Open AccessCluster and local mode‐dependent H ∞ filtering for distributed Markovian jump systems in lossy multi‐sensor networks
Author(s) -
Guan Yanpeng,
Ge Xiaohua,
Jiang Xiefu
Publication year - 2017
Publication title -
iet signal processing
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.384
H-Index - 42
ISSN - 1751-9683
DOI - 10.1049/iet-spr.2016.0374
Subject(s) - network topology , filter (signal processing) , computer science , markov process , lossy compression , mode (computer interface) , jump , stability (learning theory) , control theory (sociology) , cluster (spacecraft) , topology (electrical circuits) , mathematics , artificial intelligence , statistics , physics , quantum mechanics , operating system , control (management) , combinatorics , machine learning , computer vision , programming language
This study addresses an H ∞ filtering problem for distributed Markovian jump systems (DMJSs) in multi‐sensor networks with inaccessible global jumping modes, random data losses, random sensing topologies and incomplete mode transition rates (TRs). First, locally overlapped clusters are introduced to reformulate the DMJS to account for the scenario that the global modes of the DMJS are not accessible for filter design. Second, a new cluster and local mode‐dependent H ∞ filtering framework is presented to incorporate the simultaneous presence of Markovian data losses and sensing topologies as well as incomplete mode TRs. The proposed H ∞ filters depend on only available information of cluster modes and local modes within this cluster, and thus eliminating the requirement of complete global modes. Third, criteria for designing desired filters are derived to preserve the stochastic stability of the resulting filtering error system under a prescribed H ∞ performance index. The proposed results are shown to be more general by covering some existing results as special cases. Finally, an F404 aircraft engine system is employed to demonstrate the effectiveness of the proposed filter design approach.