FDIA Attack Detection Technique for Smart Grids Based on Graph Reconstruction and Spatio-Temporal Joint Modeling

With the widespread application of smart grids, the false data injection attack (FDIA) has become a major threat to power grid security. Traditional detection methods often have difficulty in effectively identifying such attacks, especially in complex environments. To address this problem, this paper proposes a spatial-temporal joint FDIA detection framework named HSGT-Net, which integrates the Hodge diffusion, Sparge sparsity mechanism, and time series modeling. This framework unifies the branch power and bus power data through a graph reconstruction mechanism, and combines Hodge multi-order diffusion with spatial feature modeling of Sparge-iTransformer, so it can effectively capture the local and global dependencies of the power grid. A GRU module is formulated to model time series features, with improved perception ability of dynamic attack features. Experimental results show that the HSGT-Net performs better than traditional machine learning methods (such as SVM and KNN) and deep learning models (such as MLP, CNN, and GNN+LSTM) on both IEEE- 30 and IEEE-57 standard test systems. Especially in large-scale power system scenarios, it can still maintain high accuracy and robustness. In addition, ablation experiments further verify the effectiveness of each module and prove the key role of joint modeling on spatial and temporal features in FDIA detection. In a word, the HSGT-Net demonstrates strong adaptability and computational efficiency in FDIA attack detection and has good engineering application prospects.