Tooth Root Crack Detection of Planet and Sun Gears Based on Resonance Demodulation and Vibration Separation

Due to the epicyclic motion of planet gears, the vibration picked up from planetary gearboxes is different from that observed from the fixed-axis gearboxes. In particular, the vibration transfer path is time-varying, which makes the time synchronous averaging (TSA) not directly usable to reduce the nonsynchronous interferences for improving the signal-noise ratio (SNR) of vibration. Hence, it is difficult to detect teeth faults of planet and sun gears. In order to address this issue, a novel scheme is proposed in this paper, which combines the vibration separation, equi-angle resampling, and resonance demodulation approaches. In the proposed scheme, the complex envelope of the vibration is extracted by the fast kurtogram algorithm at first. Then, the envelope is equi-angle resampled to eliminate the inevitable speed fluctuations. Subsequently, a window function is used to grab the envelope data series at a fixed angular position related to the ring gear, such that the problems caused by time-varying vibration transfer paths are solved. Then, synthetic envelopes can be constructed by connecting the data in the data blocks according to the teeth mesh sequences of the interesting gear, where a data mapping scheme and a synthetic envelope determining strategy based on maximum kurtosis are introduced. Furthermore, the TSA is applied to the synthetic gear envelope to improve the SNR. Finally, spectral analysis is utilized to expose the characteristic frequencies. The effectiveness of the proposed method is validated by comparative experiments with the normal and faulty planet and sun gears in a planetary gearbox test rig.