Li Yuan-Xin, Yang Guang-Hong
IEEE Trans Neural Netw Learn Syst. 2018 Sep;29(9):4359-4369. doi: 10.1109/TNNLS.2017.2765683. Epub 2017 Nov 9.
This paper is concerned with the simultaneous design of a neural network (NN)-based adaptive control law and an event-triggering condition for a class of strict feedback nonlinear discrete-time systems. The stability and tracking performance of the closed-loop network control system under the event-triggering strategy is formally proven based on the Lyapunov theory in a hybrid framework. The proposed Lyapunov formulation yields an event-triggered algorithm to update the control input and NN weights based on conditions involving the closed-loop state. Different from the existing traditional NN control schemes where the feedback signals are transmitted and executed periodically, the feedback signals are transmitted and executed only when the event-trigger error exceeds the specified threshold, which can largely reduce the communication load. The effectiveness of the approach is evaluated through a simulation example.
本文关注一类严格反馈非线性离散时间系统的基于神经网络（NN）的自适应控制律和事件触发条件的同时设计。基于混合框架下的李雅普诺夫理论，正式证明了事件触发策略下闭环网络控制系统的稳定性和跟踪性能。所提出的李雅普诺夫公式产生了一种事件触发算法，用于根据涉及闭环状态的条件更新控制输入和神经网络权重。与现有传统神经网络控制方案不同，在传统方案中反馈信号是周期性传输和执行的，而这里反馈信号仅在事件触发误差超过指定阈值时才进行传输和执行，这可以大大降低通信负载。通过一个仿真例子评估了该方法的有效性。
