Uncertainty Quantification in Gear Remaining Useful Life Prediction Through an Integrated Prognostics Method

Accurate health prognosis is critical for ensuring equipment reliability and reducing the overall life-cycle costs. The existing gear prognosis methods are primarily either model-based or data-driven. In this paper, an integrated prognostics method is developed for gear remaining life prediction, which utilizes both gear physical models and real-time condition monitoring data. The general prognosis framework for gears is proposed. The developed physical models include a gear finite element model for gear stress analysis, a gear dynamics model for dynamic load calculation, and a damage propagation model described using Paris´ law. A gear mesh stiffness computation method is developed based on the gear system potential energy, which results in more realistic curved crack propagation paths. Material uncertainty and model uncertainty are considered to account for the differences among different specific units that affect the damage propagation path. A Bayesian method is used to fuse the collected condition monitoring data to update the distributions of the uncertainty factors for the current specific unit being monitored, and to achieve the updated remaining useful life prediction. An example is used to demonstrate the effectiveness of the proposed method.