Railway tunnel construction tunnel face front surrounding rock water abundance three-dimensional model construction method

Abstract

The invention relates to the technical field of tunnel advanced geological forecast, and discloses a railway tunnel construction tunnel face front surrounding rock water abundance three-dimensional model construction method comprising the following steps: collecting a plurality of transient electromagnetic survey lines in a tunnel face direction, and uniformly mapping two-dimensional inversion results to a right-handed coordinate system according to survey line angles, mileages and survey point distances; and constructing a spatial water-rich database. After geometric parameters of the tunnel face are obtained, a three-dimensional cube grid is established, and apparent resistivity values are given to nodes. A kd tree is constructed based on node distribution, a normal vector and curvature are calculated, and a mark array is initialized. And completing node clustering by adopting a three-dimensional region self-growth algorithm in combination with attribute difference, a region threshold and curvature constraint to form a water-rich region. And a binary volume data field is constructed for each clustering region, contour surfaces are extracted by using Marching Cubes, and a surrounding rock water yield three-dimensional model is generated. According to the method, high-precision three-dimensional modeling of the water-rich area in front of the tunnel face is effectively constructed, and the accuracy of advanced prediction of the tunnel is remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of tunnel advanced geological prediction technology, and more specifically, to a method for constructing a three-dimensional model of the water-rich nature of the surrounding rock in front of the tunnel face during railway tunnel construction. Background Technology

[0002] During tunnel construction, the water-bearing capacity of the surrounding rock is a key factor affecting the safety and efficiency of the project. In water-rich karst geological conditions, tunnel construction is prone to sudden water and mud inrushes, causing not only huge economic losses but also potentially serious casualties. Engineering practice shows that over 80% of tunnel safety accidents are caused by sudden mud and water inrushes during construction, posing a serious threat to the safety of people and property.

[0003] As the construction of highways and high-speed railways in my country gradually extends into deep mountainous areas, the number of tunnels built in mountainous regions is increasing. The unique and complex geological structure of mountainous areas makes geological features such as the water-bearing capacity of the surrounding rock highly complex and uncertain. Quantifying these uncertainties presents a new challenge to tunnel advance geological prediction: it requires determining the true depth, boundary location, and spatial distribution of water-bearing structures ahead of the tunnel face in three-dimensional spatial positioning information, providing more detailed information data support for taking advance construction measures.

[0004] In recent years, numerous scholars have conducted systematic research on the mechanism of tunnel water inrush, water volume prediction models, and advanced detection methods, focusing on the fundamental issue of predicting and correcting water inrush volume during tunnel construction. Regarding the optimization of transient electromagnetic data acquisition methods, Li Hui et al. used multi-line, multi-directional detection and comprehensive analysis of results to improve the prediction accuracy of groundwater-developing sections ahead of the tunnel face; Wang Chunyuan et al. used COMSOL software to establish an advanced detection model for shallow-buried deep coils and studied the electromagnetic field response characteristics; Zhang Huan et al. explored the applicability of transient electromagnetic methods in advanced detection of shallow-buried tunnels through theoretical analysis and numerical simulation; Wu Xiaogang et al. proposed apparent resistivity thresholds for common water-bearing areas based on mathematical statistics methods, which helps improve interpretation efficiency; Wang Gang et al. proposed an optimized adaptive wave field inverse transform method, while Cheng Hui et al. improved detection accuracy by constructing typical electrical models and conducting numerical simulations.

[0005] In terms of transient electromagnetic forward and inverse modeling algorithms and imaging technology, Xue Guoqiang et al. introduced secondary conductivity differential parameters to draw profile images, which achieved good results in the prediction of water body diseases; Yan Shuai et al. transformed the complex inversion process into a matrix mapping form, realizing real-time and rapid imaging of transient electromagnetic data; Zhou Jin et al. proposed a particle swarm nonlinear inversion method based on the full-space model forward model algorithm, which has high accuracy in water body identification, but its computational efficiency decreases in the later stages of iteration due to Gaussian noise.

[0006] In terms of the application of integrated geophysical exploration methods, Zhang Xiaoyu et al. combined elastic wave reflection method and transient electromagnetic method to obtain information on differences in physical properties, and combined them with groundwater dynamics model to predict water inflow; Liu Kang et al. integrated multiple methods such as seismic reflection wave method, ground-penetrating radar method and transient electromagnetic method to comprehensively interpret karst and water-rich areas.

[0007] In terms of multi-detection method fusion and three-dimensional modeling, Zhang Shuo et al. first integrated the TSP method with transient electromagnetic results in three-dimensional visualization to determine the water content of faults and delineate the spatial distribution of fracture zones and water-rich areas; Zhu Qing et al. integrated earthquake prediction and transient electromagnetic data to construct a three-dimensional voxel model to express the spatial distribution of risks; Su Bo et al. integrated multiple geological and geophysical information to carry out three-dimensional prediction research on adverse geological bodies.

[0008] In the field of transient electromagnetic 3D modeling technology, Li Xiu et al. used 3D boundary element method to achieve downward continuation imaging under curved surface conditions, improving the 3D fine detection capability of target bodies; Sun Huaifeng et al. analyzed the 3D distribution law of water-bearing structures under the tunnel floor through multi-point array detection; Xing Xiuju et al. studied the spatial distribution characteristics of apparent resistivity of water-bearing structures with the help of 3D physical simulation technology; Killingbeck et al. improved the inversion algorithm and proposed a new 3D interpretation model; Cheng Ming et al. established a complete transient electromagnetic 3D data processing flow; Xing Xiuju et al. adopted a multi-directional angle data acquisition strategy to enhance the intuitiveness of the results interpretation; Xu Caijian et al. proposed a risk decision-making framework based on digital twins; Su Maoxin et al. systematically summarized the application characteristics of 3D continuation imaging technology; Jing Xu et al. reduced the amount of computation by optimizing the grid discretization method; Sun Huaifeng used the finite difference method in the time domain to complete 3D modeling; Han Ziqiang conducted in-depth research on key technical issues in high-precision 3D advanced geological prediction.

