Vibration analysis of planet gear bore-rim failure using enhanced planet time synchronous averaging

It is a worldwide challenge to detect planet gear bore-rim cracking using vibration data analytics, especially for large helicopters like the Airbus Super Puma that has experienced two fatal accidents since 2009 due to planet gear bore-rim cracking. For helicopter safety, it is of vital importance to equip helicopter vibration health monitoring systems with the capability of detecting and monitoring the progression of planet gear bore-rim cracking. When a crack reaches sufficient size in the rim of the gear, the weakening stiffness caused by the crack allows the planet gear teeth near the crack to distort as they come into and out of mesh, resulting in vibration impacts. A key challenge for detecting a planet gear bore-rim crack lies in isolating these impacts from the dominating gear mesh harmonics. In this paper, we propose a method of vibration data analytics based on enhanced planet gear time synchronous averaging (EP-TSA). The method takes advantage of visualizing the impacts from planet-ring and planet-sun gear meshes over the planet-ring hunting tooth period of the planetary gear system. We firstly process the raw vibration signal using time-synchronous averaging with respect to the relative rotation of the planet gear over a number of hunting tooth periods. The method employs a novel resampling scheme, termed boosted synchronous resampling scheme (BSRS), which facilitates simultaneous removal of multiple shaft and gear mesh harmonics and their intermodulation components using the spectrum of the averaged signal in order to derive a residual signal. By construction, the BSRS facilitates three-dimensional (3-D) plotting of the squared envelope of the residual signal, comprising multiple individual revolutions of the planet gear over the hunting tooth period. With crack-induced impacts visualized in 3-D, we can identify those planet gear revolutions where the crack-induced impacts are in close proximity to the vibration transducer to obtain a final feature signal reflecting the characteristics of the bore-rim cracking. We call the final feature signal the enhanced planet residual signals (EPRS), which is those individual residual signals at the selected planet gear revolutions with the largest kurtoses above a certain threshold. Where possible we should look at multiple planet gear revolutions to reinforce the detection of vibration impacts. Lastly, we confirm the detection of planet gear bore-rim cracking by the consistent pattern of two spikes in each planet revolution that are associated with the planet-ring and planet-sun gear meshes. The method is validated using vibration data generated from a small industrial planetary gearbox with simulated cracks inserted in the bore-rim from one side of a planet gear. The results show that the BSRS can be very effective in removing multiple sets of shaft harmonics and their intermodulation sidebands. We have demonstrated that the visualization tool can provide strong evidence of the crack-induced impacts. In comparison to other methods our results have shown that we can detect the planet gear bore-rim crack more effectively using the enhanced planet time synchronous averaging method.