Finite-Time Fault Detection Under Deception Attacks: A Hybrid Dynamic Variables-Dependent Event-Based Scheme

This article deals with the dynamic event-based finite-time fault detection (FD) issue for singularly perturbed systems under deception attacks. A hybrid dynamic variables-dependent event-triggered mechanism (ETM) is proposed, which contains both additive and multiplicative internal dynamic variables (IDVs). Such a new dynamic ETM (DETM) is employed to decide whether the measurement data packet should be released or not at each time instant. A Bernoulli random variable is utilized to describe the deception attack that occurs during the measurement data packet transmission. Attention is to design an FD filter (FDF) which guarantees that the resultant error dynamics of filtering on FD is stochastically $H_{\infty }$ finite-time bounded. Based on a new Lyapunov function containing two IDVs, sufficient conditions are derived to ensure the existence of the FDF whose parameters are designed by resorting to the feasible solutions of several matrix inequalities. Two examples are presented to demonstrate the effectiveness of the DETM-based FDF design method. It is revealed that the devised DETM has superiority over some existing ETMs in saving network resources while not degrading the FD filtering performance.