Abstract: Abstract: In the complex underground environment of coal mine, the vibration process of explosion-proof electrical equipment in coal mine will lead to the loosening of fasteners or wear of parts, change the mechanical structure of the equipment, affect the signal of explosion-proof electrical equipment, make the boundary between normal signal and fault signal become blurred, resulting in low accuracy of fault detection. Therefore, an automatic vibration fault detection method based on single-class Support Vector Machine (OCSVM) is proposed. The normal state characteristics and vibration fault state characteristics of the equipment are constructed, and the normal state feature sequence is taken as the decision boundary learning target of OCSVM kernel function. According to the nonlinear and high-dimensional characteristics of the vibration fault of the explosion-proof electrical equipment in coal mine, the kernel function of OCSVM is selected as polynomial kernel. Combined with grid search and K-cross validation, OCSVM parameters are optimized. By obtaining the optimal solution of OCSVM objective function and finding the optimal decision boundary, the automatic detection of vibration fault is realized. It is found that the proposed method can effectively detect the vibration failure of explosion-proof electrical equipment in coal mine, and the accuracy rate is up to 98.25%, which can provide a strong guarantee for the safe operation of electrical equipment.