[0009] In summary, although existing research has made significant progress in transient electromagnetic detection, inversion imaging, and 3D modeling, further exploration and improvement are still needed in the context of railway tunnel construction to effectively integrate 2D inversion data from multiple survey lines and construct a 3D model that accurately reflects the true depth, boundary location, and spatial distribution characteristics of water-bearing properties of the surrounding rock ahead of the tunnel face. Summary of the Invention

[0010] This invention aims to at least solve one of the technical problems existing in the prior art. To this end, this invention proposes a method for constructing a three-dimensional model of the water-rich nature of the surrounding rock ahead of the tunnel face in railway tunnel construction. This method effectively overcomes the shortcomings of existing technologies in terms of blurred boundaries, distorted spatial distribution, and inaccurate depth positioning in water-rich areas by using three-dimensional spatial registration of transient electromagnetic two-dimensional inversion data from multiple survey lines, spatial grid apparent resistivity reconstruction based on three-dimensional Kriging interpolation, and a three-dimensional region self-growing algorithm that integrates curvature constraints and regional difference thresholds with a MarchingCubes isosurface extraction mechanism. The goal is to improve the accuracy and reliability of advanced geological prediction for railway tunnel construction and provide high-fidelity three-dimensional spatial information support for disaster prevention and control.

[0011] To address the aforementioned problems, this invention provides a method for constructing a three-dimensional model of the water-rich nature of the surrounding rock ahead of the tunnel face during railway tunnel construction, comprising: Acquire several survey lines in a preset direction at the working face, and acquire transient electromagnetic two-dimensional inversion data for several survey lines. Each survey line includes the survey line number, distance to the origin, and corresponding apparent resistivity value. Based on the survey line acquisition angle, mileage location and measurement point distance, the transient electromagnetic two-dimensional inversion data are mapped to a unified right-handed coordinate system to construct a spatial water-rich database. Obtain the geometric parameters of the working face, construct a three-dimensional cubic mesh based on the geometric parameters of the working face, and obtain the apparent resistivity value at the mesh node using a spatial water-bearing database as input; A kd-tree structure is constructed for each node of the 3D cubic mesh, the normal vector and curvature of each node are calculated, and a marker array is initialized for region partitioning. Based on the 3D region self-growth algorithm, the differences in attributes between points, the regional difference threshold, and the curvature constraint, the 3D cube mesh nodes are clustered. A binary volume data field is constructed for each cluster region. Isosurfaces are extracted based on the MarchingCubes algorithm, and a three-dimensional surface model of the water-bearing properties of the surrounding rock is generated based on each isosurface.

[0012] Furthermore, when acquiring several survey lines in a preset direction at the tunnel face, the process includes: Acquire transient electromagnetic two-dimensional inversion data from at least 6 survey lines, including: The six survey lines shall include at least the following directions: 60° upward from the horizontal face, 30° upward from the horizontal face, horizontal direction of the face, 30° downward from the horizontal face, 60° downward from the horizontal face, and vertical direction of the face. Each survey line includes at least 12 survey points; The transient electromagnetic two-dimensional inversion data for each survey line is in the format of a triplet (survey line number, distance to origin, apparent resistivity).

[0013] Furthermore, when mapping transient electromagnetic two-dimensional inversion data to a unified right-handed coordinate system to construct a spatial water-rich database, the following steps are included: With the direction of mileage increase as the positive X-axis, vertical upward as the positive Z-axis, and the Y-axis determined by the right-hand rule, a tunnel design coordinate system is constructed. Based on the horizontal deflection angle, vertical elevation angle, and distance of each measuring point from the origin, obtain the coordinate components of each measuring point in the local coordinate system: Based on the global mileage of the current tunnel face center point and combined with the tunnel design alignment parameters, obtain the rotation and translation matrix from the local coordinate system to the global coordinate system; Based on the rotation and translation matrix, the coordinate components of each measuring point in the local coordinate system are transformed to the global coordinates of each measuring point. A spatial water-rich database is constructed based on the global coordinates of each measuring point and the corresponding apparent resistivity of each measuring point.

[0014] Furthermore, the transformation from the local coordinate system to the global coordinate system is based on the global mileage of the tunnel face center point and the horizontal and vertical curve parameters of the tunnel, which are used to construct a rotation and translation matrix.

[0015] Furthermore, when constructing a three-dimensional cubic mesh based on the tunnel face geometry and obtaining the apparent resistivity values ​​at the mesh nodes using a spatial water-bearing database as input, the following steps are taken: Based on the geometric dimensions of the tunnel face, the mesh is divided in the tunnel axis with a step size of 0.5 m, and in the horizontal and vertical directions with a step size of 0.3 m, to construct a three-dimensional cubic mesh. The apparent resistivity of known observation points in the spatial water-rich database is used as input. Three-dimensional Kriging interpolation is performed on the three-dimensional mesh nodes, where the interpolation process is as follows: The Kriging equations are constructed based on the spherical variogram model, and the weight coefficients of each observation point are solved to obtain the apparent resistivity of the grid nodes. The spherical variogram model is shown below: γ(h)=C0+ C·[1.5·(h / a)-0.5·(h / a) 3 (When h≤a); γ(h) = C0+ C (when h > a).

[0016] Furthermore, a kd-tree structure is constructed for each node of the 3D cubic mesh, the normal vector and curvature of each node are calculated, and a marker array is initialized for region partitioning, including: A kd-tree is constructed based on the spatial coordinates of 3D cubic mesh nodes and a recursive segmentation strategy, wherein the recursive segmentation strategy involves cyclically selecting the segmentation axis along the X, Y, and Z directions. Based on the spatial neighborhood query supported by kd-tree, for each grid node, obtain the set of neighboring nodes within its 1.0 m radius neighborhood, obtain the covariance matrix of the neighborhood, and use the minimum eigenvector of the covariance matrix of the neighborhood as the normal vector of the node. The curvature value of a node is determined by the standard deviation of the change in the normal vector of the node within its neighborhood.

