Abstract: One of the prominent problems of intelligent mining is its inadaptability to geological conditions, especially for the absolute accuracy of coal seam floor and thickness. 3D seismic has a high horizontal resolution and it can control the ups and downs of the coal seam. However, due to the time domain data, the calculation of the coal seam floor elevation is negatively affected by the time-depth conversion point. On the basis of the 3D seismic data of the working face and the coal thickness data during the mining process, the absolute accuracy of the time-depth conversion of the coal seam floor is improved by continuously refreshing the velocity field. At the same time, the iterative interpolation algorithm is used to continuously update the coal seam thickness of the working face, then error statistics and analysis are conducted based on the calculated data. The experiment was carried out at TJH304 working face after using coal seam floor height and thickness of working face tunnel and mining face combined with the dynamic interpretation of the 3D seismic data. The absolute error of the coal seam floor elevation and thickness values in front of the working face is reduced. In particular, the four verification points within thirty meters from the current mining face and the coal seam floor elevation error is between 0.37-0.58 m; the coal seam thickness error is between 0.32-0.44 m. The results show that the 3D seismic dynamic interpretation technology can maximize the effective combination of 3D seismic and downhole production data, continuously improve the spatial accuracy of coal seam, and provide a prospective coal seam model for intelligent mining.