Robust secure synchronization control for multiple neural networks: An event‐triggered adaptive integral sliding mode method

Abstract

AbstractThis article investigates the robust secure synchronization control of multiple neural networks (MNNs) systems with external random disturbances in the presence of actuator and sensor attacks. First, the types of attacks are analyzed in detail. Actuator attacks refers to the damage of control input signals transmitted through the network layer, and sensor attacks refers to the damage of communication channels between neural networks (NNs). Then, for the attacked NNs system, by introducing a novel neighborhood synchronization error, two novel integral sliding mode (ISM) functions with the event‐triggered mechanisms are designed. Based on this, a distributed event‐triggered adaptive ISM secure synchronization control scheme is proposed. It is worth noting that the control signal and adaptive parameters in the proposed scheme are only updated at the triggering time instants, which further saves the energy and resources consumed by the system. Next, the sufficient conditions for secure synchronization of MNNs system are obtained by constructing Lyapunov function and a positive lower bound of the internal execution time is ensured. Finally, a numerical example is given to demonstrate the effectiveness of the proposed secure synchronization control strategy.