A fault diagnosis approach for autonomous spacecraft based on transition-system model

For autonomous spacecraft, anomaly detection and fault diagnosis systems play a very important role. To meet the requirement of real-time diagnosis performance for the fault diagnosis system of an autonomous spacecraft, a new method based on transition system model is proposed. In order to improve the efficiency of fault diagnosis, we introduce the separation strategy to separate this process into two stages: the off-line stage at which to finish most computation works of conflict recognition, and the on-line stage at which to generate the fault candidate sets with the minimum matching space. In order to reduce the matching space effectively and avoid enumerating all the fault modes, we build a Conflict-Net to record the relation between the fault modes and the key variables. We have applied this method on a primary power subsystem of a certain satellite, and the result shows it can work more effectively.