Master-slave synchronization of heterogeneous dimensional delayed neural networks

In this paper, the problem on synchronization is studied for a class of master-slave delayed neural networks (DNNs) with heterogeneous dimensions. Through designing a reduced-order observer and choosing an augmented Lyapunov-Krasovskii functional, a delay-dependent stability criterion on the error system is presented and the synchronization one is formulated in terms of linear matrix inequalities (LMIs) via parameters' transformation. Especially; since some novel inequality techniques are used, those previously ignored information can be reconsidered during the discussion and the conservatism can be effectively reduced. Furthermore, the proposed condition can be conveniently presented and the controller gain can be checked by solving the proposed LMIs. Finally, two numerical examples are presented to illustrate the presented results. (C) 2016 Elsevier B.V. All rights reserved.