A stochastic collocation approach for efficient integrated gear health prognosis

Uncertainty quantification in damage growth is critical in equipment health prognosis and condition based maintenance. Integrated health prognostics has recently drawn growing attention due to its capability to produce more accurate predictions through integrating physical models and real-time condition monitoring data. In the existing literature, simulation is commonly used to account for the uncertainty in prognostics, which is inefficient. In this paper, instead of using simulation, a stochastic collocation approach is developed for efficient integrated gear health prognosis. Based on generalized polynomial chaos expansion, the approach is utilized to evaluate the uncertainty in gear remaining useful life prediction as well as the likelihood function in Bayesian inference. The collected condition monitoring data are incorporated into prognostics via Bayesian inference to update the distributions of uncertainties at given inspection times. Accordingly, the distribution of the remaining useful life is updated. Compared to conventional simulation methods, the stochastic collocation approach is much more efficient, and is capable of dealing with high dimensional probability space. An example is used to demonstrate the effectiveness and efficiency of the proposed approach.

► The proposed gPC collocation method provides a much more efficient way for integrated gear prognostics. ► Uncertainty sources are categorized to highlight necessity of gPC. ► gPC collocation calculates Bayes' likelihood and RUL density in an efficient way. ► The method reduces computing time significantly compared to random sampling. ► Life prediction converges and the uncertainty is reduced as new data are obtained.