基于系统级综合的分布式鲁棒最优滤波算法OA

This paper proposes a distributed robust optimal filtering algorithm for linear time-invariant systems subject to both white Gaussian noises and bounded-power disturbances.This method achieves the optimal estimation under bounded-power disturbances and Gaussian white noise on condition of the observability of the system.Under a sys-tem-level synthesis framework,the estimated performance is described based on the system response of error dy-namics,and then the filter gain at any single node is obtained.Then local estimation results from all nodes are dy-namically averaged among neighbor nodes for sufficient iterations to achieve estimation performance close to that of a centralized estimator.The stability of the filter is analyzed to ensure the rationality of the design and finally a nu-merical example is given to prove the effectiveness of the proposed algorithm.