Composite adaptive exponential fault-tolerant control for large-scale nonlinear systems with input saturation and actuator/sensor faults

This paper presents a composite adaptive decentralised fault-tolerant control (FTC) strategy for large-scale nonlinear systems subject to external disturbances, system uncertainties, time-varying sensor and actuator faults, as well as input saturation. The filtered backstepping technique is employed to mitigate the ‘explosion of terms’ issue inherent in conventional backstepping methods. To counteract filtering errors, a new dedicated compensation mechanism is introduced, which further improves tracking error convergence. Novel composite adaptive laws are proposed by incorporating compensated tracking errors, prediction errors and estimation errors, wherein learning adaptation terms are integrated with appropriately designed time-varying gains. The proposed framework explicitly accounts for sensor fault propagation through subsystem interconnections and adaptively handles actuator saturation without requiring auxiliary systems. By invoking Lyapunov stability theory, the controlled system is rigorously proven to be globally exponentially stable. The validity and effectiveness of the proposed FTC method are demonstrated through two illustrative application examples.