Abstract:High-speed and accurate detecting of the thickness of the top coal seam plays an important role in the caving method. It is very significant for mastering accurately caving dynamics,guarantying the coal quality and computing exactly the recovery ratio. The elastic wave reflection method is one of the effective methods of detecting the thickness of the top coal seam. To identify the signal of the weak reflected wave is a key problem to detect the thickness of the coal seam by using elastic wave. So,the detecting signals of thickness of the coal seam are processed by means of the wavelet multi-resolution analysis. The principle to choose the wavelet base for decomposing the signals is put forward according to the properties of its compact support,continuity and high-order vanishing moment. The actual detecting signals are decomposed using Db4 wavelet and the properties of their time-frequency are analyzed. Therefore,the ability to identify the weak reflected wave and the accuracy to detect the thickness of the coal seam are considerably improved.