Attention-Enhanced Bidirectional LSTM for Accurate False Data Injection Attack Detection in Smart Grid

Abstract

Due to an increase in the integration of renewable energy sources in the smart grid, the importance of the smart grid is increasing day by day. Moreover, the decentralization of the smart grids makes them prone to cyberattacks such as Distributed denial of service (DDOS) attacks, False data injection attacks (FDIA), and so forth. These attacks raise serious concerns about the integrity and stability of a smart grid. Therefore, the detection of these attacks has a prominent impact on the stability of a smart grid. For this purpose, the presented work proposes a robust detection system that leverages the fusion of Bidirectional Long Short-Term Memory (Bi-LSTM) networks with an Attention Mechanism. The proposed architecture of Bi-LSTM captures both forward and backward temporal dependencies in sensor data, enhancing the model's ability to detect anomalies that cause FDIA in time-series data. So, it amplifies the efficiency of the proposed model by bringing about stress on the most emphatic time steps while enhancing interpretability and accuracy classification (classification accuracy). The proposed model is evaluated on a time-series smart grid data set, and the experimental results have been compared with the other state-of-the-art techniques, such as LSTM-CNN and LSTM-Autoencoder. The results clearly demonstrate that the proposed model is capable enough to identifying the anomaly with an accuracy rate of about 92.32%.