为精准判识节理岩体隧道施工中的潜在失稳块体并做出针对性支护方案，提出一种融合点云与结构面重建的失稳块体判识及针对性支护方法。首先，通过三维激光扫描获取隧道掌子面点云数据，采用基于DPC（density peaking clustering）改进的DBSCAN（density-based spatial clustering of applications with noise）算法（DPC-DBSCAN），实现结构面的自适应聚类与信息提取，克服传统方法中参数依赖人工调试、效率低、精度不足的问题； 然后，对获取的结构面参数实现DFN（discrete fracture network）重建，结合现场地应力与围岩力学参数，构建合成岩体模型（SRM），并利用三维离散元程序（3DEC）模拟隧道开挖过程，实现潜在失稳块体的三维动态识别与稳定性评价； 最后，以安徽西武岭隧道为依托，提出并对比6种支护方案，明确靶向支护的关键参数与实施策略。研究表明： 1）提出的DPC-DBSCAN算法在结构面聚类中表现出更高的效率与精度，其运行时间较传统算法减少约38%，有效提升了结构面信息提取的自动化程度与可靠性； 2）基于点云提取的结构面信息参数与现场人工测量结果吻合良好，验证了该智能识别方法的正确性与工程适用性； 3）针对失稳块体提出的靶向支护方案，通过局部加长锚杆至5 m，加密环距至0.6 m，并优化锚杆角度，使失稳块体最大位移降至1.1 mm，显著提升了支护效果，同时在保证安全的前提下降低了材料用量，具备良好的工程经济性。
To accurately identify potentially unstable blocks during the construction of jointed rock tunnels and develop targeted support schemes, this study proposes a method integrating point cloud technology and structural plane reconstruction for block instability identification and support design. First, point cloud data of the tunnel face are acquired through three-dimensional (3D) laser scanning. The density-based spatial clustering of applications with noise (DBSCAN) algorithm is enhanced using density peaking clustering (DPC), yielding the DPC-DBSCAN algorithm. This approach enables adaptive clustering and information extraction of structural planes, overcoming the limitations of traditional methods that rely on manual parameter tuning and providing increased efficiency and accuracy. Subsequently, a discrete fracture network is reconstructed based on the extracted structural plane parameters. Combined with insitu stress and surrounding rock mechanical parameters, a synthetic rock mass model is established. The tunnel excavation process is simulated using 3DEC discrete element software to achieve 3D dynamic identification and stability assessment of potentially unstable blocks. Finally, the Xiwuling tunnel in Anhui, China, is used as an engineering case in which six support schemes are proposed and compared to determine the primary parameters and implementation strategies for targeted support. The main conclusions are as follows: (1) the proposed DPC-DBSCAN algorithm improves efficiency and accuracy in structural plane clustering, reducing runtime by approximately 38% compared to traditional algorithms and enhancing the automation and reliability of structural plane information extraction; (2) the structural plane parameters extracted from point cloud data agree with field manual measurements, validating the accuracy and engineering applicability of the intelligent identification method; and (3) the targeted support scheme for unstable blocks, which includes locally extending anchor bolts to 5 m, reducing bolt spacing to 0.6 m, and optimizing bolt angles, reduces the maximum displacement of unstable blocks to 1.7 mm. This markedly improves support effectiveness compared to conventional schemes while reducing material usage and maintaining safety.
