Deep Learning-Based Fault Knowledge Graph Construction for Power Communication Networks
Gao Dequan, Zhu Pengyu, Sheng Wang, Ziyan Zhao · 2024
Power communication network is a crucial infrastructure in the model power system, and its maintenance capability are crucial to ensuring the stable operation of power grid business. As an organized semantic knowledge base, the knowledge graph effectively organizes power communication network fault documentation and expert experience to enhance intelligent maintenance. This paper outlines a top-down approach to systematically construct a fault knowledge graph in the domain of power communication networks. The approach utilizes a seven-step method to establish a domain ontology model and integrates deep learning algorithms, including pre-trained language models, bidirectional long short time memory networks, convolutional neural networks and attention mechanisms. These algorithms process unstructured text to extract key entities and relationships. The effectiveness of the approach is verified through experiments using a product device document as a test case. Extracted knowledge is then visualized and stored using Neo4j database. Finally, this paper proposes a knowledge service model centered on fault knowledge graph and explores its application in fault diagnosis.