[0017] Furthermore, based on the 3D region self-growth algorithm, inter-point attribute differences, region difference threshold, and curvature constraints, when performing region clustering on 3D cubic mesh nodes, the following steps are included: Sort all unassigned nodes in the 3D cube mesh in ascending order of curvature value, and select the node with the smallest curvature as the initial seed point and add it to the seed queue Q. When the seed queue Q is not empty, take a seed point s from Q. Traverse the set N(s) of neighboring points of seed point s, and for any neighboring point n, make a determination, where: When n is unmarked and the apparent resistivity difference |ρs-ρn| is less than 30 Ω·m, n is divided into the same region as the seed point s; and when the curvature of n is less than 0.05, n is added to the seed queue Q. When the seed queue Q is empty and there are still unlabeled nodes, select the node with the smallest curvature value from the unlabeled nodes as the new seed point, and perform a second judgment until all nodes have completed the region allocation, and output the node region label array R.

[0018] Furthermore, the set of neighboring points N(s) includes 26 neighboring nodes in the three-dimensional cube mesh that are adjacent to the face, edge, or corner of the seed point.

[0019] Furthermore, when constructing a binary volumetric data field based on each cluster region and extracting isosurfaces using the MarchingCubes algorithm to generate a three-dimensional surface model of the water-bearing properties of the surrounding rock, the process includes: For each clustering region r, construct the corresponding three-dimensional volume data Vr. If the grid node (i,j,k) belongs to region r, then let Vr(i,j,k)=1, otherwise let V_r(i,j,k)=0, so as to construct the binary volume data field of the region. Traverse each cube cell in Vr, generate an 8-bit index code based on the binary state combination of its 8 vertices, and use the index code to look up the edgeTable of the MarchingCubes algorithm to determine the intersection edge between the isosurface and the cube cell; For each edge that is determined to intersect, the coordinates of the intersection point between the isosurface and the edge are obtained according to the linear interpolation formula; Generate the corresponding triangular facets by searching the triTable based on the determined sequence of intersections; Merge the triangular patches generated from all the cubic units to construct local isosurfaces for each cluster region, and then construct a three-dimensional surface model of the water-rich surrounding rock.

[0020] Furthermore, it also includes: spatially comparing the generated 3D model with the actual excavation and exposure area, wherein: Based on spatial overlap analysis, Boolean intersection operation is performed on the regions in the 3D surface model with apparent resistivity below 200 Ω·m and the actual excavation and exposure area to obtain the product overlap rate and root mean square of boundary deviation between the 3D surface model and the actual excavation and exposure area. The 3D surface model is validated based on the product overlap rate and the root mean square of the boundary deviation. When the product overlap rate is greater than or equal to the preset product overlap rate and the root mean square of the boundary deviation is less than the preset root mean square of the boundary deviation, the 3D surface model is determined to be free of adjustment.

[0021] Compared with existing technologies, the advantages of this invention are as follows: By utilizing the acquisition angle, mileage location, and distance between measurement points of the survey line, the two-dimensional inversion results are mapped to a unified tunnel design coordinate system, constructing a structurally complete and spatially continuous water-rich database, overcoming the limitation that traditional two-dimensional profiles cannot reflect three-dimensional spatial relationships. Secondly, by constructing a high-resolution three-dimensional cubic mesh and performing Kriging interpolation on the nodes, the apparent resistivity of water-rich areas is continuously expressed in three-dimensional space. Combining kd-tree structure, neighborhood normal vectors, and curvature calculation makes region identification more accurate, and water-rich boundaries can be captured stably and reliably. Furthermore, based on a three-dimensional region self-growth algorithm, the introduction of point attribute differences, region difference thresholds, and curvature constraints enables automatic clustering and division of water-rich regions, effectively avoiding the uncertainty caused by human experience-based demarcation, and giving water-rich blocks clear spatial boundaries and connectivity structures. The three-dimensional water-rich surface model generated based on a binary volume data field and using the MarchingCubes algorithm to extract isosurfaces has high accuracy in terms of geometric morphology, boundary structure, and spatial connectivity. The three-dimensional water-rich model constructed by the method of this invention can not only accurately reflect the boundary position of the water-rich area, but also comprehensively present its spatial distribution in front of the tunnel. It provides detailed, intuitive and reliable three-dimensional information data support for the early identification of water-rich risks in tunnel construction, and helps construction units to formulate support, advanced detection and drainage measures in advance, thereby improving construction safety and decision-making accuracy. Attached Figure Description

[0022] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart illustrating a method for constructing a three-dimensional model of the water-rich surrounding rock in front of the tunnel face during railway tunnel construction, as provided in an embodiment of the present invention. Figure 2 A flowchart illustrating a method for constructing a three-dimensional model of water-rich surrounding rock in front of the tunnel face during railway tunnel construction, as provided in an embodiment of the present invention. Figure 3 Transient electromagnetic scanning profile of the tunnel face during road construction, as provided in this embodiment of the invention; Figure 4 A comparison chart of three-dimensional Kriging interpolation results provided in an embodiment of the present invention; Figure 5 A basic topology diagram provided for embodiments of the present invention; Figure 6 A flowchart of the three-dimensional model construction process provided in the embodiments of the present invention; Figure 7 A three-dimensional model diagram of the water-bearing properties of the surrounding rock provided in an embodiment of the present invention; Figure 8 This is a water-rich area map of a three-dimensional model of the water-rich properties of the surrounding rock provided in an embodiment of the present invention. Detailed Implementation

[0023] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0024] The tunnel face refers to the working face currently being excavated during tunnel excavation, and it is a key reference surface for advanced geological forecasting and construction safety control.

[0025] Transient electromagnetic two-dimensional inversion data refers to the apparent resistivity distribution data of underground media obtained by two-dimensional inversion calculation of the time-domain electromagnetic response collected by a transient electromagnetic instrument. It is used to reveal the electrical characteristics of underground structures and water-rich anomalies.

[0026] Apparent resistivity is the equivalent resistivity obtained from observation data in electromagnetic detection. It is a physical quantity used to reflect the comprehensive electrical characteristics of underground media.

[0027] A survey line is a data acquisition path laid out in a predetermined direction during geophysical exploration or engineering surveying, used to acquire continuous observation data along the line.

[0028] A spatial water-richness database is a dataset that stores the spatial coordinates of each measuring point and the corresponding water-richness indicators such as apparent resistivity in the same three-dimensional coordinate system, which is used to support subsequent three-dimensional interpolation, clustering and model reconstruction analysis.

