隧道及其隐伏不良地质体三维多尺度集成建模研究

The design and construction of tunnel engineering depend on the multidimensional, multi-scale and visual representation of the indetectable unfavorable geologic body (IUGB) information. Due to the limitation of the existing multi-dimensional data model for the description of the integrated model between IUGB and the tunnel, there are some shortcomings in the multi-scale representation of the existing multi-dimensional data model. As a result, the information expression of IUGB is not comprehensive enough to meet the demand of IUGB information in every stage of well and tunnel engineering. Therefore, this project will focus on multi-scale expression of IUGB. The key problems such as 3D integrated modeling data model between IUGB and tunnel are studied. The main research contents include: (1) Based on 3D map synthesis and semantic description, the multi-scale representation method of IUGB will be researched; (2) Through the analysis of the multi-source geological exploration data, the 3D modeling data model, data structure and modeling method of the IUGB will be put forward, and the three-dimensional simulation theoretical model of the IUGB is constructed; (3) Aiming at the integration between IUGB and tunnel model, the theory and method of integrated modeling between IUGB and tunnel based on the above data model will be realized; (4) According to research and develop the experimental system, it will modify and perfect the integrated data model and modeling method. The results are expected to provide technical support for high quality and efficient shaft and tunnel engineering design and construction, and to avoid the waste of resources and potential traffic disasters caused by blind construction.

隧道工程设计与施工依赖于隐伏不良地质体信息的多维、多尺度表达。由于现有多维数据模型对隐伏不良地质体与隧道集成模型描述的局限性及在多尺度表达方面存在不足，导致不良地质体信息表达不够全面，已不能满足井巷工程各阶段对不良地质体信息的需求。为此，本项目将围绕不良地质体多尺度表达、不良地质体与隧道三维集成建模数据模型等关键问题进行研究，主要研究内容包括：（1）基于三维地图综合与语义描述，提出不良地质体的多尺度表达方法；（2）通过多源地质勘探数据分析，提出不良地质体的三维建模数据模型、数据结构及建模方法，构建不良地质体的三维模拟理论模型；（3）针对不良地质体与隧道模型的集成问题，实现基于上述数据模型的不良地质体与隧道集成建模理论与方法创新；（4）通过研发实验系统，对集成数据模型和建模方法进行完善。该研究的成果可望为优质、高效的井巷工程设计、施工提供技术支持，避免因盲目施工导致资源浪费和潜在的交通灾害。

针对隧道及线路工程遇到的隐伏不良地质体及病害的检测识别困难等问题，依据项目研究内容和研究目标，通过4年的项目研究，提出了隐伏不良地质体探测、识别及建模的新方法。（1）通过分析隐伏不良地质体在雷达图像中的典型特征，经过一维滤波、静校正、图像增益、二维滤波、带通滤波和滑动平均，从而得到三维地层信息切片；基于探地雷达二维纵剖面图像，提出一种基于三次样条插值和F-X域道插值的三维探地雷达数据重建方法，实现探地雷达数据三维重建；提出一种三维探地雷达杂波抑制与分析方法，对三维探地雷达杂波具有一定抑制作用，有效提高了杂波抑制之后的信噪比；为提取和分析探地雷达路面数据的纹理特征，提出基于线性型、对数型和指数型三种转换类型的探地雷达数据灰度级转换方法，并采用灰度共生矩阵和小波变换提取转换后的探地雷达图像纹理特征。（2）基于顶点插入的增量Delaunay算法，提出约束Delaunay四面体剖分方法，并对空间四面体格网质量进行控制/优化，实现了基于四面体的不良地质体与地层、隧道的集成建模。（3）提出一种路面裂缝灌缝病害识别方法，采用残差网络(ResNet)作为主干网络，加入空洞卷积层，提高网络对细小病害的识别性能；提出了基于改进HRNet模型的路面裂缝分割算法。该算法弥补丢失的细节信息，避免了分割边缘模糊，同时，对不同特征层的融合重新标定权重。结果表明该算法可以很好的满足实际工程的需求。.成果方面：在国际、重要会议和期刊上发表论文20多篇，培养学生方面：硕士生6名。完成预期目标。