Learning Tracking Control for Cyber–Physical Systems

This article investigates the problem of optimal tracking control for cyber-physical systems (CPSs) when cyber realm is attacked by Denial-of-Service (DoS) attacks which can prevent signal transmitting to actuator. Attention focused on how design scheme without using system dynamics and analyze impact DoS performance. First, a Riccati equation augmented system, including model reference derived under framework dynamic programming. The existence uniqueness its solution are proved. Second, successful attack probability performance analyzed. A critical value given, beyond cannot converge. controller be designed. Third, reinforcement learning introduced schemes, in not necessary known. Finally, both dc motor an F16 aircraft used evaluate proposed schemes this article.