Data-Based Security Control for Partially Unknown Networked Linear Systems: An Event-Triggered Strategy

Abstract

This work studies the data-based security control problem of networked control systems (NCSs) under aperiodic denial-of-service (DoS) attacks, in which system parameters are partially unknown; meanwhile, the DoS attack is subject to only attack duration constraints. In order to reduce unnecessary data transmission, an event-triggered mechanism (ETM) is introduced in the S/C channel. Moreover, by means of partially known parameters and precollected state/input measurements, a new data representation of the system model is constructed as an extension of de Persis and Tesi’s work with completely unknown parameters. Then, by solving linear matrix inequalities (LMIs) related to only partially known parameters and measurement data, an event-triggered sliding mode controller is designed, under which the exponential stability and reachability of the closed-loop system are analyzed by constructing an attack parameter-dependent time-varying Lyapunov function. It is shown that despite the partial availability of system parameters, the proposed parameter- and data-based hybrid control method can effectively mitigate the impact of attacks. Finally, the effectiveness and superiority of the proposed strategy are verified by simulation results.