[0029] The spherical variogram model is a commonly used form of variogram in Kriging interpolation, used to describe the changing pattern of spatial variables where the correlation gradually weakens as distance increases.

[0030] A normal vector is a vector in spatial geometry that is perpendicular to a surface or a local neighborhood fitting plane of a point, and is used to describe the local surface orientation characteristics of that point.

[0031] A binary volumetric data field is a set of three-dimensional discrete data in which each voxel in a three-dimensional mesh is assigned a value of 0 or 1 according to specific conditions to indicate whether a region belongs to the target structure.

[0032] Boolean Intersection Operation refers to finding the overlapping area occupied by two three-dimensional spatial objects to obtain the spatial intersection result.

[0033] The Marching Cubes Algorithm is a computer graphics method that generates smooth, continuous 3D isosurface models by traversing a 3D scalar field and interpolating isosurfaces within cube cells.

[0034] The triTable (Triangle Table) is a lookup table in the MarchingCubes algorithm used to indicate the connection methods of triangular facets under different vertex state combinations, and is used to quickly determine the triangular topology of isosurfaces in a cube.

[0035] Binary State Combination refers to representing each vertex of a 3D mesh cell as 0 or 1 to indicate whether it belongs to the target region, thus forming a binary combination pattern for isosurface extraction or classification.

[0036] like Figures 1-2As shown in some embodiments of this application, this embodiment provides a method for constructing a three-dimensional model of the water-rich nature of the surrounding rock in front of the tunnel face during railway tunnel construction, including: Step S100: Obtain several survey lines in a preset direction at the working face, and obtain transient electromagnetic two-dimensional inversion data for several survey lines. Each survey line includes a survey line number, distance to the origin, and corresponding apparent resistivity value.

[0037] Specifically, when acquiring several survey lines along a preset direction at the tunnel face, the process includes: acquiring transient electromagnetic two-dimensional inversion data for at least six survey lines, wherein the six survey lines include at least the following directions: 60° upward from the horizontal face, 30° upward from the horizontal face, horizontal direction of the tunnel face, 30° downward from the horizontal face, 60° downward from the horizontal face, and vertical direction of the tunnel face; each survey line includes at least 12 survey points; and the data format of the transient electromagnetic two-dimensional inversion data for each survey line is a triplet (survey line number, distance to the origin, apparent resistivity).

[0038] To achieve three-dimensional modeling of the water-bearing properties of the surrounding rock ahead of the tunnel face, this invention first lays out multiple transient electromagnetic survey lines (such as...) in a predetermined direction at the tunnel face. Figure 3 As shown in the figure, this is used to collect underground electromagnetic response information representing different spatial directions. Based on the scanning layout principles of the "Engineering Rock Mass Classification Standard" (GB / T 50218—2014), six standard observation directions are arranged around the working face, including: 60° upward horizontal, 30° upward horizontal, horizontal, 30° downward horizontal, 60° downward horizontal, and vertical. By moving the transmitting and receiving coils, a transient electromagnetic measurement profile is formed in each direction, thus constituting six survey lines with spatial distribution characteristics.

[0039] Each survey line contains at least 12 valid survey points. For each point, a set of transient electromagnetic two-dimensional inversion results is obtained, represented as a triplet: (survey line number, distance to the origin, apparent resistivity). The apparent resistivity reflects the electrical characteristics of the surrounding rock below the survey point, thus indirectly indicating water abundance. To ensure data standardization and consistency, each acquisition is processed using transient electromagnetic data processing software to export six corresponding .dat files. Each file contains at least the three columns of valid data mentioned above. The survey line numbers are integers between 1 and 12, representing the horizontal distribution from 165° to 0°.

[0040] Step S200: Based on the survey line acquisition angle, mileage location and measurement point distance, the transient electromagnetic two-dimensional inversion data is mapped to a unified right-handed coordinate system to construct a spatial water-rich database.

[0041] Specifically, when mapping transient electromagnetic two-dimensional inversion data to a unified right-handed coordinate system to construct a spatial water-bearing database, the following steps are taken: A tunnel design coordinate system is constructed with the direction of mileage increase as the positive X-axis, vertical upward as the positive Z-axis, and the Y-axis determined according to the right-hand rule; the coordinate components of each measuring point in the local coordinate system are obtained based on the horizontal deflection angle, vertical elevation angle, and distance of the measuring point from the origin for each survey line; the rotation and translation matrix from the local coordinate system to the global coordinate system is obtained based on the global mileage of the current tunnel face center point and the tunnel design alignment parameters; the coordinate components of each measuring point in the local coordinate system are transformed to the global coordinates of each measuring point based on the rotation and translation matrix; and a spatial water-bearing database is constructed based on the global coordinates of each measuring point and the corresponding apparent resistivity.

[0042] Specifically, the transformation from the local coordinate system to the global coordinate system is based on the global mileage of the tunnel face center point and the horizontal and vertical curve parameters of the tunnel, which are used to construct a rotation and translation matrix.

[0043] Based on step S100, to achieve three-dimensional fusion of survey data from different directions, this invention maps the two-dimensional inversion data of the six survey lines to the tunnel design coordinate system as shown in Table 1. By referencing the survey line direction, the distance of the measuring point from the origin, and the tunnel face mileage, precise registration of each measuring point in three-dimensional space is achieved. After coordinate transformation, all measuring points form a three-dimensional spatial water-bearing database containing spatial locations (X, Y, Z) and corresponding apparent resistivity, providing a complete, continuous, and usable data foundation for subsequent three-dimensional mesh interpolation, clustering partitioning, and water-bearing three-dimensional structural modeling.

