As the informatization of power systems advances, the secure operation of power systems faces various potential network attacks and threats. The false data injection attack (FDIA) is a common attack mode that can lead to abnormal system operations and serious economic losses by injecting abnormal data into terminal links or devices. The current research on FDIA primarily focuses on detecting its existence, but there is relatively little research on the localization of the attacks. To address this challenge, this study proposes a novel FDIA localization method (GA-CNN-LSTM) that combines convolutional neural networks (CNNs), long short-term memory (LSTM), and a genetic algorithm (GA) and can accurately locate the attacked bus or line. This method utilizes a CNN to extract local features and combines LSTM with time series information to extract global features. It integrates a CNN and LSTM to deeply explore complex patterns and dynamic changes in the data, effectively extract FDIA features in the data, and optimize the hyperparameters of the neural network using the GA to ensure an optimal performance of the model. Simulation experiments were conducted on the IEEE 14-bus and 118-bus test systems. The results indicate that the GA-CNN-LSTM method achieved F1 scores for location identification of 99.71% and 99.10%, respectively, demonstrating superior localization performance compared to other methods.

The rapid development of large language models has significantly advanced research on parameter-efficient fine-tuning and preference alignment for domain-specific applications. In cybersecurity, existing studies have primarily focused on general-purpose security assistance and performance improvement for specific tasks. However, most approaches still jointly model heterogeneous knowledge types, organizational structures, and expression paradigms, causing shared knowledge and structure-specific knowledge to become entangled and resulting in parameter interference, structural confusion, and unstable category-specific generation. This paper proposes SPOR-LoRA, a low-rank adaptation method that decouples knowledge content from knowledge organization and expression patterns. Unlike conventional LoRA, which maps all samples into a single shared low-rank subspace, SPOR-LoRA constructs a shared–private low-rank adaptation architecture for multiple knowledge domains, enabling the separate modeling of cross-domain common knowledge and heterogeneous structural patterns. Differential privacy is further incorporated to protect sensitive training data. Experimental results demonstrate that the proposed method achieves a better balance among cybersecurity knowledge preservation, generation accuracy, and privacy robustness.