带时间相关乘性噪声多传感器系统的分布式融合估计  

Distributed Fusion Estimation for Multi-sensor Systems With Time-correlated Multiplicative Noises

作　　者：马静[1] 杨晓梅 孙书利[2] MA Jing;YANG Xiao-Mei;SUN Shu-Li(School of Mathematics Science,Heilongjiang University,Harbin 150080;School of Electronic Engineering,Heilongjiang University,Harbin 150080)

机构地区：[1]黑龙江大学数学科学学院,哈尔滨150080 [2]黑龙江大学电子工程学院,哈尔滨150080

出　　处：《自动化学报》2023年第8期1745-1757,共13页Acta Automatica Sinica

基　　金：国家自然科学基金(61573132,61873087);黑龙江省信息融合估计与检测重点实验室资助。

摘　　要：研究带时间相关乘性噪声多传感器系统的分布式融合估计问题,其中时间相关的乘性噪声满足一阶高斯-马尔科夫过程.通过引入虚拟状态和虚拟过程噪声,构建了虚拟状态的递推方程.首先,基于新息分析方法,分别对系统状态和虚拟状态设计局部一步预报器.然后,基于一步预报器设计状态的局部线性滤波器、多步预报器和平滑器.推导了任意两个局部状态估计误差之间的互协方差矩阵.接着,基于线性最小方差意义下的矩阵加权、对角矩阵加权和标量加权融合算法,给出相应的分布式融合状态估值器.最后,分析算法的稳定性.仿真研究验证了该算法的有效性.This paper is concerned with the distributed fusion estimation problem for multi-sensor systems with time-correlated multiplicative noises,where the time-correlated multiplicative noises satisfy first-order Gaussian-Markov processes.By introducing the friction states and friction process noises,the recursive equation of the friction state is established.Local one-step predictors for system state and friction states are designed based on an innovation analysis approach.Then,the local linear filter,multi-step predictor and smoother for the state are designed based on one-step predictor.The cross-covariance matrices between any two local state estimation errors are derived.Further,the corresponding distributed fusion state estimators are given based on the fusion algorithms weighted by matrices,diagonal matrices and scalars in the linear minimum variance sense.At last,the stability of the proposed algorithms is analyzed.The simulation research verifies the effectiveness of the proposed algorithms.

关 键 词：时间相关乘性噪声 多传感器系统 分布式融合估值器 互协方差矩阵 虚拟状态 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置] O212.1[自动化与计算机技术—控制科学与工程]