Research on innovative methods for converter station operation and maintenance based on knowledge graphs

Efficient handling of various faults and issues during the operation of converter station equipment is essential for enhancing maintenance effectiveness. An innovative operation and maintenance (O&M) method based on knowledge graph technology is proposed. Relevant converter station equipment data is first collected, followed by data cleaning, denoising, text clause splitting, and manual annotation. Two models are then constructed: BERT-BiLSTM-CNN-MHSA-CRF for named entity recognition and BERT-BiLSTM-CNN-Attention for relationship extraction. Both models yield favorable results. The resulting knowledge graph is stored in Neo4j and visualized on web pages using JavaScript, facilitating fault queries and knowledge updates.