Abstract:The equipment health status of gas turbine units is easily influenced by various factors, and there exist complex coupling and transmission relationships among these factors. Traditional fault diagnosis systems often face the problem of one symptom corresponding to multiple fault modes. To address this issue, the mainstream solution currently is the fault diagnosis method based on knowledge graph. However, there is limited research on the construction and knowledge updating of the gas turbine fault knowledge base, which is why this article focuses on introducing the knowledge base construction and maintenance of gas turbine fault diagnosis based on knowledge graph. Firstly, the basic theory of ontology and the semantic reasoning principle based on ontology are explained. Secondly, the process of knowledge base construction is introduced through the acquisition and representation of domain knowledge, which avoids conflicts and redundancies in the integration of new knowledge. Finally, the feasibility of the method is verified through examples, and the stability of knowledge updating is ensured. This effectively solves the problem of "one-to-many" reasoning in conventional intelligent diagnosis, and provides a more accurate and stable knowledge graph updating method, which offers a new approach for fault detection of other mechanical equipment.