Remaining Useful Life Estimation for Degrading Systems under Time-Varying Operational Conditions

This paper presents a method to estimate the remaining useful life for degrading systems operating under time-varying operational conditions. This method considers a non-monotone degradation process that is significantly affected by stochastically-evolving operational conditions. The failure zone instead of the deterministic failure threshold is used to identify the failures, and different operational conditions may have different failure zones. The method is developed using a semi-Markov decision process framework, and illustrated by the prognostic problem of the 2008 PHM Conference Data Challenge Competition.