面对煤矿井下低照度、弱纹理、高粉尘等复杂条件，现有煤矿巷道三维建模方法存在成本高、时效性差和精度低等问题，提出了一种基于已知点约束的高精度煤矿巷道三维点云建模方法。首先，通过体素滤波器对激光雷达点云数据进行降采样，并对降采样后的激光雷达点云数据利用迭代最近点（ICP）匹配提取出局部点云地图，结合惯性测量单元（IMU）数据对点云数据进行畸变校正；其次，利用ICP配准局部点云地图和畸变校正后的点云地图，以提高前端配准的精度和效率，并在后端加入回环检测来提高煤矿巷道定位与建图精度；然后，通过附合导线控制测量获取煤矿巷道分段已知点坐标，为点云建模提供全局约束条件；最后，将已知点和激光雷达同时定位与建图（SLAM）确定的测站点进行联合平差计算，对测站点坐标进行校正，并进一步利用非线性优化方法校正全局点云地图坐标，从而提高三维点云建模精度。实验结果表明：该方法构建的煤矿巷道三维点云地图具有较好的全局一致性和几何结构真实性，在煤矿井下具有较高的定位与建图精度。

In response to the complex conditions such as poor lighting, weak texture, and high concentrtaions of dust in underground coal mines, the existing 3D modeling methods for coal mine roadways have the disadvantages of high costs, poor timeliness, and low accuracy. A high-precision 3D point cloud modeling method for coal mine roadways based on known point constraints was proposed. The LiDAR point cloud data was downsampled by voxel filter, followed by the use of iterative closest point (ICP) matching for the downsampled LiDAR point cloud data to extract local point cloud maps. The point cloud data was then distortion-corrected using inertial measurement unit (IMU) data. ICP was utilized to align the local point cloud maps with the distortion-corrected point cloud maps, improving the accuracy and efficiency of front-end registration. Loopback detection was incorporated in the back-end to enhance the accuracy of coal mine roadway localization and mapping. The coordinates of the known points of the coal mine roadways were obtained through control measurements using connecting traverse, providing global constraints for point cloud modeling. A combined adjustment calculation was performed on the known points and the station points determined by LiDAR simultaneous localization and mapping (SLAM). The station point coordinates were corrected, and a nonlinear optimization method was further employed to adjust the global point cloud map coordinates, thereby improving the accuracy of 3D point cloud modeling. Experimental results demonstrated that the 3D point cloud map of coal mine roadways constructed by this method had high global consistency and geometric structure authenticity, achieving high localization and mapping accuracy in underground coal mines.