Detection of False Data Injection Attack in Smart Grid Based on Extended Kalman and Smooth Variable Structure Filter

Abstract

False Data Injection Attack (FDIA) is a prevalent cyber threat in modern smart grids, capable of altering measurement data and circumventing Bad Data Detection (BDD) mechanisms, which results in inaccurate state estimations. This paper proposes a combined filtering algorithm, the extended Kalman filter and smooth variable structure filter (EK-SVSF), to address state estimation issues under FDIA and to overcome the limitations of traditional bad data detection (BDD) algorithms in detecting sophisticated FDIA. The EKF introduces a smooth variable structure to force the estimates to stay within the region of the true state trajectory by switching the gains, keeping the state estimates close to the true values. This approach enhances the robustness and stability of the estimation, effectively addressing the accuracy challenges of EKF based estimation under FDIA. Second, a state estimation-based FDIA detection method is proposed by leveraging the different characteristics presented by static state estimation Weighted Least Squares (WLS) and dynamic state estimation EK-SVSF under attack conditions. The Euclidean distance between the dynamic state estimates from EK-SVSF and the static estimates from WLS is calculated and compared to a predefined detection threshold to identify FDIA effectively. Finally, simulations conducted on IEEE 14-bus and IEEE 30-bus systems demonstrate that the proposed method effectively detects FDIA.

Keywords

• Power system state estimation
• false data injection attacks
• EK-SVSF
• detection
• smart grid