In order to improve the capability of fault diagnosis and localization of secondary equipment in smart substation, a fault diagnosis method of secondary equipment based on big data of smart substation is proposed. the self-test status information, power supply status information, telegram information and sampling values of secondary equipment can be monitored and analyzed by establishing a fault diagnosis information database. The characteristics of abnormal information can be studied and the information of communication breakage, data abnormality, synchronization abnormality and alarms can be represented by fault feature set simultaneously. Using deep learning to train big data, a fault diagnosis model of secondary equipment in smart substations based on recurrent neural network (RNN) is established. The effectiveness of the proposed method is verified through simulation, and the precise location of secondary equipment faults is realized.