Yunfei Zhang, Tiantian Wu, Chen Lu, Xingzhi Xu, Weiwen Xu
Anti-misoperation logic, substation, entity alignment algorithm,power grid dataset, multimodal information
Aiming at the challenges of multi-source heterogeneous data alignment difficulty and high real-time requirements in substation anti-misoperation logic verification, an entity alignment method integrating multimodal and structural information is proposed. This method supports dynamic verification of anti-misoperation logic and real-time blocking decisions through precise cross- system entity matching. During the entity matching process, a cross-modal attention mechanism is employed to fuse equipment text, attributes, and topological features, generating robust joint embeddings. Simultaneously, the fused Gromov–Wasserstein optimisation objective is incorporated to achieve joint optimisation of semantic alignment and topological consistency. Furthermore, a block-based acceleration strategy is proposed to reduce the computational complexity of neighbourhood structures. Experiments on the power system dataset PowerGrid demonstrate that the proposed method achieves Hits@1, Hits@10, MRR, and F1-Score values of 98.6%, 97.8%, 94.6%, and 95.2%, respectively, representing average improvements of 9.1%, 9.5%, 11.5%, and 12.5% compared to the existing classical method FGWEA. The experimental results validate the effectiveness and real-time performance of the proposed method in substation anti-misoperation logic verification, offering a novel approach for multi-source heterogeneous data alignment and contributing to the enhancement of the intelligence level and reliability of anti-misoperation systems. ∗ Wuxi Power Supply Branch of State Grid Jiangsu Electric Power Co., Ltd., Wuxi 214000, China; e- mail: [email protected]; [email protected]; [email protected]; [email protected] ∗∗ College of Computer Science and Software Engineer- ing, Hohai University, Nanjing 211100, China; e-mail: [email protected] Corresponding author: Yunfei Zhang Recommended by Simon X. Yang
Important Links:
Go Back
