Abstract: In order to solve the problem that it is difficult to accurately identify the fault line when single phase grounding faults occur in coal mine distribution networks, a method of single phase grounding fault line selection based on improved adaptive noise complete set Empirical Mode decomposition（ICEEMDAN） based on grey Wolf algorithm optimization support vector machine（GWO-SVM） is proposed. ICEEMDAN was used to decompose the zero-sequence current of each line under different working conditions. The first modal component IMF₁ after decomposition of each line was determined to obtain the margin factor, and the residual was determined to obtain the relative energy coefficient and comprehensive correlation coefficient. Then the three eigenvalues were combined to form the feature vector, which was finally input into GWO-SVM for pattern recognition. A 10 kV distribution network simulation model containing four feeders was built by Matlab/Simulink for verification. The simulation results show that the proposed method is not affected by the fault location, initial fault Angle and transition resistance, and has a high line selection accuracy.