To identify coal and gangue in a fully mechanized top-coal caving face, a coal-gangue identification approach based on IMF energy moment and SVM was proposed.Simulated signals used in the present research suggest that IMF energy moment, extracted from this approach, can effectively characterize the distribution of the signal energy along the time axis and better reflect signal characteristics in contrast with IMF energy.For data processing, firstly, the vibration signals generated by the impact of coal and gangue on the hydraulic support tail beam during top coal caving were collected; then these signals were decomposed into a series of intrinsic mode functions by the ensemble empirical mode decomposition; the first eight IMF components containing primary signal information were selected based on decomposing results to further extract the energy moment; each IMF energy moment was input into the trained support vector machine to identify two working conditions of coal caving and gangue caving.The experimental results show that this approach can effectively identify the coal-gangue vibration sample data and the average identification accuracy rate is up to 90%.