With the increasing digitalization and intelligence of smart grids, their reliance on communication and computing infrastructures has made them vulnerable to cyber threats, particularly false data injection attacks (FDIAs), which can severely compromise state estimation accuracy and threaten grid stability. Existing detection methods often face limitations in localization accuracy and recovery performance under varying attack intensities. To overcome these challenges, we propose a hybrid framework named ECN-Swin Transformer (EST), which combines an edge convolutional network (ECN) for capturing the grid’s topological features with a Swin Transformer for extracting both local and global temporal characteristics, thereby improving the detection and localization of FDIA-targeted nodes. Furthermore, for the recovery of compromised state variables, a SwinUNet model—an encoder-decoder architecture based on the Swin Transformer with U-Net-style skip connections—is developed to enable high-fidelity reconstruction under physical constraints. Extensive simulations on IEEE 14-, 39-, 118-, and 300-bus systems with varying FDIA intensities verify that the proposed EST+SwinUNet framework achieves superior localization accuracy, enhanced recovery quality, and better real-time adaptability compared to conventional methods, offering a robust and scalable solution for cyber-resilient smart grids.

In this paper, we consider decomposing an image into its cartoon and texture components. Traditional methods, which mainly rely on the gradient amplitude of images to distinguish between these components, often show limitations in decomposing small-scale, high-contrast texture patterns and large-scale, low-contrast structural components. Specifically, these methods tend to decompose the former to the cartoon image and the latter to the texture image, neglecting the scale features inherent in both components. To overcome these challenges, we introduce a new variational model which incorporates an$L_{0}$-based total variation norm for the cartoon component and an$L_{2}$norm for the scale space representation of the texture component. We show that the texture component has a small$L_{2}$norm in the scale space representation. We apply a quadratic penalty function to handle the non-separable$L_{0}$norm minimization problem. Numerical experiments are given to illustrate the efficiency and effectiveness of our approach.

The smart grid consists of two parts, one is the physical power grid, the other is the information network. In order to study the cascading failure, the vulnerability analysis of the smart grid is done under a kind of community attack style in this paper. Two types of information networks are considered, i.e. topology consistency and scale-free cyber networks, respectively. The concept of control center is presented and the controllable power nodes and observable power lines are defined. Minimum load reduction model(MLRM) is given and described as a linear programming problem. A index is introduced to assess the vulnerability. New England 39 nodes system is applied to simulate the cascading failure process to demonstrate the effectiveness of the proposed MLRM where community the attack methods include attack the power lines among and in power communities.