Abstract: Seismic waveform classification technology has the characteristics of statistics of the overall change of the seismic signal and reflects the distribution of this change. It is an important extension of seismic attribute analysis technology. It has a good application effect on the reflected wave changes caused by geological anomalies, which is similar to the conventional single attribute prediction. Compared, it has the characteristics of sensitive reflection and reliable results. High density 3D seismic data have the characteristics of high signal-to-noise ratio, high resolution and high fidelity. This paper attempts to use waveform classification technology to predict the occurrence of coal seam and magmatic intrusion area reflected by high density 3D seismic data, and studies the interpretation method of collapse column. Through actual exposure and drilling verification in underground roadways, the prediction results have high accuracy and accurate bounds, which can provide accurate geological data for coal mining.