Event-Triggered Impulsive Synchronization for Neural Networks Subject to Deception Attacks: Saturation Control and Optimization

• Published in: IEEE transactions on automation science and engineering Vol. 22; pp. 12071 - 12081
• Main Authors: Sun, Wenjing, Tang, Ze, Feng, Jianwen, Park, Ju H.
• Format: Journal Article
• Language: English
• Published: IEEE 2025
• Subjects: actuator saturation; Actuators; Control systems; Couplings; deception attacks; domain of attraction; Electronic mail; Event detection; hybrid event-triggered mechanism; impulsive effects; Mean-square synchronization; Optimization; Sun; Symmetric matrices; Synchronization; Vectors
• ISSN: 1545-5955; 1558-3783
• DOI: 10.1109/TASE.2025.3540772

This paper is concerned with the mean-square synchronization issue of the coupled neural networks (CNNs) subject to deception attacks, where the saturation constraint on impulsive control signal is adequately taken into account. To compensate the effect of deception attacks, a hybrid event-triggered mechanism is developed and a minimum inter-event interval is introduced simultaneously to avoid the Zeno phenomena. Under deception attacks, sufficient conditions to guarantee the achievement of the mean-square exponentially synchronization for the CNNs are presented with a novel method combined with mathematical induction, polyhedral representation of saturation nonlinearity, proof by contradiction, which address the actuator saturation in discrete-time control signals effectively. Meanwhile, in consideration of different impulsive effects, three optimization problems are constructed for the sake of acquiring the maximum estimation of the domain of attraction, the feasible maximal impulsive interval and the admissible feasible minimum impulsive interval, respectively. Finally, two numerical simulations are carried out to illustrate the validity of the proposed theoretical analysis. Note to Practitioners-In the engineering and industry, the synchronization control of CNNs exists in many different fields, including image processing, fluid dynamics, and secure communication. The control signals and plant information are transmitted over communication network which is vulnerable to deception attacks. To save communication resources and compensate the effect of deception attacks, a hybrid event-triggered scheme is proposed. Simultaneously, the synthesis of impulse effects and actuator saturation are considered to make results more practical. In addition, three optimization problems are proposed to derive the maximum estimation of the domain of attraction, some related parameters selection of event-triggered scheme.