An Integrated Prognostics Method for Failure Time Prediction of Gears Subject to the Surface Wear Failure Mode

Surface wear is one of the main failure modes that gears suffer from due to the sliding contact in the mesh process. However, the existing gear prognostics methods mainly focused on the fatigue cracking failure mode and the existing prediction methods considering surface wear are physics based without utilizing condition monitoring data. This paper proposes the first integrated prognostics method for failure time prediction of gears subject to the surface wear failure mode, utilizing both physical models, i.e., Archard's wear model and condition monitoring data, i.e., inspection data on gear mass loss in this study. By noticing the importance of the wear coefficient in Archard's model, the proposed method can result in a more accurate value of the wear coefficient so that the wear evolution in the future is forecasted with more accuracy. To achieve this, a Bayesian update process is implemented to incorporate the mass loss observation at an inspection point to determine the posterior distribution of the wear coefficient. With more mass loss data available, this posterior distribution gets narrower and its mean approaches the actual value of the coefficient. To use Archard' model, the gear mesh geometry and Hertz contact theory are applied to compute the sliding distance and the contact pressure for different points on the tooth flank. The proposed method is validated using run-to-failure experiments with a planetary gearbox test rig.