Synchronization and robust synchronization of fractional-order coupled neural networks (FCNNs) are considered in this paper. Different with the most published works on synchronization based on a special solution of an isolate node of the networks, we remove this restriction and introduce a more widely accepted definition of synchronization. Meanwhile, because of parametric uncertainties of network models, robust synchronization for FCNNs is investigated. In addition, by utilizing pinning control strategies, several sufficient conditions are derived to make sure that the considered networks can realize pinning synchronization and robust pinning synchronization. Finally, the correctness of the obtained results is substantiated by two given numerical examples.
A multi‐weighted coupled neural networks (MWCNNs) model with event‐triggered communication is studied here. On the one hand, the passivity of the presented network model is studied by utilising Lyapunov stability theory and some inequality techniques, and a synchronisation criterion based on the obtained output‐strict passivity condition of MWCNNs with event‐triggered communication is derived. On the other hand, some robust passivity and robust synchronisation criteria based on output‐strict passivity of the proposed network with uncertain parameters are presented. At last, two numerical examples are provided to testify the effectiveness of the output‐strict passivity and robust synchronisation results.
3 February 2019 in International Journal of Control
In this paper, we firstly investigate synchronization and passivity in finite time of coupled memristive neural networks (CMNNs) with and without coupling delays. The key contributions of our work lie on two aspects. On the one hand, by designing suitable controllers, several finite-time synchronization criteria for the CMNNs with and without coupling delays are established. On the other hand, some sufficient conditions for ensuring finite-time passivity of these considered networks are presented. In addition, two numerical examples are provided to guarantee the availability for the acquired results.