为解决在有限煤层地质数据条件下难以构建高精度煤层的难题，提出了一种基于多源煤层数据的煤层DEM（Digital Elevation Model，数字高程模型）构建与动态精细修正方法。首先，基于采区煤层变化特征，提出结合双轨扫掠与加权融合的初始煤层DEM构建方法，通过该方法所构建的煤层DEM能够反映煤层的变化趋势，垂直方向上的模型节点RMSE(Root Mean Square Error,均方根误差)与MAE(Mean Absolute Error,平均绝对误差)分别达到0.25 m与0.2 m以下；其次，引入采区内离散分布的煤层地质数据，在原有模型基础上计算煤层DEM中相应节点处高程误差值，基于这些误差值通过拟合的方法得到残差曲面，并通过残差曲面对DEM节点进行整体性偏移。通过该方法对煤层DEM进行修正后，模型RMSE与MAE降低至0.14 m与0.12 m水平；最后，引入工作面上连续煤层地质数据，在该工作面处产生新的模型分段面，对同一区域的DEM进行划分，新划分的区域中重新构建区域煤层DEM，覆盖原有DEM，从而达到煤层DEM的动态精细修正。通过该方法对煤层DEM进行修正后，模型局部RMSE与MAE均达到0.12 m以下，模型精度水平能够为无人工作面的实现提供可靠的地质信息保障基础。

In order to solve the difficulty of constructing high-precision coal seam with limited geological data, a coal seam DEM (Digital Elevation Model) construction and dynamic fine correction method based on multi-source coal seam data was proposed. Firstly, based on the characteristics of coal seam change in mining area, an initial coal seam DEM construction method combining double-track sweep and weighted fusion is proposed. The DEM constructed by this method can reflect the change trend of coal seam,and the Model nodes RMSE and MAE in the vertical direction are less than 0.25 m and 0.2 m respectively. Secondly, the coal seam geological data distributed discretely in the mining area are introduced to calculate the elevation errors of corresponding nodes in coal seam DEM on the basis of the original model. Based on these errors, the residual surface is obtained by fitting, and the DEM nodes are integrated offset by the residual surface. The RMSE and MAE of coal seam DEM are improved to 0.14 m and 0.12 m after the DEM of coal seam is modified by this method. Finally, the continuous coal seam geological data on the working face is introduced to generate a new model segment surface at the working face, and the DEM of the same area is divided. The regional coal seam DEM is reconstructed in the new divided area to cover the original DEM, so as to achieve dynamic fine correction of coal seam DEM. After the DEM of coal seam is modified by this method, the local RMSE and MAE of the model reach below 0.12 m, which can provide a reliable basis for geological information guarantee for the realization of unmanned working face.