Bearing and gear compound faults diagnosis based on the Vold-Kalman generalized demodulation under time-varying speeds

The spectral kurtosis based filtering algorithms cannot work well for distinguishing the differences between the two resonance bands excited by bearing and gear multi-fault. Hence, the traditional band-pass based methods were given up, and based on the extraction, reset and quantitative expressing of time-varying nonstationary fault characteristics, a new strategy was proposed for the bearing and gear multi-fault diagnosis under time-varying speeds based on the Vold-Kalman generalized demodulation. The essence of the proposed method is as follows: extract the time-varying nonstationary fault feature components from the envelope of raw signals using the Vold-Kalman filtering algorithm, reset the extracted filtering components by the generalized demodulation transform, quantitatively express these demodulated components using the FFT, and lastly determine the faults by comparing the theoretical frequency points and the peaks in spectrums. The fault characteristic frequency functions and their corresponding phase functions and frequency poisnts were calculated according to the rotational frequency equation of the target bearing and the mechanical structure parameters. The analysis results of some practical simulated and experimental multi-fault signals under time-varying rotating speeds validate that the proposed method is reliable to diagnose faulted bearings and gears without any frequency band location and angular resampling. The contrastive analysis results further verify its superiority in the interference terms elimination. © 2019, Editorial Office of Journal of Vibration and Shock. All right reserved.