|
Deep learning-based approach for detecting false data injection attacks in smart grids
|
Alireza Soltani   , Hamid Reza Baghaee * , Mahmoud-Reza Haghifam |
| Tarbiat Modares University , hrbaghaee@modares.ac.ir |
|
|
Abstract: (19 Views) |
| With the expansion of smart distribution networks and the increase in cyber threats, securing power systems has become a critical necessity. False Data Injection (FDI) attacks are among the most destructive threats, jeopardizing system stability by bypassing traditional detection mechanisms. This research presents a detection model based on a Graph Autoencoder that proactively identifies these attacks in the IEEE 33-bus distribution network. The proposed model is designed within a deep learning architecture and, by utilizing the capabilities of Graph Neural Networks, learns the normal operational patterns of the network and identifies attacks through reconstruction error calculation. The results indicate that the presented model is capable of effectively detecting anomalies with over 98% accuracy and leads to an improvement in the performance of the protection system. |
|
| Keywords: Anomaly Detection, Cyber Attack, Deep Learning, Graph Autoencoder, Graph Neural Networks Smart Grid |
|
|
Full-Text [PDF 927 kb]
(15 Downloads)
|
Type of Study: Research |
Subject:
Smart Grids
Received: 2025/12/3 | Accepted: 2026/01/26 | Published: 2025/05/31
|
|
|
|
|
|
|
| Send email to the article author |
|
|
|
| Add your comments about this article |
|
|
|
