IEEE Trans Cybern. 2019 Dec;49(12):4296-4307. doi: 10.1109/TCYB.2018.2862828. Epub 2018 Sep 10.
This paper addresses the distributed adaptive event-triggered H filtering problem for a class of sector-bounded nonlinear system over a filtering network with time-varying and switching topology. Both topology switching and adaptive event-triggered mechanisms (AETMs) between filters are simultaneously considered in the filtering network design. The communication topology evolves over time, which is assumed to be subject to a nonhomogeneous Markov chain. In consideration of the limited network bandwidth, AETMs have been used in the information transmission from the sensor to the filter as well as the information exchange among filters. The proposed AETM is characterized by introducing the dynamic threshold parameter, which provides benefits in data scheduling. Moreover, the gain of the correction term in the adaptive rule varies directly with the estimation error and inversely with the transmission error. The switching filtering network is modeled by a Markov jump nonlinear system. The stochastic Markov stability theory and linear matrix inequality techniques are exploited to establish the existence of the filtering network and further derive the filter parameters. A co-design algorithm for determining H filters and the event parameters is developed. Finally, some simulation results on a continuous stirred tank reactor and a numerical example are presented to show the applicability of the obtained results.
本文针对一类具有时变和切换拓扑的滤波网络中的扇形界非线性系统，研究了分布式自适应事件触发 H 滤波问题。在滤波网络设计中，同时考虑了滤波器之间的拓扑切换和自适应事件触发机制（AETMs）。通信拓扑随时间演变，假设它服从非齐次马尔可夫链。考虑到网络带宽有限，AETMs 已被用于从传感器到滤波器的信息传输以及滤波器之间的信息交换。所提出的 AETM 的特点是引入了动态阈值参数，这在数据调度方面具有优势。此外，自适应规则中的修正项增益与估计误差成正比，与传输误差成反比。切换滤波网络被建模为一个马尔可夫跳跃非线性系统。利用随机 Markov 稳定性理论和线性矩阵不等式技术，建立了滤波网络的存在性，并进一步推导出滤波器参数。开发了一种用于确定 H 滤波器和事件参数的协同设计算法。最后，通过连续搅拌釜式反应器和一个数值实例进行了一些仿真结果，以展示所得到的结果的适用性。
