Design of Switching-Type Zonotopic Unknown Input Observers for Attack Reconstruction and Detection in Uncertain Cyber–Physical Systems

The security problem plays a critical role in managing the operation of cyber–physical systems (CPSs). This article investigates state estimation, attack reconstruction and attack detection for uncertain CPSs with unknown but bounded (UBB) noise. First, uncertain CPS dynamics are characterized using Takagi–Sugeno fuzzy logic. To address critical modeling challenges arising from complex non-Gaussian stochastic processes, the UBB processes are systematically analyzed. Within this framework, false data injection (FDI) attacks on actuators or sensors are regarded as unknown inputs that can compromise the integrity and authenticity of the system information. Second, a new switching-type zonotopic unknown input observer (SZUIO) is proposed to enhance estimation accuracy while mitigating attack impacts. This observer handles system uncertainties and partial unknown inputs through a switching mechanism that activates multiple operational modes, thereby relaxing state estimation constraints and reducing conservatism. Moreover, the optimal SZUIO for each time step is selected from a set of candidate observers, based on the dynamic information of MF. Third, to enhance proactive defense against attacks, an integrated framework is proposed for state estimation, attack reconstruction and detection in uncertain CPSs. Finally, the feasibility of the proposed method is demonstrated through examples.