A Robust Fault Detection Algorithm Using q-LPV Interval Observer Together with an Accommodation System Based on Actuators Reconfiguration

This paper presents robust fault detection based on adaptive thresholds for a satellite described by means of a linear parameter varying (LPV) model. Adaptive threshold is generated using an interval LPV observer that produces a band of predicted outputs considering both bounded parametric modeling errors and measurements noises. This strategy propagates the effect of satellite parameter uncertainties into the alarm limits and so enhances the robustness of fault detection system at the decision making stage. The developed method can minimize the missing alarm rates due to modeling uncertainties; also this approach detects small or incipient faults more effectively than the classical robust fault detection algorithms with constant thresholds. In this paper also an accommodation system based on reconfiguration of available actuators has been designed. Accordingly, after isolation of faulty reaction wheels, they turned off and replaced by suitable magnetic torquers. Therefore, despite occurrences of several failures in the ACS, attitude control error was kept limited. In the presented simulation results, some scenarios are developed to verify the performance of the proposed method.