【摘要】 在矿山智能化建设中，三维巷道模型的快速、精确和自动生成是迫切需要解决的难题之一，由于巷道空间复杂交错，如何实现三维巷道模型空间交汇自动分割与连接是解决这一难题的关键。在分析巷道模型空间交汇方式的基础上，提出了一种空间交汇巷道自动分割连接方法。该方法基于分层处理框架，通过粗筛阶段、精修阶段与容错阶段的协同优化，实现了“粗筛—精修—容错”一体化处理，显著提升了三维巷道模型空间交汇处处理的稳定性与效率。首先通过AABB相交检测快速筛选潜在相交的巷道三角面，并对包围盒相交的三角面进行精确求交运算；然后将闭合交线分别加入到巷道三角面网中，基于约束Delaunay剖分对相交区域进行局部拓扑重建，利用相交多边形将巷道三角面网分割为多个子区域；完成分割之后，确定在原始三角面网内部的子三角面网，并将其删除；最终将剩余的三角面网组合形成新的巷道网格结构，实现巷道空间交汇处的完全贯通。以某矿的巷道为例，运用该方法成功实现了巷道空间交汇处的自动分割与连接，交线闭合度达100%，拓扑错误率低于2%，验证了该方法在复杂三维巷道自动建模中的有效性。更多还原

【Abstract】 In the field of intelligent mine construction, the swift, accurate, and automated generation of threedimensional roadway models remains a formidable challenge. The complex spatial interconnections inherent in roadways necessitate effective automatic segmentation and linkage at spatial intersections to address this issue. Conventional approaches frequently depend on substantial manual adjustments at roadway intersections, which are not only time-consuming and labor-intensive but also susceptible to inconsistencies in model accuracy. To mitigate these limitations, this study introduces an automatic method for spatial intersection segmentation and connection of roadways, based on a hierarchical processing framework. This framework amalgamates AxisAligned Bounding Box(AABB) pre-screening(coarse screening phase), constrained Delaunay triangulation(refinement phase), and tolerance mechanisms(error correction phase) to establish a cohesive “coarse-refinecorrect” workflow, thereby markedly enhancing the stability and efficiency of processing intricate 3D roadway intersections. The methodology is structured as follows: initially, potential intersecting triangular meshes are swiftly identified through AABB intersection detection. Subsequently, precise intersection calculations are conducted on the detected meshes to generate closed intersection lines. These lines are then integrated into the roadway’s triangular mesh network, and constrained Delaunay triangulation is employed to reconstruct the local topology of intersecting regions. By utilizing intersecting polygons, the original mesh is partitioned into multiple sub-regions. Sub-meshes located within the original mesh are identified and removed, while the remaining sub-meshes are amalgamated to create a seamless, interconnected roadway structure. This methodology facilitates the construction of high-precision and reliable 3D roadway models, significantly reducing the need for manual intervention and enhancing modeling efficiency. Validation using a real-world mine roadway model demonstrated the method’s efficacy: 63 spatial intersections(comprising 58 roadway junctions and 5 crossings) were automatically processed, achieving 100% intersection line closure and a topological error rate of less than 2%. These findings confirm the method’s capability to address complex 3D roadway modeling scenarios, offering a robust solution for intelligent mine planning and digital twin applications.更多还原