Scoring Models for Fault Detection in Spacecraft

This paper presents two scoring models for the momentum wheel speed error telemetry data of a satellite. The wheel speed error data are first transformed into orbital segments which form new vectors, and then the vectors are clustered into groups where each group is represented by its center. Based on the centers, two scoring models are formed where one uses distance measurements and the other uses an order-one Markov transition model to score new observed data. Simulation results are presented and discussed