Time-Frequency Analysis of Torsional Vibration Signals in Resonance Region for Planetary Gearbox Fault Diagnosis Under Variable Speed Conditions

Translational vibration-based methods have been widely used for machinery fault diagnosis. However, because of the unique gear configuration and complex kinetics, planetary gearbox translational vibration signals have complex modulation features due to gear faults and time-varying vibration transmission paths. This results in complex frequency components of translational vibrations, and adds difficulty to gear fault signature extraction. Under variable speed conditions, the resultant time-variant frequency components and complex sidebands may overlap in frequency domain, thus making it more difficult to pinpoint fault features. To address this issue, torsional vibration signals are exploited, because they are free from the extra modulation effect due to time-varying transmission paths and have simpler frequency contents. Gear faults generate impacts, thus exciting resonances and leading to modulations on resonances. Therefore, torsional resonance frequency band is concentrated to extract gear fault information. The time variant but symmetric sideband characteristics in the resonance region are derived based on the explicit time varying amplitude modulation and frequency modulation signal model. Resonance frequencies are identified under variable speed conditions by virtue of their independence on running conditions. Furthermore, time frequency analysis is utilized to extract time-variant gear fault frequencies. The proposed method is validated using both numerical simulation and lab experimental data. Localized faults of the sun, planet, and ring gears are diagnosed under variable speed conditions.