Adaptive estimator-based exponential fault-tolerant tracking control for a class of uncertain MIMO nonlinear systems with simultaneous actuator/sensor faults

In this article, we propose an adaptive estimator-based exponential fault tolerant control (FTC) for a class of Multiple-Input Multiple-Output (MIMO) nonlinear systems under model uncertainties, external disturbances, along with both multiplicative and additive time-varying actuator/sensor faults. To tackle the inherent “explosion of terms” issue in the standard backstepping method, we employ the command filtered backstepping (CFB) framework, introducing a novel compensating system to alleviate the effect of filtering errors and improve the convergence of tracking errors. The adaptive estimator-based estimation laws are meticulously designed separately from the tracking system, integrating both proportional and integral prediction errors derived from estimator outputs along with estimation errors of faulty terms. Instead of directly employing high learning gains to achieve fast estimation—a practice that may induce undesired high-frequency oscillations, especially in transient process—we incorporate time-varying learning gains with appropriate convergence rates that may be initialized with low values and gradually increased to high bounded gains. Furthermore, by introducing modification terms with appropriate time-varying gains, we demonstrate, based on Lyapunov theory, that the resulting system is globally exponentially stable. Two application examples are considered in simulation to illustrate the effectiveness and benefits of the presented FTC approach. •The adaptive estimator-based FTC system is built under both actuator/sensor faults.•The proposed method introduces a novel compensating system into the CFB framework.•The on-line estimation schemes are constructed independently of the tracking system.•The proposed adaptive laws incorporate time-varying modification and learning gains.•The closed-loop system is rigorously proven to be globally exponentially stable.