A new fixed-time stabilization approach for neural networks with time-varying delays

Abstract

In this article, we investigate the problem of fixed-time stabilization (FXTSB) of delayed neural networks (DNNs). Firstly, some new general conditions on the control law are established to guarantee the FXTSB of DNNs. Secondly, specific linear matrix inequalities FXTSB conditions are obtained by constructing different kinds of controller which include a delay-dependent and free ones. Furthermore, the FXTSB of DNNs with unbounded activation functions is investigated and the restriction of differentiability of the time-varying delay is removed. Finally, three numerical examples accompanied by graphical illustrations are given to illuminate our theoretical results and based on chaotic synchronization, our approach has been successfully applied to secure communication which can be realized with a time delay.