Abstract: Since the Three Mile Island Nuclear Power Plant accident happened in U.S., the role of fault diagnosis in nuclear power plants has been gotten more attention. To explore the fault diagnosis of nuclear power plants in China, the intelligent fault diagnosis method named DUCG (dynamic uncertain causality graph) was described, which can be applied to a complex system of uncertainty. A model based on Unit 1 of Ningde Nuclear Power Plant, CGNPC (China Guangdong Nuclear Power Group) was established and the model was validated by the corresponding simulator, ensuring the validation of effectiveness. Validation results show that the DUCG method can accurately identify the typical nuclear power plant faults and can also show the development process of faults, successfully getting what caused the failure event and the corresponding probabilities.