This article is concerned with the challenge of guaranteeing output constraints for fault-tolerant control (FTC) of a class of unknown multi-input single-output (MISO) nonlinear systems in the presence of actuator faults. Most industrial systems are equipped with redundant actuators and a fault detection-isolation mechanism for accommodating unexpected actuator faults. To simplify the system design and reduce the risk of false alarm or missed detection brought by the detection unit, a learning-based switching function scheme is proposed to automatically activate different sets of actuators in a rotational manner without human intervention. By this means, no explicit fault detection mechanism is needed. An additional step has been made to guarantee that the system output remains in user-defined time-varying asymmetric output constraints all the time during the occurrence of failures by utilizing error transformation techniques. The stability of the transformed system can equivalently deliver the result that the original system output stays in the required bounds. Hence, system crash or further catastrophic outcomes can be avoided. A neural network is integrated to embody the adaptive FTC design for dealing with unknown system dynamics. The dynamic surface control (DSC) technique is also invoked to decrease complexity. Furthermore, the stability analysis is carried out by the standard Lyapunov approach to guarantee that all the signals of the closed-loop system are semiglobally uniformly ultimately bounded. Finally, the simulation results are provided to verify the effectiveness of the proposed scheme.

中文翻译：

---

通过基于学习的切换方案对不确定非线性系统的性能保证容错控制

本文关注在存在执行器故障的情况下保证一类未知多输入单输出 (MISO) 非线性系统的容错控制 (FTC) 的输出约束的挑战。大多数工业系统都配备了冗余执行器和故障检测隔离机制，以适应意外的执行器故障。为简化系统设计，降低检测单元带来的误报或漏检风险，提出了一种基于学习的切换功能方案，无需人工干预，自动旋转激活不同组执行器。通过这种方式，不需要明确的故障检测机制。通过利用误差转换技术，还采取了额外的步骤，以确保在发生故障期间系统输出始终保持在用户定义的时变非对称输出约束中。转换后系统的稳定性可以等效地提供原始系统输出保持在所需范围内的结果。因此，可以避免系统崩溃或进一步的灾难性后果。集成了神经网络以体现用于处理未知系统动力学的自适应 FTC 设计。还调用动态表面控制 (DSC) 技术来降低复杂性。此外，稳定性分析是通过标准的李雅普诺夫方法进行的，以保证闭环系统的所有信号都是半全局一致最终有界的。最后，