Mission-based fault reconfiguration for spacecraft applications

A study was conducted to propose a fault-reconfiguration scheme that considered the uncertainty in fault detection and the possibility of failure of a reconfiguration action, along with incorporating the mission objectives and current policy execution state into the decision-making process. The value iteration approach was used to compute an optimal reconfiguration policy off-board that maximized the expected discounted performance reward over a finite horizon, which was a function of the Markov decision process (MDP) state. The problem addressed in this study involved developing a framework for constructing fault-reconfiguration policies that were optimal with respect to minimizing the possibility of incorrect reconfiguration and maximizing the possibility of achieving mission objectives while accounting for the possibility of failure of the reconfiguration action.