Entity Information Recognition Method for Power Grid Dispatch Fault Handling Plan Based on DK-UIE Framework

To address the structural parsing challenge of multi-type entities and complex hierarchical relationships in power grid fault handling plan, this paper proposes an intelligent parsing method for power grid dispatch fault handling plan based on knowledge graph technology. First, we establish a knowledge unit tagging system for fault handling plan, which generates standardized definitions for entity types and attribute features. Subsequently, we propose a Domain Knowledge-enhanced Universal Information Extraction (DK-UIE) framework to train the mapping relationships between fault handling plan entities and their labels in high-dimensional space. Following the logical association rules of dispatch fault handling plan, we construct a structured knowledge graph for power grid dispatch fault handling management. Furthermore, an application scheme for entity information recognition in power grid dispatch fault handling plan is developed based on the knowledge graph. Validation using data from a dispatch control center demonstrates that the proposed knowledge extraction model achieves average precision, recall, and F1-scores of 97.50%, 96.62%, and 97.05% respectively, showing superior recognition accuracy compared with existing methods. The constructed knowledge graph effectively supports risk identification in power grid dispatch fault handling.