联邦滤波算法与Gauss-Markov估计的统一性分析  

Uniformity of Federated Filter Algorithm and Gauss-Markov Estimation

作　　者：王勇军[1] 徐景硕[1] 李林[1] 马伟丽 

机构地区：[1]海军航空工程学院青岛校区 [2]解放军91286部队

出　　处：《火力与指挥控制》2013年第12期48-51,共4页Fire Control & Command Control

基　　金：国防科技重点实验室基金资助项目(20120224006)

摘　　要：从矩阵论和数理统计理论的角度,阐述了联邦滤波与Gauss-Markov估计的内在联系,论证了联邦全局最优滤波与Gauss-Markov估计的统一性。推导了基于Gauss-Markov估计的联邦滤波信息融合算法,并与最小二乘估计的信息融合算法以及集中卡尔曼滤波算法在舰载组合导航系统中应用进行了对比。仿真结果表明,基于Gauss-Markov估计的联邦滤波信息融合算法与集中式卡尔曼滤波的精度相当,两者的估计精度均高于最小二乘估计,前者具有全局最优性。From the perspective of matrix and mathematical statistics theory, internal connection of federated filter and Markov estimation is formulated. The uniformity of federated global optimal filtering and Gauss-Markov estimation is proved. Meanwhile,the fusion algorithm of federated filtering based on Gauss-Markov estimation and the fusion algorithm based on least squares estimation are taken a deduction and applied to the shipborne integrated navigation system. Simulation results show that filtering accuracy of fusion algorithm of federated filtering based on Gauss-Markov estimation is equal to those of centralized kalman filter. Their estimation accuracy are both higher than those of least squares estimation. The former possesses global optimality.

关 键 词：联邦滤波 Gauss—Markov估计 最优估计 融合算法 

分 类 号：U666.1[交通运输工程—船舶及航道工程]