Abstract This article put forward an adaptive neural fault-tolerant control strategy for a class of switched nonlinear systems subject to actuator fault by means of the command filter approach. By using neural networks, the unknown nonlinear functions of the system under consideration are approximated, while its unmeasurable states are estimated by establishing a switched observer. Furthermore, the “explosion of complexity” issue, which arises from the derivatives of virtual controllers, is addressed with the command filter method. In order to reduce filter errors and overcome the drawbacks of the most traditional approaches, such as the ones based on dynamic surface control techniques, an error compensation mechanism was developed. In summary, by taking advantage of the command filter approach, backstepping algorithm, and average dwell time method, an adaptive NNs fault-tolerant controller is established for the system under consideration. Finally, the designed controller has the ability to make the reference signal which can be tracked by the output of the system as close as possible and the boundness of all signals within the closed-loop system can be guaranteed. The usefulness of the designed controller is illustrated by the simulation example.

中文翻译：

---

基于指令滤波技术的非线性切换系统自适应容错控制

摘要 本文针对一类受执行器故障影响的非线性切换系统，通过指令滤波方法提出了一种自适应神经容错控制策略。通过使用神经网络，所考虑系统的未知非线性函数被逼近，而其不可测量状态则通过建立一个切换观察器来估计。此外，通过命令过滤方法解决了由虚拟控制器派生而来的“复杂性爆炸”问题。为了减少滤波器误差并克服最传统方法的缺点，例如基于动态表面控制技术的方法，开发了一种误差补偿机制。总之，通过利用命令过滤器方法，反步算法，和平均停留时间方法，一个自适应的神经网络容错控制器被考虑的系统被建立。最后，所设计的控制器具有使系统输出可跟踪的参考信号尽可能接近的能力，保证闭环系统内所有信号的有界性。仿真示例说明了所设计控制器的有用性。