DOI number:10.1109/TSMC.2018.2791575
Journal:IEEE Transactions on Systems, Man, and Cybernetics: Systems
Key Words:Coupled systems, finite-horizon H∞ estimation, Markov jump systems, neural networks
Abstract:This paper investigates an H ∞ estimator design for time-varying coupled neural networks (NNs) over a finite-horizon. In order to reduce the information exchanged among the NNs, a periodic inner-coupling strategy is proposed. In addition, a Markov driven transmission scheme is introduced to overcome the communication capacity constraint between the NNs and the estimators, where an inner-coupling-dependent Markov chain is used to improve the efficiency of the communication channel. Subsequently, the time-varying Markov estimators are designed to enhance the performance of the estimators. A recursive matrix inequality (RMI)-based sufficient condition is established to ensure that the time-varying estimation error system meets the finite-horizon H ∞ performance. Afterward, the estimator gains are designed by transforming the RMIs into linear RMIs. Finally, a numeral example is used to illustrate the developed results.
First Author:Yong Xu
Indexed by:Journal paper
Discipline:Engineering
Document Type:J
Volume:50
Issue:1
Page Number:211-219
ISSN No.:2168-2216
Translation or Not:no
Date of Publication:2018-01-31
Links to published journals:https://ieeexplore.ieee.org/document/8276597