Event-based passification of delayed memristive neural networks

As well known, neural network is a mathematical model obtained by simulating biological synapse for information processing [1,2]. Memristor was first predicted by Chua in [3]. Since the successful establishment of the physical model of memristor in 2008 [4], scientists have been enthusiastic about the dynamics of memristive neural networks [5]. As a kind of nonlinear resistor with memory function, the non-linear nature of memristor can generate chaotic circuits with rich applications in secure communications [6]. In the circuit implementation of neural networks, memristors can also be used to replace resistors to describe the connection weights as state-dependent functions in the mathematical model, then we can get the memristive neural networks [7]. In addition, the dynamic behaviors of memristive neural networks has also attracted widespread attention [8-10]. In the practical application of neural networks, due to the switching speed of amplifiers, the transmission speed of elec This paper focuses on the passification issue of delayed memristive neural networks via the event-based control. First, by designing an appropriate controller based on a static event trigger scheme, the passification conditions are deduced for delayed memristive neural networks. Then, under the same controller, the passivity is discussed for the delayed memristive neural network system with a more economical and realistic dynamic event trigger rule. Meanwhile, in order to ensure these two event trigger control schemes are Zeno free, the existence of positive lower bounds are approved for the inter event time. Finally, illustrative examples are elaborated to support the theoretical results. (c) 2021 Elsevier Inc. All rights reserved.