Jiang Chenhui, Tang Ze, Park Ju H, Feng Jianwen
IEEE Trans Neural Netw Learn Syst. 2024 Feb;35(2):1821-1832. doi: 10.1109/TNNLS.2022.3185586. Epub 2024 Feb 5.
In this article, the quasi-synchronization for a kind of coupled neural networks with time-varying delays is investigated via a novel event-triggered impulsive control approach. In view of the randomly occurring uncertainties (ROUs) in the communication channels, the global quasi-synchronization for the coupled neural networks within a given error bound is considered instead of discussing the complete synchronization. A kind of distributed event-triggered impulsive controllers is presented with considering the Bernoulli stochastic variables based on ROUs, which works at each event-triggered impulsive instant. According to the matrix measure method and the Lyapunov stability theorem, several sufficient conditions for the realization of the quasi-synchronization are successfully derived. Combining with the mathematical methodology with the formula of variation of parameters and the comparison principle for the impulsive systems with time-varying delays, the convergence rate and the synchronization error bound are precisely estimated. Meanwhile, the Zeno behaviors could be eliminated in the coupled neural network with the proposed event-triggered function. Finally, a numerical example is presented to prove the results of theoretical analysis.
本文通过一种新颖的事件触发脉冲控制方法，研究了一类具有时变延迟的耦合神经网络的准同步问题。鉴于通信通道中随机出现的不确定性（ROUs），考虑的是在给定误差范围内耦合神经网络的全局准同步，而非讨论完全同步。提出了一种基于ROUs并考虑伯努利随机变量的分布式事件触发脉冲控制器，其在每个事件触发脉冲时刻起作用。根据矩阵测度方法和李雅普诺夫稳定性定理，成功推导了实现准同步的几个充分条件。结合变分参数公式的数学方法和时变延迟脉冲系统的比较原理，精确估计了收敛速度和同步误差界。同时，利用所提出的事件触发函数可消除耦合神经网络中的芝诺行为。最后，给出一个数值例子以验证理论分析结果。