[0044] Table 1 Two-dimensional inversion data

[0045] Specifically, to unify the multi-directional, multi-section two-dimensional inversion results into the overall tunnel space, this invention establishes a unified three-dimensional spatial framework using the tunnel design coordinate system. The X-axis represents the direction of mileage increase, the Z-axis represents the vertical upward direction, and the Y-axis is determined according to the right-hand rule. Based on the horizontal deflection angle, vertical elevation angle, and distance between measuring points for each survey line, the three-dimensional components of the measuring points in the local coordinate system at the tunnel face are first calculated, including Δx, Δy, and Δz along the forward direction, lateral direction, and vertical direction. Subsequently, based on the mileage position of the survey line and the tunnel horizontal curve and longitudinal slope parameters at that mileage, a rotation and translation matrix is ​​constructed from the local coordinate system to the global design coordinate system, mapping the local three-dimensional components to global coordinates (x, y, z).

[0046] Specifically, when mapping the data of each survey line from two-dimensional to three-dimensional coordinates, the following steps are taken: based on the horizontal deflection angle a, the vertical elevation angle b, and the distance L between the survey points, the three-dimensional coordinate components Δx, Δy, and Δz of the survey points relative to the local coordinate system of the tunnel face can be calculated.

[0047] Calculate the horizontal plane components based on the horizontal angle α: (1) (2) Calculate the vertical and forward components based on the pitch angle b: (3) (4) This yields the three-dimensional vector of the measuring point in the local coordinate system of the tunnel face: (5) The conclusion is , , Three directional components , , Then, based on mileage Calculate mileage Transformation matrix from the design coordinate system to the local coordinate system Use Formula 5 to calculate the coordinates in the design coordinate system. Thus, the apparent resistivity in the design coordinate system is obtained. All of the 6 cross-sections The data is structured and stored in the database to obtain a spatial water-rich database.

[0048] Step S300: Obtain the geometric parameters of the working face, construct a three-dimensional cubic mesh based on the geometric parameters of the working face, and obtain the apparent resistivity value at the mesh node using a spatial water-rich database as input.

[0049] Specifically, when constructing a three-dimensional cubic mesh based on the tunnel face geometry and using a spatial water-bearing database as input to obtain the apparent resistivity values ​​at the mesh nodes, the process includes: dividing the tunnel into meshes with a step size of 0.5 m along the tunnel axis and with a step size of 0.3 m in both the horizontal and vertical directions, based on the geometric dimensions of the tunnel face, to construct a three-dimensional cubic mesh; using the apparent resistivity of known observation points in the spatial water-bearing database as input, performing three-dimensional Kriging interpolation on the three-dimensional mesh nodes. The interpolation process involves: constructing a Kriging equation system based on a spherical variogram model and solving for the weight coefficients of each observation point to obtain the apparent resistivity of the mesh nodes. The spherical variogram model is as follows: γ(h) = C0 + C·[1.5·(h / a) - 0.5·(h / a)] 3 ] (when h ≤ a); γ (h) = C0 + C (when h > a).

[0050] Specifically, based on the actual geometric parameters of the working face, a three-dimensional cubic water-rich spatial mesh is divided with the same gradient. Using the water-rich database as input, the apparent resistivity value of the node position of the three-dimensional cubic water-rich spatial mesh is calculated based on the three-dimensional Kriging interpolation algorithm.

[0051] The specific parameter definitions and calculations for the 3D Kriging interpolation algorithm are as follows: The maximum distance between any two points; The number of known points in the water-rich database; : Lag distance dimension (6) : Coordinate space distance gradient partitioning interval (7) range: Spatial influence distance (beyond this distance, the independent variable is unrelated); : Indicates small-scale error (randomness); : Indicates the population variance; : Represents gradient Inner and gradient The difference between the average values ​​of the spatial distance differences within the interior space; : Represents gradient Inner and gradient The difference between the mean values ​​of the internal spatial attribute differences; and The calculation process is as follows: 1. Calculate the average of the distance difference and the average of the attribute difference between any two points within each gradient range. = The average of the distance differences between any two points with the maximum gradient The average of the distance differences between any two points with the minimum gradient.

[0052] 2. The variogram model is for coordinate points. , and parameter residuals , The kernel function is reparameterized into a linear function, let: (8) Then we can obtain: Spherical: (9) Calculate based on the given conditions. , can be obtained and : (10) 3. For any known point Place Calculation: Known Given points, calculate any point. arrive distance The calculation method is as follows: (11) according to , , and calculate : (12) Calculate any point using formulas 11 and 12 arrive of The value is then used to calculate the apparent resistivity. value: (13) Based on the aforementioned three-dimensional Kriging interpolation algorithm in this application, the apparent resistivity in each profile space is first predicted using data from five horizontal profiles in the water-bearing database. Values ​​were imaged separately, and the results were compared with those in the experimental report (e.g., Figure 4 The results (shown) are compared to demonstrate the correctness of the 3D Kriging interpolation algorithm, where: As can be seen, group a presents the prediction results of the 3D Kriging interpolation algorithm, while group b presents the results reported in the report. It is readily apparent that the method proposed in this application is essentially consistent with the results given in practical applications.

[0053] Step S400: Construct a kd-tree structure for each node of the 3D cube mesh, calculate the normal vector and curvature of each node, and initialize the marker array for region partitioning.

[0054] Specifically, a kd-tree structure is constructed for each node of the 3D cubic mesh, the normal vector and curvature of each node are calculated, and a label array is initialized for region partitioning. This includes: constructing a kd-tree based on the spatial coordinates of the 3D cubic mesh nodes and a recursive partitioning strategy, wherein the recursive partitioning strategy involves cyclically selecting the partitioning axis along the X, Y, and Z directions; based on the spatial neighborhood query supported by the kd-tree, the set of neighboring nodes within a 1.0 m radius neighborhood of each mesh node is obtained, the covariance matrix of this neighborhood is obtained, and the normal vector of the node is taken as the minimum eigenvector of the covariance matrix of this neighborhood; the curvature value of the node is taken as the standard deviation of the change in the normal vector of the nodes within the neighborhood.

