Remaining useful life prediction for a unit under time-varying operating conditions

The reliability of a unit during field use is important in many critical applications such as the power generation industry. In particular, the propensity of achieving zerodowntime emphasizes the need for remaining useful life (RUL) prediction for an individual unit. The task becomes very challenging when the unit is subject to time-varying operating conditions. This paper provides an enhanced approach for predicting the RUL of a unit under time-varying operating conditions using the results from both accelerated degradation testing (ADT) and in-situ condition monitoring. Specially, the degradation process of the unit is modeled as a Brownian motion which evolves in response to the operating conditions. The model is combined with the in-situ degradation measurements of the interested single unit and the operating conditions through a Bayesian updating technique. An analytical solution technique is developed for RUL prediction under piecewise constant operating conditions. A simulation experiment is provided to demonstrate the predictive capability and accuracy of the proposed approach.