Optimal Mission Abort Planning for Partially Observable System

Abstract

Mission abort is proved to be an effective approach to improve the survivability of systems. The existing researches on mission abort generally believe that the degradation state of the system can be completely observed. The prevailing research on mission abort typically assumes fully observability of the system’s state, but in fact, many systems cannot detect the health state completely during the execution of a mission. This paper study the control policy that timely aborts the mission system whose state information cannot be completely monitoring. A partially observable Markov decision process is used to solve the problem by minimizing the cost probably incurred by the failure of mission and system. Some properties of the model are presented and proved.