As the key equipment of electric power plant, the safe operation of steam turbine plays a vital role in practical production. At the same time, the normal and stable operation of excitation system, water-oxygen cooling system, regenerative system and other set system also play a decisive role in the safe operation of the unit. Fault diagnosis in the past mainly relied on the domain experts to judge the type of fault based on its experience, which has too much limitation, it is difficult to make effective diagnosis for the unconspicuous fault. The paper analyzed the common failures of the steam turbine regenerative system and established a typical fault set of the regenerative system. On the basis of using the fuzzy rules to establish knowledge base of fault symptom of regenerative system, a fault diagnosis method of regenerative system based on support vector machine multi-classification algorithm was proposed. Finally, the method was applied to the fault diagnosis of the regenerative system of a steam turbine. The experimental results showed that the model could effectively identify the fault of the regenerative system.