Robust Finite-Time Synchronization of Recurrent Neural Networks via Saturated Control

This brief studies the finite-time synchronization of a class of recurrent neural networks, where the dual effects of the saturation structure and uncertain factors on the system are fully considered. By designing different classes of saturated controllers, some sufficient conditions guaranteeing robust finite-time synchronization are proposed, where the estimations of the settling-time and the domain of attraction are obtained. In particular, in the framework of saturation structure, the designed controller can be improved a simpler form to further acquire globally robust finite-time synchronization. Finally, a numerical example is demonstrated to verify the presented results.

Automated summary

This study considers the effects of saturation structure and uncertain factors on a class of recurrent neural networks. By designing various saturated controllers, sufficient conditions for robust finite-time synchronization are proposed, along with the estimations of settling-time and domain of attraction. Notably, the designed controller can be simplified for globally robust finite-time synchronization within the saturation structure framework. A numerical example is provided to validate the presented results.