Data-driven event-triggered control for switched systems based on neural network disturbance compensation

Abstract

This paper investigates data-driven event-triggered tracking control for switched systems with dual unreliable communication links. Consider modeling constraints with unknown dynamics, a data-driven model-free adaptive control (MFAC) method which relies on system input and output data is adopted. To reduce system resource consumption caused by continuous calculation of control execution, an event-triggering scheme is applied. A radial basis function (RBF) neural network is constructed to approximate the unknown disturbance. Accordingly, the disturbance estimation and triggered system data are respectively transmitted through dual network channels, which are vulnerable to malicious denial-of-service (DoS) attacks. Then, a data dependent anti-attack mechanism and an event-triggered MFAC with disturbance compensation are proposed, respectively. Moreover, in view of the stability analysis of the MFAC switched systems, by introducing Lyapunov functional and average dwell time technique, a set of novel sufficient conditions is derived for ensuring the boundedness of tracking error. Finally, simulation examples verify the effectiveness of the proposed method.