Attack Detection and Secure State Estimation for Discrete-Time Nonlinear Systems

This paper studies the issues of attack detection and secure state estimation for nonlinear cyber-physical systems (CPSs) with unknown-but-bounded (UBB) noises. First, a nonlinear estimation approach based on first-order Taylor expansion with Lagrangian remainder is employed, and a zonotopic observer is proposed for state estimation. Then, a novel attack detection mechanism is presented to improve detection sensitivity. It consists of two steps: 1) judging the intersection between the predicted state set and the strip, and 2) checking whether the measurement state set is empty set. In addition, a secure state estimation method is proposed to protect the estimated state against cyber attacks. Finally, an unmanned vehicle system is leveraged to demonstrate the validity of the main results.