Abstract: Due to the limitations of knowledge reasoning models, traditional blast furnace expert systems in China have gradually shown disadvantages in the increasingly complex and changeable smelting environment, making it difficult to achieve large-scale promotion across blast furnace scenarios. Knowledge graph is a new information management retrieval technology based on graph structure data. Firstly, the construction mechanism of China′s traditional blast furnace expert system was sorted out, including three knowledge reasoning modes based on rules and logic, typical cases and metallurgical mathematical models, and the current pain points were analyzed in depth.Secondly, the construction method of knowledge graph under the background of big data of blast furnace was explored. Based on the previous work, an example of knowledge graph network with vanadium content in hot metal as the target node was constructed, and the relevant parameter regulation enrichment network was drawn. The advantages of knowledge graph in decoupling multi-parameter correlation were verified. Finally, the application directions of knowledge graph in blast furnace expert systems was summarized, including but not limited to the retrieval and management of multimodal data, decoupling the effects of multiple parameters, dynamic control and optimization of processes, intelligent diagnosis of furnace conditions, dynamic generation and recommendation of intelligent decisions. Knowledge graph can effectively address issues such as the high complexity of traditional blast furnace expert systems, one-way retrieval, and insufficient adaptive capabilities, transforming traditional rule-based parameter control methods into data-driven intelligent decisions in all aspects. The efficient use of knowledge graph technology is an important direction for the transformation and upgrading of blast furnace expert system in the future.