Detection of False Data Injection Attacks in Smart Microgrid using Deep Learning to Improve Classification Algorithms

Detecting false data injection (FDI) attacks poses a significant challenge in smart microgrids. Recently, classification algorithms have been employed to address this issue. However, the effectiveness of these methods is hindered by the similarity between normal states and attack states. Consequently, the classification algorithm's accuracy is low, leading to undetected attacks. In the past few years, deep learning has emerged as a powerful approach for modeling complex nonlinear relationships and enhancing accuracy in various domains. In this study, we leverage a deep neural network (DNN) to improve the accuracy of classification algorithms in FDI attack detection. To accomplish this, we first employ the DNN to assign a value to each sample in the dataset, which consists of N features. This value is then added as an additional feature, resulting in a dataset with 1+N features. Subsequently, the augmented dataset is utilized for training and testing the desired classification algorithm instead of the original dataset. Experimental results demonstrate that augmenting the dataset with the feature obtained from the DNN significantly enhances the accuracy of classification algorithms in detecting FDI attacks.