False Data Injection Attack Against the Optimal Energy Management of Active Distribution Networks: A Bilevel Programming Model

Abstract

With the wide integration of advanced measurement equipment and communication networks into active distribution networks (ADNs), the ADN cyber-physical system (CPS) is facing the risk of false data injection attacks (FDIAs). This paper proposes a bilevel attack model to obtain optimal false data, which maximizes power losses, photovoltaic (PV) curtailment, and load shedding while ensuring the concealment of the attack. Different from the FDIA for the transmission networks, the FDIA targeting ADNs not only injects false data into load demand but also further injects false data into PV output, achieving an attack through the cooperation of the two types of false data injection. Moreover, differing from a single-level attack model, the proposed bilevel model takes the optimal energy management model executed by the ADN system operators as the lower-level model, using a master-slave game to obtain the optimal false data that considers operator responses, making it more in line with practical engineering. Using the Karush-Kuhn-Tucker (KKT) conditions, the bilevel model is transformed into a single-level model for ease of solution. Simulation results based on the IEEE 33-bus test case confirm the effectiveness of the proposed model and reveal the risk of the ADN under FDIA.