Attack Tolerant Fault Diagnosis Based on Unknown Input Interval Observer

For cyber-physical systems that are adversely af-fected by actuator attacks, faults, and disturbances, this paper proposes a secure and robust fault estimator designed for real-time detection and estimation of potential fault signals, laying the groundwork for subsequent secure fault tolerance strategies. Specifically, the designed fault estimator integrates the features of unknown input observers and interval observers, enabling it to decouple disturbance signals and generate upper and lower bounds for the fault signal even in the presence of actuator attacks. Through mathematical derivation, linear solvable conditions have been provided that ensure the stability of the nominal error system, non-negativity of the error system matrix, and robustness of the system under attack. Finally, a set of simulation experiments demonstrates the effectiveness of this fault diagnosis method.