Knowledge Graph-Based Approach for Main Transformer Defect Grade Analysis

The effectivemaintenance of power grid equipment is critical for ensuring the safe and stable operation of the power grid. In recent years, knowledge graphs have emerged as a powerful tool for representing complex relationships and knowledge in a structured and accessible format. In this paper, we proposed a knowledge graph-based approach for analyzing and diagnosing defects in power grid transformers. We first designed an ontology for defect data in the field of main transformers in power grids. The ontology included equipment information, defect descriptions, and industry-standard classification criteria. We then performed named entity recognition(NER) on textual data in the field of main transformers using the Bert-Bilstm-CRF [1-3] model to extract entities. The extracted entity information was represented using the ontology, and the ontology was embedded into a knowledge graph using models such as TransE [4]. We conducted knowledge graph completion experiments to achieve diagnosis and analysis of the defect level. Our experimental results demonstrated that this method efficiently and automatically constructs a knowledge graph of main transformers in power grids. The well-designed ontology and effective knowledge graph completion experiments also support the analysis of defect levels in main transformers in power grids. Additionally, this method can promote the understanding and management of complex systems in the field of power grid equipment.