Mission Reliability Evaluation and Integrated Repair-Stopping Planning for Safety-Critical Equipment

Failure of safety-critical equipment during mission implementation can lead to serious consequences and financial losses. Performing maintenance and mission abort management simultaneously based on real-time degradation data can help alleviate failure risks and improve mission reliability. In this study, we focus on a safety-critical equipment that degrades as a Wiener process and fails when the degradation exceeds a pre-defined threshold. The drift parameter of Wiener process is unknown and can only be inferred from degradation signals by means of maximum likelihood estimation. The problem is constituted as a Markov decision process. We conduct an in-depth study of the structural properties, establishing the existence and monotonicity of both repair and abort thresholds. We transform the joint maintenance-mission abort problem into an optimal threshold learning problem. Numerical examples are employed that illustrate the model applicability.