In order to recognize the coal-rock interface in the top coal caving, the vibration signals of the tail beam of the hydraulic support are investigated. A new method based on empirical mode decomposition (EMD) and neural network is proposed. Firstly, EMD is employed to decompose the original signal into a number of stationary intrinsic mode functions (IMF). Subsequently, a further analysis of the IMF is chosen to extract the characteristic parameters. Finally, the parameters that contained the main information compose feature vectors and are chosen as the input of the neural network to identify the coal-rock interface. The experimental result shows that kurtosis and crest factor of the IMFs can all be used to identify the coal-rock interface and a higher identification rate is achieved using energy than the other two methods.