Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
4948234 | Neurocomputing | 2017 | 11 Pages |
Abstract
This paper investigates the distributed Hâ-consensus filtering problem for a class of discrete time-varying systems with random parameters and event-triggering protocols. An event-triggering protocol for each node is employed to reduce the burden of the network communication. A novel matrix named by information matrix is proposed to describe the complicated correlations among the elements of random matrix. By virtue of the presented information matrix, a weighted covariance matrix can be easily obtained to analyze the system with random parameters. With the aid of the newly constructed dissipation matrix and vector supplied rate functions, a set of local coupled conditions for each node is obtained such that the stochastic vector dissipativity-like over the finite-horizon of the filtering error dynamics can be guaranteed. As well, these sufficient conditions together could effectively solve the distributed Hâ-consensus filtering problem. Notably, the designed filtering algorithm can be implemented on each node to obtain the desirable distributed filter gains. Finally, the effectiveness and applicability of the proposed algorithm is illustrated by a numerically simulative example.
Related Topics
Physical Sciences and Engineering
Computer Science
Artificial Intelligence
Authors
Fei Han, Yan Song, Sunjie Zhang, Wangyan Li,