Visualization Maintenance Method of Relay Protection Systems Based on Knowledge Graphs

Aiming at addressing issues such as high module coupling, difficulties in fault tracing, and low maintenance efficiency in relay protection systems, this paper proposes a visual operation and maintenance method based on a multi-task learning model and knowledge graph. Firstly, a multi-task learning model for joint entity-relationship extraction and relationship fragment identification is constructed, employing a relationship fragment expansion strategy to resolve multi-layer logical nesting issues in operation logs. Subsequently, a knowledge graph for the relay protection system is developed to visualize module coupling relationships, enabling technicians to directly assess system risks and health status. Experimental results demonstrate that the proposed method achieves F1 scores of 89.1% and 87% for entity and relationship recognition, respectively, providing an effective visual decision-support tool for operation and maintenance management for relay protection systems.