This paper proposes a method for power grid fault diagnosis based on fault coding technology, which first forms the optimal code set of fault space and then makes template matching. Aiming at the problem that the diagnostic accuracy of fault diagnosis algorithm is not high due to the lack of cleaning algorithms for front-end fault remote signal data processing, this paper proposes an algorithm of establishing discrete Hopfield neural network model for front-end data cleaning of fault remote signal data. The correlation between the faulted remote signal data is used to group the remote signal displacement data, and the algorithm proposed in this paper is used for data cleaning, and the modified domain of the algorithm is obtained by exhaustive input state data. Thus the DHNN cleaning model is established. Finally, an intelligent power grid fault diagnosis method with error correction capability is formed to realize the diagnosis of fault elements in the fault diagnosis space. The validity of Hopfield neural network information correction model and fault diagnosis model in fault component diagnosis of power grid is verified by testing the fault remote communication data of actual power grid. This work is supported by National Natural Science Foundation of China (No. 51777119) and Science and Technology Project of State Grid Shanghai Electric Power Company (No. 520900170024).