[0055] Specifically, a kd-tree is constructed using the spatial coordinates of all nodes in a 3D cubic mesh as input, employing a recursive partitioning strategy. The partitioning axes are cyclically selected in the order of X, Y, and Z, ensuring efficient organization of nodes in 3D space and supporting fast neighborhood queries. Secondly, after establishing the kd-tree structure, a spatial neighborhood search is performed on each mesh node to query its set of neighboring nodes within a 1.0 m radius (or, based on a limit on the number of neighboring points, such as selecting the 20 nearest neighbors). For each node's neighborhood set, its covariance matrix is ​​calculated, and the eigenvector corresponding to the smallest eigenvalue is obtained through eigenvalue decomposition. This vector is used as the node's normal vector, reflecting the principal direction of the local spatial structure. After obtaining the node's normal vector, the standard deviation of the change in normal vectors within its neighborhood is used to characterize the curvature of the local geometric surface, and this standard deviation is defined as the node's curvature value. A larger curvature value indicates a more drastic change in the local spatial surface; a smaller curvature value indicates a smoother structure in the region.

[0056] Step S500: Based on the three-dimensional region self-growth algorithm, the differences in attributes between points, the regional difference threshold, and the curvature constraint, perform region clustering on the three-dimensional cube mesh nodes.

[0057] Specifically, based on the 3D region self-growth algorithm, point attribute differences, region difference threshold, and curvature constraint, when performing region clustering on the nodes of the 3D cube mesh, the following steps are taken: All unassigned nodes in the 3D cube mesh are sorted in ascending order of curvature value, and the node with the smallest curvature is selected as the initial seed point and added to the seed queue Q. When the seed queue Q is not empty, a seed point s is taken from Q. The set of neighboring points N(s) of the seed point s is traversed, and for any neighboring point n, a determination is made: when n is unlabeled and the apparent resistivity difference |ρs-ρn| is less than 30 Ω·m, n is divided into the same region as the seed point s; and when the curvature of n is less than 0.05, n is added to the seed queue Q. When the seed queue Q is empty and there are still unlabeled nodes, the node with the smallest curvature value is selected from the unlabeled nodes as the new seed point, and a second determination is performed until all nodes have completed region assignment, and the node region label array R is output.

[0058] Specifically, the set of neighboring points N(s) includes 26 neighboring nodes in the three-dimensional cube mesh that are adjacent to the face, edge or corner of the seed point.

[0059] Specifically, after constructing a kd-tree to achieve fast spatial neighborhood retrieval and obtaining the normal vector and curvature of each grid node, all nodes in the 3D mesh are sorted according to their curvature values ​​from smallest to largest. The smaller the curvature, the smoother the local surface, and the more likely it is to be located within the same water-rich structure. Therefore, the node with the smallest curvature is selected as the initial seed point for region growth and added to the seed queue Q.

[0060] During the region growth process (as shown in Table 2), when queue Q is not empty, a current seed point s is taken from it, and its neighboring point set N(s) is queried. The neighboring point set consists of 26 nodes that are face-adjacent, edge-adjacent, or corner-adjacent to s in the cubic grid. For any neighboring point n, if it has not yet been classified and the apparent resistivity difference |ρs-ρn| between it and s is less than 30 Ω·m, that is, the two points have sufficient consistency in water-bearing properties, then the neighboring point is assigned to the same region as s. At the same time, to ensure that the region boundary is more accurate and smoother, when the curvature value of the neighboring point n is less than 0.05, it is considered that its local structure change is small and can be added to queue Q as a new growth node.

[0061] When queue Q is empty but there are still unassigned nodes, the node with the smallest curvature value is selected from the remaining nodes as the new seed point. The region growth process is repeated until all nodes are assigned to the corresponding regions, and finally the region label array R is formed.

[0062] Table 2 Region Growing Algorithm ; As can be seen, the region growing algorithm shown in Table 2 can gradually construct a coherent water-rich distribution region based on the combined driving force of three-dimensional geometric features and apparent resistivity features.

[0063] Step S600: Construct a binary volume data field for each clustered region, extract isosurfaces based on the MarchingCubes algorithm, and generate a three-dimensional surface model of the water-bearing properties of the surrounding rock based on each isosurface.

[0064] Specifically, when constructing a binary volume data field based on each cluster region and extracting isosurfaces using the MarchingCubes algorithm to generate a three-dimensional surface model of water-bearing surrounding rock, the process includes: for each cluster region r, constructing corresponding three-dimensional volume data Vr; if a grid node (i,j,k) belongs to region r, then Vr(i,j,k)=1, otherwise V_r(i,j,k)=0, to construct the binary volume data field of that region; traversing each cube cell in Vr, generating an 8-bit index code based on the binary state combination of its 8 vertices, and using this index code to look up the edgeTable of the MarchingCubes algorithm to determine the intersection edge between the isosurface and the cube cell; for each edge determined to be intersecting, obtaining the coordinates of the intersection point between the isosurface and the edge according to the linear interpolation formula; generating corresponding triangular facets by looking up the triTable based on the determined intersection point sequence; merging the triangular facets generated from all cube cells to construct the local isosurfaces of each cluster region, and constructing a three-dimensional surface model of water-bearing surrounding rock.

[0065] To achieve a three-dimensional geometric representation of the regional clustering results, this invention constructs a binary volumetric data field for each clustered region based on the MarchingCubes algorithm and extracts the corresponding isosurfaces to generate a three-dimensional surface model of the water-bearing properties of the surrounding rock (e.g., Figure 6 Specifically, for each clustering region r, a corresponding three-dimensional volumetric data field Vr is constructed. When a cube grid node (i,j,k) belongs to region r, its voxel value is set to 1; otherwise, it is set to 0, thus forming binary volumetric data reflecting the spatial distribution of the region. This binary volumetric data serves as input, enabling accurate identification of region boundaries during isosurface extraction.

[0066] During the isosurface reconstruction phase, the algorithm traverses every cube cell in Vr. Each cell consists of 8 grid nodes, which can be combined to form an 8-bit binary index code based on their corresponding binary state combinations. Since each vertex may be inside (1) or outside (0) the isosurface, a cube has a total of 256 possible topological states, which can be simplified to 15 basic topological patterns (such as...). Figure 5(As shown). MarchingCubes quickly determines which edges of the cube cells intersect with the isosurface by looking up the edgeTable and using the index code. For the edges identified as intersecting, the algorithm uses linear interpolation to calculate the voxel scalar values ​​between the two endpoints, thereby accurately obtaining the coordinates of the intersection point between the corresponding isosurface and that edge.

