As the core equipment of traction power supply and transformation system, transformer faults will affect the safety and stability of rail transit operation. Therefore, timely detection of potential faults is conducive to reducing fault costs and improving the reliability of the system. At present, many researches on early fault diagnosis of transformer are focused on the application of monitoring quantity, and the diagnosis and recognition results are fuzzy. In this paper, the technical methods of transformer fault diagnosis and the monitoring parameters used in the fault diagnosis process are summarized. Then the development of transformer fault diagnosis technology is summarized according to the difference of transformer fault diagnosis model method and monitoring state. The shortcomings of transformer fault diagnosis technology as well as the future development direction and challenges are discussed in this paper, which provides some reference value for fault prediction.