Abstract: The causes of equipment failures of hydropower stations are complex and highly correlated, so the study on the sequence relationship and probability of failure is helpful to quickly determining the cause of failure and making a troubleshooting plan.According to the expert experience and historical fault data, a knowledge graph of hydropower equipment was constructed, and an intelligent fault diagnosis algorithm based on the knowledge graph was designed.An approximate reasoning algorithm was implemented by using the Noisy Or model to realize the quantitative analysis of root causes, and the optimization calculation of troubleshooting suggestions was realized based on the graph reasoning analysis of related phenomena and entropy theory.The system provides comprehensive and detailed suggestions and explanation information, allows users to interact freely, and helps users quickly find faults.The system has the advantages of being independent on historical data, high accuracy, strong interpretability and dynamic updating, which provide an advanced platform for construction of intelligent hydropower stations.