Approximation-based Nussbaum-gain adaptive neural fault-tolerant prescribed-performance control for saturated MIMO nonlinear systems: a non-backstepping approach

In this paper, an active fault-tolerant (AFT) adaptive tracking control problem, with guaranteed a priori prescribed performance, is developed for a class of multi-input multi-output (MIMO) pure-feedback nonlinear systems subject simultaneously to input saturation, output constraints, and unknown actuator faults. In the suggested fault-tolerant adaptive control method, neural networks (NNs) are used to understand unknown nonlinear behaviours, reduce the impact of input saturation, and manage three additive and one multiplicative actuator faults without requiring a separate fault unit. The proposed AFT controller design avoids the difficulty caused by the recursive design procedure in controlling cascading MIMO nonlinear systems by reformulating the original constrained MIMO control problem into an unconstrained MIMO feedback control problem in controllable form, with a dynamic control input, to avoid the algebraic loop problem usually encountered in AFT control. To estimate the unknown states of the newly derived canonical system, an adaptive NN observer is adopted, and the knowledge of control directions is circumvented by using Nussbaum-type functions. Furthermore, the proposed AFT controller is designed to guarantee a priori prescribed performance with only a few adaptive parameters by using the common Lyapunov stability method. Simulations on a model and a quadrotor UAV validate the proposed control scheme.