In order to improve the accuracy of fault diagnosis of power transformers, this paper presents a fault diagnostic method based on the back propagation algorithm evolving fuzzy Petri nets. Based on BP network algorithm, the optimization of FPN parameters can be determined when the initial value of weight of FPN on the arcs, threshold value and credibility is given. Using BP network algorithm with its capacity of self-learning and self-adaption, on FPN structure, the training is carried out on the sample data of DGA. Throughout the process, the value of FPN parameter approaches the real value. The results of numerical examples show that the algorithm proposed has good classifying capability of both single fault and multiple fault samples. At the same time, the correctness and effectivity of the proposed method are verified.