Predefined-time synchronization of competitive neural networks.

Abstract

In this paper, the predefined-time synchronization of competitive neural networks (CNNs) is researched based on two different predefined-time stability theorems. In view of the bilayer structure of CNNs, we design two bilayer predefined-time controllers. The first controller utilizes sign function, while the second controller utilizes exponential function and Lyapunov function. In these two controllers, the predefined time is set as a controller parameter, and it can be an arbitrary positive constant. Under these two controllers, the considered CNNs can achieve synchronization within the predefined time regardless of the initial values. A specific example is presented to validate the theoretical results.