A Method for Constructing the Knowledge Graph of Digital Twin Power Facilities Based on Deep Learning

We propose a method for constructing the knowledge graph of digital twin power facilities based on deep learning to solve the problem which is difficult to effectively manage and apply complex and diverse information of power facilities. This method systematically integrates structured data and unstructured data such as facility attributes, functions, locations, operating status, and detection status, generates digital twin models of power facilities. It constructs knowledge graph through entity recognition and fills in the lacks in the knowledge graph. The experiments show that the accuracy of entity recognition by this method reaches 90.89%, and the accuracy of knowledge graph completion is 83.8%. Finally, we constructed the knowledge graph of digital twin power facilities. This knowledge graph contains 29564 entities and 104615 relationships, with entity recognition accuracy of 81.68% and relationship recognition accuracy of 79.35%. This knowledge graph can serve as the core of data and model management for our power facility digital twin platform for model management and facility retrieval.