A memory behavior related hybrid event-triggered mechanism for an improved robust control on neural networks

Yang Liu,Zhenzhen Zhang,Hao Chen,Shouming Zhong
DOI: https://doi.org/10.1016/j.matcom.2023.04.023
IF: 3.601
2023-04-24
Mathematics and Computers in Simulation
Abstract:This paper addresses an H∞ control approach on neural networks with hybrid-triggered mechanism (HTM) under deception attacks. With the aim of mitigating the burden of the transmission network, an HTM is introduced to handle unforeseen non-ideal environment influence, which is characterized by Bernoulli distribution. The weight combination coefficients εj related to historical information are conducted to develop an improved HTM. By taking into account network-induced delay, and the randomly happened deception attacks in transmission network, a Lyapunov-Krasovskii functional (LKF) is constructed. Using linear matrix inequality (LMI), sufficient conditions are formed to render the system asymptotically stable and the H∞ hybrid-triggered controller is designed. Finally, Simulation examples are executed to validate the feasibility of the developed method.
mathematics, applied,computer science, software engineering, interdisciplinary applications
What problem does this paper attempt to address?