The invention discloses a transformer fault diagnosis method based on a conditional generative adversarial network. The method includes the following steps: (1) collecting transformer state historicalmonitoring data, forming a transformer fault diagnosis data set, and dividing the transformer fault diagnosis data set into a training set and a test set; (2) constructing a conditional generative adversarial network into a multi-layer sensor structure, wherein the conditional generative adversarial network has a generator and a discriminator and N outputs of the discriminator comprise: N-1 faulttypes and a judgment result about whether the data comes from analog fault data of the generator; (3) training and testing the generator and discriminator; and (4) inputting collected real-time monitoring data of the state of a transformer into the discriminator, and obtaining a fault diagnosis result of the transformer from the output of the discriminator. In addition, the invention discloses asystem based on a conditional generative adversarial network. The transformer fault diagnosis method can realize automatic identification of defects of transformer equipment.