具有衰减测量和概率通讯延迟的随机非线性系统的事件触发分布式一致滤波  

Event-Based Distributed Consensus Filtering for Stochastic Nonlinear Systems with Fading Measurements and Probabilistic Communication Delays

作　　者：于强 胡军 陈东彦[1,3] YU Qiang;HU Jun;CHEN Dongyan(Department of Mathematics,Harbin University of Science and Technology,Harbin 150080;School of Automation,Harbin University of Science and Technology,Harbin 150080;Heilongjiang Provincial Key Laboratory of Optimization Control and Intelligent Analysis for Complea Systems,Harbin University of Science and Technology,Harbin 150080)

机构地区：[1]哈尔滨理工大学数学系,哈尔滨150080 [2]哈尔滨理工大学自动化学院,哈尔滨150080 [3]黑龙江省复杂系统优化控制与智能分析重点实验室,哈尔滨150080

出　　处：《系统科学与数学》2022年第9期2279-2296,共18页Journal of Systems Science and Mathematical Sciences

基　　金：国家自然科学基金(12171124,12071102);黑龙江省教育科学规划重点课题(GJB1422069)资助课题。

摘　　要：文章针对传感器网络下一类具有衰减测量和概率通讯延迟的随机时变非线性系统,研究该类系统的分布式一致滤波算法设计问题.首先,考虑到传感器网络的拓扑结构,状态估计信息可在相关节点间进行交换.注意到相邻节点间的信息传输会产生概率通讯延迟现象,提出事件触发传输方案旨在减少交换传感器节点之间误差较大的估计信息.每个传感器节点使用相邻节点发送的信息来修正自身的状态,从而构造相应的分布式一致滤波器.通过最小化滤波误差协方差上界的迹来求出滤波器增益以及一致增益.此外,对滤波误差协方差上界矩阵的单调性进行理论分析,并通过数值仿真验证所提出的滤波算法的可行性和有效性.In this paper,a distributed consensus filter design problem is studied for a class of stochastic nonlinear systems with fading measurement and probabilistic communication delays in sensor networks.Considering the topology of the network,the state estimation information is packaged and exchanged among various nodes.It is noted that probabilistic communication delay occurs in the information transmission between adjacent nodes,and an event-triggered transmission scheme is proposed to reduce the estimation information with large error between the sensor nodes.Each sensor node uses the information sent by its neighbors to correct its own state,and the corresponding distributed consensus filter is constructed accordingly.The filter gain and consensus gain are obtained by minimizing the trace on the upper bound of the filter error covariance.In addition,the monotonicity of the upper bound matrix of the filtering error covariance is analyzed theoretically,and the feasibility and effectiveness of the proposed filtering algorithm are verified by numerical simulation.

关 键 词：传感器网络 时变随机系统 衰减测量 事件触发通信协议 概率通讯延迟 卡尔曼一致滤波 

分 类 号：TP212.9[自动化与计算机技术—检测技术与自动化装置] TN929.5[自动化与计算机技术—控制科学与工程]