Detection And Estimation Of Actuator Fault In Satellite Attitute Control System

It’s very important to ensure the reliability and safety in satellite attitude control system, which affects that if the project can achieve its goals and if the space mission can be completed smoothly. Fault detection and diagnosis(FDD) plays an important role in this aspect. In this paper, the fault detection and estimation of actuators, such as reaction wheels, in satellite attitude control system is discussed. The main scheme is to detect faults and compensate the system state with the means of Generalized Likelihood Ratio(GLR) test, and estimate the magnitude of faults with Two-Stage Extended Kalman Filter(TSEKF). Following aspects are included in this paper.Firstly, the satellite attitude description is given, and the dynamics and kinematics equations of satellite attitude are built. Based on the analysis of unit faults in satellite attitude control system, the faults of reaction wheels are researched and modeled.Secondly, a linear scheme for actuators’ fault detection and estimation based on original N-order GLR algorithm and Augmented State Kalman Filter(ASKF) is proposed, considering that the model of satellite control system can be approximately linearized under the hypothesis of infinitesimal angular displacement. It is useful for detection and estimation of actuator fault when the attitude is stable, which avoids multiple alerts of one fault, improves the accuracy of estimation, and reduces the convergence time of estimation process.Thirdly, based on bias-separated method and Extended Kalman Filter(EKF), a nonlinear scheme for detection and estimation based on TSEKF is proposed, considering that the model of satellite control system cannot be approximately linearized when attitude maneuvering. It is useful for estimation of actuator fault when the attitude is maneuvering which improves the accuracy of estimation. Furthermore, a nonlinear scheme for detection and estimation based on the method of GLR test and TSEKF is proposed, which has the advantage of GLR algorithm and TSEKF algorithm. The nonlinear GLR-TSEKF algorithm can detect and estimate the magnitude of state and faults accurately and timely, and shorten the convergence time of estimation process greatly.