Observer‐Based Sensor Attack Diagnosis for Cyber‐Physical Systems via Zonotope Theory

Abstract This paper investigates the attack detection, isolation and reconstruction problems for cyber‐physical systems (CPSs) under sensor attacks. We deal with all the issues in view of an overall system instead of every subsystem. To begin with, a Luenberger‐like observer is designed and a state interval estimation method is proposed based on the zonotope theory. And then, a sensor attack detection scheme is developed by using the interval estimation of the output error as the dynamic residual. Besides, the sensor detection method is modified to form a sensor attack isolation method to locate the exact subsystems suffering from sensor attacks. Moreover, by treating sensor attacks as part of the states, an augmented descriptor system is constructed and the state and sensor attack asymptotic estimation problems are discussed in detail. Finally, a simulation example is given to verify the effectiveness of the proposed methods.