Location of False Data Injection Attacks in Power System

Recent research indicates that a new type of network attack called false data injection (FDI) attack can bypass the bad data detection mechanism in power system state estimation, which is posing a serious threat to the security and reliable operation of the smart grid. To better protecting the smart grid system, a supervised FDI attacks location method is proposed based on statistical features and tree boosting (XGBoost) technique. Threshold-based algorithm and SHapley Additive exPlanations (SHAP) value based algorithm are presented for single and multiple bus node attacks respectively. The proposed method is tested on the IEEE 14-bus system using the real-time load data from Open Energy Information (OpenEI). Simulation results demonstrate that our proposed method can recognize the manipulated bus node at a high detection and location rate.