Fault-tolerant tracker for a class of unknown interconnected nonlinear systems with input constraint

This paper presents the decentralized fault-tolerant tracker based on the model predictive control (MPC) for a class of unknown interconnected large-scale sampled-data nonlinear systems. Due to the computational requirements of MPC and the system information is unknown, the observer/Kalman filter identification (OKID) method is utilized to determine decentralized appropriate (low-) order discrete-time linear models. Then, to overcome the effect of modeling error on the identified linear model of each subsystem, the improved observers with the high-gain property based on the digital redesign approach will be presented. Once fault is detected in each decentralized controller, one of the backup control configurations in each decentralized subsystem is switched to using the soft switching approach. Thus, the decentralized fault-tolerant control with the closed-loop decoupling property can be achieved through the above approach with high-gain property decentralized observer/tracker.