This paper proposes a multiple fault diagnosis (MFD) scheme based on an observer method for satellite attitude control systems (ACSs) subject to nonlinearity and external disturbances of the system. The essential idea is to develop a fault diagnosis scheme and design the corresponding observers for satellite ACSs. The nonlinearity, space external disturbances, sensor uncertainties, and multiple faults problem of the satellite ACS are all taken into account. The proposed MFD scheme is developed at two different levels. First, at the system level, two nonlinear observers based on analytical redundancy are designed for the MFD of the satellite ACS; this level roughly reveals the fault source. Then, at the component level, a bank of sliding mode observers activated by the results of the previous diagnosis is designed to precisely diagnose multiple faults of actuators in the satellite ACS; this level further precisely reveals the fault source. Both the nonlinear observers and the sliding mode observers are confirmed to be asymptotically stable via Lyapunov stability theory. Finally, numerical simulations of a satellite ACS are performed to illustrate the effectiveness of the proposed MFD scheme.
This study investigates the problem of fault diagnosis observer design in the finite frequency domain for discrete‐time Lipschitz nonlinear systems simultaneously subjected to actuator and sensor faults and unknown input disturbances. An augmented descriptor system is developed by constructing an augmented state consisting of system states and sensor faults. A novel
H
− /
L
∞ unknown input observer (UIO) is proposed in the finite frequency domain using the generalized Kalman–Yakubovich–Popov lemma to render the residual sensitive to actuator faults and robust against disturbances, which can achieve a better fault diagnosis result compared with existing observers in the full frequency domain. In the fault diagnosis scheme, a time‐varying threshold is presented based on the L∞ performance index to detect actuator faults, and the sensor fault estimation is simultaneously achieved through an asymptotical estimation of the augmented state. Design conditions for the proposed
H
− /
L
∞ UIO are derived and converted into linear matrix inequalities to solve its design matrices together. The effectiveness of the developed
H
− /
L
∞ UIO and the fault diagnosis scheme are validated via simulations of an example of a single‐link flexible joint robot.