A Novel False Data Injection Attack Detection Method for Digital Substations

The digital transformation of control and communication equipment has significantly enhanced the efficiency and performance of modern power systems, but has also increased their vulnerability to cyber-attacks, particularly false data injection attacks (FDIA) that aim to manipulate system measurements and disrupt operational decision-making. To address this challenge, this study proposes a novel detection method based on machine learning techniques that predicts the dynamic characteristics of the system state variations to identify FDIA attacks effectively. The approach involves (i) Training a machine learning model on historical state data to predict current system state variations, (ii) Deriving probability distributions from both historical and current system state variations (iii) Comparing discrepancies using the Hellinger Distance (HD) to identify the FDIA attack. The proposed approach is validated through hardware-in-the-loop testing on a typical 400 kV UK substation. The method demonstrated high accuracy, offering an effective solution for strengthening the power system's security against FDIA attacks.