Induction of on-board fault management for remote and autonomous systems

The author presents a novel approach for producing the diagnostic logic that could be used as a basis for onboard satellite fault management system (FMS) design. He uses a qualitative model, exhaustive simulation, and rule induction to produce rules that cover all failure modes. The approach is applied to the attitude and orbital control subsystem (AOCS) of a communications satellite. Two stages of model building are performed: (i) without any FMS logic, and (ii) with FMS logic and recovery. Through qualitative modeling the AOCS, it is demonstrated that an induced knowledge-based FMS can be incorporated into a model and exhibit fully autonomous behaviour in the presence of faults.<
>