The accurate identification of water‐bearing structures is urgently required for the safe construction of tunnel engineering. Currently, the direct current resistivity method is an effective method for detecting water‐bearing structures in tunnels. In the advanced detection of the direct current resistivity based on the finite element method, the traditional hexahedron mesh performs poorly for the discretization of models of complex tunnel structure sections such as horseshoe‐shaped and round sections. Therefore, this study adopts unstructured grid generation technology combining tetrahedra and hexahedra to achieve more accurate modelling of complex structures, such as round and horseshoe‐shaped sections, and establishes a forward modelling method of the direct current resistivity in tunnels based on an unstructured mesh. The maximum error between the numerical simulation and theoretical results for an infinite tabular body in full space is less than 0.8%. It is more complicated to calculate the sensitivity matrix and model constraint term for the inversion region containing two types of grid than for one. For this purpose, the sensitivity matrix of different types of grid areas is calculated, a model constraint term based on the dual constraints of volume and distance is constructed, and finally, a partitioned domain‐weighted least‐squares inversion method based on an unstructured mesh is proposed. Synthetic examples of typical water‐bearing structures are analysed, and the results show that the proposed forward and inverse methods of the direct current resistivity in tunnels based on an unstructured mesh can effectively capture the position and morphology of the water‐bearing structure. Finally, an on‐site application was conducted in the Yellow River Diversion Project in central Shanxi. The proposed method could effectively identify the water body in front of the tunnel face and guide the on‐site construction of the project. These results can improve the interpretation of the direct current resistivity data in tunnels and play a positive role in promoting the use of the direct current resistivity method to prevent and control water‐inrush disasters in tunnels with complex structures.

中文翻译：

---

基于非结构网格的直流电阻率正反演方法及其在隧道工程中的应用

准确识别含水结构是隧道工程安全施工的迫切需要。直流电阻率法是目前检测隧道含水结构的有效方法。在基于有限元法的直流电阻率高级检测中，传统六面体网格对于马蹄形、圆形断面等复杂隧道结构断面模型的离散化效果较差。因此，本研究采用四面体和六面体相结合的非结构化网格生成技术，实现对圆形、马蹄形断面等复杂结构的更精确建模，建立了基于非结构化网格的隧道直流电阻率正演建模方法。全空间无限板状体的数值模拟与理论结果之间的最大误差小于0.8%。包含两种网格的反演区域的灵敏度矩阵和模型约束项的计算比一种网格更为复杂。为此，计算不同类型网格区域的灵敏度矩阵，构造基于体积和距离双重约束的模型约束项，最后提出基于非结构化网格的分区域加权最小二乘反演方法被提议。对典型含水结构的综合算例进行了分析，结果表明，所提出的基于非结构化网格的隧道直流电阻率正反演方法能够有效捕捉含水结构的位置和形态。最后在晋中引黄工程中进行了现场应用。该方法可有效识别掌子面前方水体，指导工程现场施工。这些成果可以完善对隧道直流电阻率数据的解释，对推广应用直流电阻率法防治复杂结构隧道突水灾害具有积极作用。