IEEE Trans Cybern. 2019 May;49(5):1669-1679. doi: 10.1109/TCYB.2018.2809576. Epub 2018 Mar 6.
The adaptive neural network tracking control problem is investigated for a class of interconnected switched systems. The considered systems are with unmodeled dynamics, some of which do not satisfy the input-to-state stable (ISS) condition. By utilizing the neural network to approximate the composite unknown nonlinear functions, the corresponding decentralized tracking controller is designed for each subsystem with the help of dynamic surface control method. Some subsystems are stable with the designed controller, while other subsystems may not be stable because of non-ISS unmodeled dynamics, but they have some special properties with the designed controller. Then, a novel switching signal scheme is established such that the interconnected switched system is stable in the sense of semi-global boundedness, and the tracking errors can converge to predefined residual sets with prescribed performance index. Moreover, the switching scheme allows the number of switches to grow faster than traditional average dwell time method. Finally, a numerical example is provided to demonstrate the effectiveness of the presented results.
研究了一类互联切换系统的自适应神经网络跟踪控制问题。所考虑的系统具有未建模动态，其中一些不满足输入到状态稳定（ISS）条件。通过利用神经网络来逼近组合未知非线性函数，借助动态面控制方法为每个子系统设计了相应的分散跟踪控制器。一些子系统在设计的控制器下是稳定的，而其他子系统由于非 ISS 未建模动态可能不稳定，但它们在设计的控制器下具有一些特殊的性质。然后，建立了一种新的切换信号方案，使得互联切换系统在半全局有界意义下是稳定的，并且跟踪误差可以收敛到具有规定性能指标的预定残差集。此外，该切换方案允许开关的数量比传统的平均驻留时间方法更快地增长。最后，通过一个数值例子验证了所提出的结果的有效性。
