In order to accurately and quickly detect the faults of distribution network and ensure the stable power supply, a distribution network fault detection method based on edge computing is proposed. The edge computing model is constructed through multiple fault detectors, cloud computing centers and terminal equipment. Based on the edge side, combined with multi-dimensional S-transform fusion algorithm and phase modulus transformation matrix method, the relevant fault features of distribution network are extracted. According to the feature extraction results, the correlation matrix of each detection device branch and its element value conditions of the edge computing model are set after the fault information is corrected, the fault information matrix is constructed, and the fault on the line section is determined according to the value of this matrix element. Experimental results show that the distribution network fault detection method based on edge computing can effectively detect different kinds of faults, and the response time is less than 0.005 s, the detection accuracy is more than 95%, which meets the practical application requirements.