Secure H∞ Control Against Time-Delay Attacks in Cyber-Physical Systems

Dan Jin,Mingnan Hu,Bo Chen,Li Yu
DOI: https://doi.org/10.1080/23307706.2021.2011792
2021-01-01
Journal of Control and Decision
Abstract:This paper is concerned with the secure H-infinity control problem for cyber-physical systems (CPSs) under time-delay attacks, which cause the systems' information to be received with delays. Through transforming the H-infinity control problem to a two-player zero-sum game problem, the H-infinity controller can be obtained by solving the saddle point for the game. Due to the unknown system parameters, a model-free reinforcement learning (RL) method is applied, and the measurement feedback controller is designed based on the historical inputs and the tracking errors. To obviate the requirement of the initial admissible policy, the value iteration (VI) of the adaptive dynamic program (ADP) algorithm is proposed to solve the approximate saddle point. Moreover, the convergence of the proposed algorithm is proved. Finally, a networked circuit system is employed to show the effectiveness of the proposed methods.
What problem does this paper attempt to address?