Fault-Tolerant Controller Design for a Class of Nonlinear MIMO Discrete-Time Systems via Online Reinforcement Learning Algorithm
针对一类多输入多输出非线性离散时间系统，提出一种基于强化学习的容错控制方法，利用神经网络在线调整权重，在发生渐近或突变故障时最小化长期性能指标，减少浪费和能耗。
This paper concentrates on the reinforcement learning (RL)-based fault-tolerant control (FTC) problem for a class of multiple-input-multiple-output (MIMO) nonlinear discrete-time systems. Both incipient faults and abrupt faults are taken into account. Based on the approximation ability of neural networks (NNs), an RL algorithm is incorporated into the FTC strategy, in which an action network is developed to generate the optimal control signal and a critic network is used to approximate the novel cost function, respectively. Compared with the existing results, a novel fault tolerant controller is proposed based on an RL method to reduce a long-term performance index after a fault occurs. The meaning of minimizing the performance index after a fault occurs in an MIMO system is that waste will be decreased and energy will be saved. Note that the weights of NNs are adjusted online rather than offline. Then, it is proven that the adaptive parameters, tracking errors, and optimal control signals are uniformly bounded even in the presence of the unknown fault dynamics. Finally, a numerical simulation is provided to show the effectiveness of the proposed FTC approach.