Predefined-time adaptive fuzzy control for nonlinear cyber-physical systems under sensor and actuator attacks

Abstract

This paper introduces an adaptive fuzzy control framework with predefined-time convergence for nonlinear cyber-physical systems (CPSs) in strict-feedback form, subject to unmeasured states and cyber-attacks. Specifically, the system is exposed to false data injection (FDI) attacks on the sensor side, as well as both multiplicative and additive disturbances on the actuator side. These sensor attacks involve unknown, time-varying interference in measurement data, while actuator attacks cause both faults and deliberate disruptions. To handle unknown system nonlinearities and counteract the effects of these cyber threats, fuzzy logic systems are employed as universal approximators. Additionally, a fuzzy state observer is designed to estimate the system’s unobservable states and the unknown sensor-side attack signals. An observer-based control scheme is then developed to ensure predefined-time convergence. To overcome singularity problems and reduce computational complexity, a predefined-time filter and a dynamic surface control (DSC) technique are integrated into the design. The resulting control method guarantees that all closed-loop signals are semi-globally practically predefined-time stable (SGPPTS), and that the system output tracks the desired trajectory within a predefined-time despite the presence of FDI attacks. The effectiveness of the proposed strategy is demonstrated through two simulation examples.