This article is concerned with the event-triggered finite-time $H_{\infty }$ estimator design for a class of discrete-time switched neural networks (SNNs) with mixed time delays and packet dropouts. To further reduce the data transmission, both the measured information of system outputs and switching signal of the SNNs are only allowed to be accessible for the constructed estimator at the certain triggering time instants. Under this consideration, the simultaneous presence of the switching and triggering actions also leads to the asynchronism between the indices of the SNNs and the designed estimator. Unlike the existing event-triggered strategies for the general switched linear systems, the proposed event-triggered mechanism not only allows the occurrence of multiple switches in one triggering interval but also removes the minimum dwell-time constraint on the switched signal. In light of the piecewise Lyapunov–Krasovskii functional theory, sufficient conditions are developed for the estimation error system to be stochastically finite-time bounded with a finite-time specified $H_{\infty }$ performance. Finally, the effectiveness and applicability of the theoretical results are verified by a switched Hopfield neural network.

中文翻译：

---

具有事件触发策略的切换离散时间延迟神经网络的有限时间 H∞ 估计器设计

本文关注事件触发的有限时间 $H_{\infty }$具有混合时间延迟和数据包丢失的一类离散时间交换神经网络 (SNN) 的估计器设计。为了进一步减少数据传输，系统输出的测量信息和 SNN 的切换信号都只允许在某些触发时刻被构造的估计器访问。在这种考虑下，切换和触发动作的同时存在也导致 SNN 的索引与设计的估计器之间的不同步。与现有的用于一般切换线性系统的事件触发策略不同，所提出的事件触发机制不仅允许在一个触发间隔内发生多个切换，而且消除了对切换信号的最小驻留时间约束。 $H_{\infty }$表现。最后通过切换Hopfield神经网络验证了理论结果的有效性和适用性。