[0067] Subsequently, the triTable is searched based on the index code of the cubic element to determine the triangular facet topology that should be constructed for the current element's isosurface. The triTable stores all possible combinations of triangular facet connections under all possible topological conditions, so a set of triangles can be directly generated based on the table entries, thus forming a geometric approximation of the local isosurface of the element.

[0068] Finally, by connecting and merging the local triangular patches generated from all the cubic units, a complete three-dimensional surface of the corresponding region can be constructed. Then, by using a region-based self-growing three-dimensional model generation algorithm, a three-dimensional model of the surrounding rock's water-bearing properties is generated, as follows: Figure 7 As shown, it can effectively reflect the distribution of water-rich areas in front of the surrounding rock.

[0069] Step S700: Spatial comparison is performed between the generated 3D model and the actual excavation area to verify whether the generated 3D surface model needs adjustment. Specifically: Based on spatial overlap analysis, a Boolean intersection operation is performed between the areas of the 3D surface model with apparent resistivity below 200 Ω·m and the actual excavation area to obtain the product overlap rate and root mean square deviation of the boundary deviation between the 3D surface model and the actual excavation area. The 3D surface model is verified based on the product overlap rate and root mean square deviation. If the product overlap rate is greater than or equal to a preset product overlap rate and the root mean square deviation of the boundary deviation is less than a preset root mean square deviation, then the 3D surface model does not need adjustment. If the product overlap rate is less than a preset product overlap rate and the root mean square deviation of the boundary deviation is greater than or equal to a preset root mean square deviation, then the 3D surface model needs adjustment. If the product overlap rate is greater than or equal to a preset product overlap rate and the root mean square deviation of the boundary deviation is greater than or equal to a preset root mean square deviation, then the 3D surface model needs adjustment.

[0070] Specifically, to verify the accuracy of the constructed three-dimensional model of the water-bearing properties of the surrounding rock, this invention uses spatial overlap analysis to perform a spatial Boolean intersection operation between the water-bearing areas in the model with apparent resistivity below 200 Ω·m and the actual excavated area. By calculating the ratio of the volume of the intersection area to the overall volume of the water-bearing area, the product overlap rate between the model and the actual situation can be obtained, which is used to measure the spatial coverage predicted by the model. Simultaneously, the corresponding point set between the boundary of the water-bearing area in the three-dimensional model and the actual exposed boundary is extracted, and the root mean square deviation of the boundary between the two is calculated to evaluate the geometric fit of the model's boundary prediction.

[0071] Based on the calculation results of the product overlap rate and the root mean square of the boundary deviation, this invention constructs a model verification judgment criterion: when the product overlap rate is greater than or equal to a preset threshold and the root mean square of the boundary deviation is less than a preset threshold, it indicates that the spatial prediction accuracy of the 3D model meets the requirements and no adjustment is needed; when the product overlap rate is lower than the threshold or the root mean square of the boundary deviation is greater than or equal to the threshold, it indicates that the model's prediction results deviate from reality and the model needs to be readjusted in conjunction with new field data or parameters; when the product overlap rate meets the requirements but the boundary deviation still exceeds the preset range, it is also determined that the model has local boundary deviation and local optimization is required.

[0072] After quality inspection using the aforementioned verification strategies, this invention employs a region-based self-growing 3D model generation algorithm to construct a 3D model of the water-bearing properties of the surrounding rock. As shown in Figure 8, this algorithm can accurately reconstruct the spatial morphology of the water-bearing area. By hiding areas with apparent resistivity greater than 200 Ω·m, it can further clearly present the connectivity, morphological contour, and true burial depth of the water-bearing area. Experimental results demonstrate that the region-based self-growing 3D model generation method can effectively reflect the true depth, boundary location, and 3D distribution characteristics of the water-bearing area in front of the tunnel face, providing reliable data support for advance prediction and safety decision-making during tunnel construction.

[0073] In the above embodiments, by utilizing the acquisition angle, mileage location, and distance between measurement points of the survey line, the two-dimensional inversion results are mapped to a unified tunnel design coordinate system, constructing a structurally complete and spatially continuous water-rich database, overcoming the limitation that traditional two-dimensional profiles cannot reflect three-dimensional spatial relationships. Secondly, by constructing a high-resolution three-dimensional cubic mesh and performing Kriging interpolation on the nodes, the apparent resistivity of water-rich areas is continuously expressed in three-dimensional space. Combining kd-tree structure, neighborhood normal vectors, and curvature calculation makes region identification more accurate, and water-rich boundaries can be stably and reliably captured. Furthermore, based on a three-dimensional region self-growing algorithm, the introduction of point attribute differences, region difference thresholds, and curvature constraints enables automatic clustering and partitioning of water-rich regions, effectively avoiding the uncertainty caused by human experience-based demarcation, and giving water-rich blocks clear spatial boundaries and connectivity structures. The three-dimensional water-rich surface model generated based on a binary volume data field and using the MarchingCubes algorithm to extract isosurfaces exhibits high accuracy in terms of geometric morphology, boundary structure, and spatial connectivity. The three-dimensional water-rich model constructed by the method of this invention can not only accurately reflect the boundary position of the water-rich area, but also comprehensively present its spatial distribution in front of the tunnel. It provides detailed, intuitive and reliable three-dimensional information data support for the early identification of water-rich risks in tunnel construction, and helps construction units to formulate support, advanced detection and drainage measures in advance, thereby improving construction safety and decision-making accuracy.

[0074] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program goods. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program goods embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0075] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program goods according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0076] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0077] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for constructing a three-dimensional model of the water-rich nature of the surrounding rock ahead of the tunnel face during railway tunnel construction, characterized in that, include: Acquire several survey lines in a preset direction at the working face, and acquire transient electromagnetic two-dimensional inversion data for several survey lines. Each survey line includes the survey line number, distance to the origin, and corresponding apparent resistivity value. Based on the survey line acquisition angle, mileage location and measurement point distance, the transient electromagnetic two-dimensional inversion data are mapped to a unified right-handed coordinate system to construct a spatial water-rich database. Obtain the geometric parameters of the working face, construct a three-dimensional cubic mesh based on the geometric parameters of the working face, and obtain the apparent resistivity value at the mesh node using a spatial water-bearing database as input; A kd-tree structure is constructed for each node of the 3D cubic mesh, the normal vector and curvature of each node are calculated, and a marker array is initialized for region partitioning. Based on the 3D region self-growth algorithm, the differences in attributes between points, the regional difference threshold, and the curvature constraint, the 3D cube mesh nodes are clustered. A binary volume data field is constructed for each cluster region. Isosurfaces are extracted based on the MarchingCubes algorithm, and a three-dimensional surface model of the water-bearing properties of the surrounding rock is generated based on each isosurface.

