A Secure Hybrid Dynamic-State Estimation Approach for Power Systems Under False Data Injection Attacks

摘要

Dynamic-state estimation plays a critical role in achieving real-time wide-area monitoring of power systems. On the other hand, false data injection (FDI) attacks are substantial threats, which can undesirably ruin the state estimation results. To tackle this problem, an effective secure hybrid dynamic-state estimation approach that involves a dynamic model of the attack vector is proposed in this article. In the proposed method, an initial estimation of the system states is first obtained using a designed unknown input observer (UIO). Subsequently, based on the system, UIO models, and the initial estimations of the states, a dynamic model for the attack vector is extracted. Ultimately, the attack model is augmented with the main system model for coestimation of the attack and the system states using a Kalman filter. The onset of the FDI attack is rapidly detected by the accurate estimation of the attack vector. The effectiveness of the proposed approach is demonstrated under different FDI attack scenarios by a thorough theoretical analysis as well as simulations on IEEE 14-bus and 57-bus test systems. In order to show that the proposed method can keep up with typical scan rates of commercial phasor measurement units, a series of software-in-the-loop experiments are also conducted and the real-time feasibility of the proposed approach is guaranteed.