Multi-Band Torsional Vibration Amplitude Demodulation for Planetary Gearbox Fault Diagnosis Under Time-Varying Speed Conditions

Torsional vibration signal analysis is promising for diagnosing gear faults in planetary gearboxes. However, early-stage fault signatures are relatively-weak and may be ignored or wrongly identified in conventional demodulation analysis, especially under time-varying speed conditions. Two main contributions are made in this paper. Firstly, an analytical model of planetary gearbox torsional vibration signal is established, by jointly considering the gear mesh vibration, torsional resonance, and other time-invariant components. The gear fault characteristics under time-varying speeds are accordingly revealed explicitly. Furthermore, inspired by the unique time-frequency structure of the torsional vibration signal, the multi-band torsional vibration amplitude demodulation method is proposed for gear fault feature extraction. The proposed method combines the gear fault features in multiple characteristic frequency bands in torsional vibration signal, so that the influences from adjacent carrier frequencies, independent frequency harmonics and random noises can be eliminated in amplitude demodulation. As a result, the obtained amplitude demodulated order spectrum outperforms conventional envelope order spectrum and narrow-band amplitude demodulated order spectrum in revealing relatively weak gear fault features. Both numerical simulation and laboratory experiments are conducted, to demonstrate the correctness of the established analytical model and the advantages of the proposed multi-band torsional vibration amplitude demodulation method. Localized fault on the sun, planet and ring gears are successfully diagnosed.