and wavelet spatial characteristics spectral entropy in frequency-time domain. The case study showed that these four information entropies could evaluate rotor vibration state comprehensively. Information entropy distance curves in multiple points and multiple speeds can effectively improve the accuracy and distinguish single and coupled faults. The information entropy distance between the test signal and its corresponding fault is the least and the information entropy distance curve is located in the bottom of the axis

so its fault category can be diagnosed.