This paper investigates a novel strategy which can address the fault-tolerant control (FTC) problem for nonlinear strict-feedback systems containing actuator saturation, unknown external disturbances, and faults related to actuators and components. In such method, the unknown dynamics including faults and disturbances are approximated by resorting to Neural-Networks (NNs) technique. Meanwhile, a back-stepping technique is employed to build a fault-tolerant controller. It should be stressed that the main advantage of this strategy is that the NN weights are updated online based on gradient descent (GD) algorithm by minimizing the cost function with respect to NNs approximation error rather than regarding weights as adaptive parameters, which are designed according to Lyapunov theory. In addition, the convergence proof of NN weights and the stability proof of the proposed FTC method are given. Finally, simulation is performed to demonstrate the effectiveness of the proposed strategy in dealing with unknown external disturbances, actuator saturation and the faults related to the components and actuators, simultaneously.

本文研究了一种新策略，该策略可以解决非线性严格反馈系统的容错控制 (FTC) 问题，该系统包含执行器饱和、未知外部干扰以及与执行器和组件相关的故障。在这种方法中，包括故障和扰动在内的未知动态是通过采用神经网络（NN）技术来逼近的。同时，采用反步技术构建容错控制器。需要强调的是，该策略的主要优点是基于梯度下降 (GD) 算法，通过最小化与 NN 近似误差相关的成本函数，而不是将权重视为自适应参数，从而在线更新 NN 权重，这是根据设计的李雅普诺夫理论。此外，给出了NN权重的收敛证明和所提出的FTC方法的稳定性证明。最后，进行仿真以证明所提出的策略在同时处理未知外部干扰、执行器饱和以及与组件和执行器相关的故障方面的有效性。