Abstract: In order to improve the accuracy of fault location in secondary equipment in the smart substation, a fault location method for secondary equipment of the smart substation based on deep learning is proposed. According to the characteristic information of different modules in the secondary equipment, the reasoning knowledge base of the fault location is combed. Combined with the self-test information of the secondary equipment, the subscription relationship of the messages and the sampled values, the characteristic information representation of the fault section is proposed. With the training rules of deep learning, a secondary equipment fault location model based on recurrent neural network is established and the location steps are given. Taking the typical smart substation line spacing as an example, the simulation verifies the effectiveness and accuracy of the proposed fault location method. It can obtain good diagnostic results under the unreliable information, and its fault tolerance performance is good.