In this chapter, the proposed fault detection, isolation and identification (FDII) methodology was employed for fault diagnosis of reaction wheel actuators of the attitude control subsystem (ACS) of a 3-axis stabilized low earth orbit (LEO) satellite. To be able to properly verify and validate the effectiveness of the proposed FDII techniques, first a high-fidelity ACS simulator of a 3-axis stabilized satellite was developed. The ACS simulator consisted of the nonlinear model of the spacecraft attitude dynamics, a high-fidelity nonlinear model of reaction wheels, and the mathematical models of environmental disturbances acting upon the satellite. Furthermore, a decentralized proportional-integral-derivative (PID) control strategy was designed to stabilize the spacecraft attitude, and thus provide the necessary framework for validating the FDII algorithm. Numerous simulation results were presented for evaluating the performance of the proposed fault diagnosis schemes in detecting, isolating, and identifying faults in the reaction wheels of the ACS and in the presence of external disturbances and measurement noise. The simulation results demonstrated the effectiveness of the proposed fault diagnosis schemes. Numerous qualitative and quantitative observations were made regarding the performance capabilities of each FDII scheme. Furthermore, a comprehensive analysis was performed on the robustness of the two FDII schemes with respect to measurement noise. Finally, the performance of the FDII schemes under partial-state measurements was validated using simulations. It was assumed that only the speed of the wheel is measured and the motor current was estimated from the speed measurements using the neural state estimator (NSE).
Application to a satellite's attitude control subsystem
2009
119 Seiten
Aufsatz/Kapitel (Buch)
Englisch
Design of Neural Network for Satellite's Attitude Control Systems
British Library Conference Proceedings | 2012
|