This study has conducted research on entity extraction and relationship extraction for intelligent highways using an enhanced BERT model along with the BGRU model and has integrated knowledge representation methods to fundamentally complete the construction of a knowledge graph in the intelligent highway domain. Initially, by developing the schema layer of the intelligent highway knowledge graph, entities were classified into demand entities, service entities, functional entities, technological entities, and facility entities, and the primary relationships among them were established, including driving, undertaking, providing, and other relationships. The enhanced BERT model was used to extract entities from intelligent highway-related articles, achieving an F1 score of 86.8%, while the relationship extraction model reached an F1 score of 88.3%, both showing a certain degree of improvement over traditional models for entity and relationship extraction. Building on this foundation, through a process of filtering and refining, knowledge representation was ultimately completed in Neo4j, resulting in the construction of a comprehensive intelligent highway knowledge graph.