Evaluating Deep Learning for Detecting Data Integrity Attacks in Energy Smart Grids

The quality of energy services tightly conditions the transition to Industry 4.0 and smart cities. Modern smart grid systems are changing the game's rules by opening the door for green energy integration and enhancing the grid's connectivity. However, the connectivity implemented in smart grid systems imposes many data security vulnerabilities, including data integrity attacks. This paper explores the ability of various intelligent learning models to detect data integrity attacks in smart grid systems. The models analyzed include Deep Q-Network (DQN), Convolutional Neural Network (CNN), and Long Short-Term Memory (LSTM). The dataset is balanced using the SMOTE model to improve the model's performance. The CNN model shows promising results in detecting integrity attacks in smart grid systems. Moreover, the DQN model shows a limited ability to classify the attacks. The study presents recommendations and directions to enhance data integrity attack detection in smart grid systems.