A two-stage recovery strategy against false data injection attacks in smart grids

Abstract

False Data Injection Attacks (FDIAs) exploit vulnerabilities in the bad data identification mechanisms of power systems by tampering with measurement data. These attacks compromise state estimation, posing a critical threat to the safety and stability of power system operations. To enhance system resilience and enable rapid recovery after an attack, this paper proposes a novel two-stage recovery strategy specifically designed to address FDIAs. In the first stage, a detection method based on Normalized State Variable Residuals (NSVR) is introduced. By leveraging the distinct convergence behaviors of Weighted Least Squares (WLS) and the Extended Kalman Filter (EKF) under FDIAs, NSVR serves as an effective metric for detecting compromised measurements. The second stage employs a multi-step prediction compensation (MSPC) approach for attack correction. After identifying the attacked buses in the first stage, the prediction step of the EKF is utilized to compute the predicted values of the measurements. These predicted values are then selectively used to replace tampered measurements, reconstructing a set of corrected measurements for subsequent state estimation. Comprehensive simulation experiments were conducted on IEEE 14-bus and IEEE 30-bus test systems. The results confirm the efficacy of the proposed algorithm in accurately detecting FDIAs and achieving recovery of the power system.