Detection and Localization of False Data Injection Attacks Using Matrix Decomposition Methods
Hongji Wu, Hang Zhang, Wenshao Bu · 2025
As a core component of contemporary Energy Management Systems, state estimation is inherently vulnerable to the detrimental effects of False Data Injection Attacks (FDIA), posing significant challenges to system reliability. Traditional algorithms utilized for the identification of false data are insufficient in their capacity to detect such attacks, let alone ascertain the origin of the attack and the subsequent recovery of the measurements data. The present paper sets out a precise methodology for the identification of FDIA and recovery measurements; the FDIA detection problem is considered to be a matrix decomposition problem. The determination of this is contingent upon two factors: first, the low rank of the measured data; and secondly, the sparsity of the false data injection attack in a continuous time period. Applying Moving Target Defense (MTD) to the attack identification and mitigation method can contribute to a larger equivalent deviation of the tampered measurements from the attack-free measurements since the attacks are constructed based on the pre-MTD line parameters. Therefore, MTD has the potential to improve the attack identification and mitigation performance under this challenging circumstance. Through extensive experimentation on the IEEE 30-bus test systems, the validity of the proposed method is substantiated.