Abstract: The DGA data of oil-immersed transformer contains a lot of fault information, which can be excavated and analyzed to achieve fault diagnosis of oil-immersed transformer. However, between characteristic gas information and transformer fault type and degree, there is a complex nonlinear mapping relationship. Therefore, it is very difficult to judge the fault of transformer. This paper summarizes the traditional fault analysis methods, such as three ratio method and the existing intelligent diagnosis methods, such as expert system, fuzzy theory and machine learning, and analyzes the principles and shortcomings of each method. In addition, using the strong classification performance of decision tree, a transformer fault diagnosis model based on decision tree is proposed. The experimental results show that this method has certain advantages over the traditional three ratio method. Finally, some ideas are provided for the future research of DGA data intelligent algorithm to improve the accuracy of transformer fault diagnosis.