Designing an adaptive robust observer for underactuation fault diagnosis of a remote sensing satellite

Abstract

SummaryThis paper mainly aims at the underactuated case diagnosis due to an actuator failure for a remote sensing satellite. We consider healthy initial conditions for the actuators to investigate the underactuation fault in an attitude tracking control problem. Reaction wheels are the onboard actuators, and a new fault injection mechanism on the speed characteristic is introduced using the multi‐fault function. The fault detection and diagnosis strategy is based on the sliding mode and adaptive sliding mode observers in a finite‐time decision window. The robust observer and adaptive threshold are designed to detect the fault, and the adaptive robust observer estimates the fault model. The residual signal is evaluated by forming the window with fixed thresholds, and in the decision steps, priority is given to the underactuation fault diagnosis. System performance is presented for two actuator fault scenarios in a snapshot imaging mode to understand this case clearly. Numerical simulations close to in‐orbit conditions confirm the satisfactory performance of the proposed strategy that underactuated case is diagnosed by injecting stuck and idle faults around 2.19 and 3.52 s. The times obtained in this study are guaranteed with a 3‐sigma confidence level statistical characteristic. Furthermore, the results of the present work are validated using an air‐bearing experimental test bed to illustrate a more realistic behavior of an underactuated satellite.