Mission reliability analysis and optimal stopping planning oriented to shock-induced-degrading process under heterogeneity

Malfunctions of safety-critical assets during missions can lead to severe consequences and financial losses. In order to manage mission risk and improve mission reliability, a mission management policy, namely mission abort, is usually adopted based on real-time degradation signals. This paper focuses on the abort management problem of a safety-critical asset that performs a mission during a certain duration of time. The system encounters continuous degradation accumulation due to random shocks, and fails when degradation magnitude reaches a certain level. Considering the process heterogeneity, the shock arrival intensity and core degradation parameters are stochastic and need to be accurately estimated based on real-time health observation. We develop the abort management policy integrating parameter learning, modeled as a Markov decision process to minimize the expected total cost. We study the structural properties of the value function and cast the optimal mission abort policy into an optimal control limit policy. The applicability over cost management and reliability support is exemplified via a numerical experiment.