Neural-Network-Based Adaptive Control of Time-Delayed Non-Linear Cyber-Physical Systems With Power Uncertainty Against Deception Attacks

Abstract

At this job, the adaptive control problem is investigated for a class of non-linear cyber-physical systems (CPSs), where the CPSs considered are not only subject to deception attacks and time delay, but also contain uncertain input powers. The deception attacks result in the actual values of the system state being unavailable and control gains being unknown. On the basis of the theory of Lyapunov stability, a new adaptive neural-networks-based control scheme is designed to guarantee the stability of the closed-loop system and mitigate the impact of deception attacks. Compared with the existing works in literature, (1) the input powers of the CPSs considered in this article are unknown and new controllers are constructed based on the neural network approximation technique; (2) the influence of unknown time delay is eliminated by using a novel Lyapunov–Krasovskii function. Furthermore, in order to address unknown gains caused by deception attacks, the Nussbaum gain technique is firstly extended to the CPSs with power uncertainties. Finally, the simulation results confirm the effectiveness of the control strategy presented in this work.