2. The method for constructing a three-dimensional model of the water-rich nature of the surrounding rock ahead of the tunnel face in railway tunnel construction as described in claim 1, characterized in that, When obtaining several survey lines in a preset direction at the working face, the following are included: Acquire transient electromagnetic two-dimensional inversion data from at least 6 survey lines, including: The six survey lines shall include at least the following directions: 60° upward from the horizontal face, 30° upward from the horizontal face, horizontal direction of the face, 30° downward from the horizontal face, 60° downward from the horizontal face, and vertical direction of the face. Each survey line includes at least 12 survey points; The transient electromagnetic two-dimensional inversion data for each survey line is in the format of a triplet (survey line number, distance to origin, apparent resistivity).

3. The method for constructing a three-dimensional model of the water-rich nature of the surrounding rock ahead of the tunnel face in railway tunnel construction as described in claim 2, characterized in that, When mapping transient electromagnetic two-dimensional inversion data to a unified right-handed coordinate system to construct a spatial water-rich database, the following steps are included: With the direction of mileage increase as the positive X-axis, vertical upward as the positive Z-axis, and the Y-axis determined by the right-hand rule, a tunnel design coordinate system is constructed. Based on the horizontal deflection angle, vertical elevation angle, and distance of each measuring point from the origin, obtain the coordinate components of each measuring point in the local coordinate system: Based on the global mileage of the current tunnel face center point and combined with the tunnel design alignment parameters, obtain the rotation and translation matrix from the local coordinate system to the global coordinate system; Based on the rotation and translation matrix, the coordinate components of each measuring point in the local coordinate system are transformed to the global coordinates of each measuring point. A spatial water-rich database is constructed based on the global coordinates of each measuring point and the corresponding apparent resistivity of each measuring point.

4. The method for constructing a three-dimensional model of the water-rich nature of the surrounding rock ahead of the tunnel face in railway tunnel construction as described in claim 3, characterized in that, When constructing a three-dimensional cubic mesh based on the working face geometry and using a spatial water-bearing database as input to obtain the apparent resistivity values ​​at the mesh nodes, the following steps are taken: Based on the geometric dimensions of the tunnel face, the mesh is divided in the tunnel axis with a step size of 0.5 m, and in the horizontal and vertical directions with a step size of 0.3 m, to construct a three-dimensional cubic mesh. The apparent resistivity of known observation points in the spatial water-rich database is used as input. Three-dimensional Kriging interpolation is performed on the three-dimensional mesh nodes, where the interpolation process is as follows: The Kriging equations are constructed based on the spherical variogram model, and the weight coefficients of each observation point are solved to obtain the apparent resistivity of the grid nodes. The spherical variogram model is shown below: γ(h)=C0 + C·[1.5·(h / a) - 0.5·(h / a) 3 (when h ≤ a); γ(h) = C0+ C (when h > a).

5. The method for constructing a three-dimensional model of the water-rich nature of the surrounding rock ahead of the tunnel face in railway tunnel construction as described in claim 4, characterized in that, A kd-tree structure is constructed for each node of the 3D cubic mesh. The normal vector and curvature of each node are calculated, and a marker array is initialized for region partitioning, including: A kd-tree is constructed based on the spatial coordinates of 3D cubic mesh nodes and a recursive segmentation strategy, wherein the recursive segmentation strategy involves cyclically selecting the segmentation axis along the X, Y, and Z directions. Based on the spatial neighborhood query supported by kd-tree, for each grid node, obtain the set of neighboring nodes within its 1.0 m radius neighborhood, obtain the covariance matrix of the neighborhood, and use the minimum eigenvector of the covariance matrix of the neighborhood as the normal vector of the node. The curvature value of a node is determined by the standard deviation of the change in the normal vector of the node within its neighborhood.

6. The method for constructing a three-dimensional model of the water-rich nature of the surrounding rock ahead of the tunnel face in railway tunnel construction as described in claim 5, characterized in that, Based on the 3D region self-growth algorithm, point attribute differences, region difference threshold, and curvature constraints, region clustering of 3D cubic mesh nodes includes: Sort all unassigned nodes in the 3D cube mesh in ascending order of curvature value, and select the node with the smallest curvature as the initial seed point and add it to the seed queue Q. When the seed queue Q is not empty, take a seed point s from Q. Traverse the set N(s) of neighboring points of seed point s, and for any neighboring point n, make a determination, where: When n is unmarked and the apparent resistivity difference |ρs-ρn| is less than 30 Ω·m, n is divided into the same region as the seed point s; and when the curvature of n is less than 0.05, n is added to the seed queue Q. When the seed queue Q is empty and there are still unlabeled nodes, select the node with the smallest curvature value from the unlabeled nodes as the new seed point, and perform a second judgment until all nodes have completed the region allocation, and output the node region label array R.

7. The method for constructing a three-dimensional model of the water-rich nature of the surrounding rock ahead of the tunnel face in railway tunnel construction as described in claim 6, characterized in that, The set of neighboring points N(s) includes 26 neighboring nodes in the 3D cube mesh that are adjacent to the face, edge or corner of the seed point.

8. The method for constructing a three-dimensional model of the water-rich nature of the surrounding rock ahead of the tunnel face in railway tunnel construction as described in claim 7, characterized in that, When constructing a binary volumetric data field based on each cluster region and extracting isosurfaces using the MarchingCubes algorithm to generate a three-dimensional surface model of the water-bearing properties of the surrounding rock, the following steps are included: For each clustering region r, construct the corresponding three-dimensional volume data Vr. If the grid node (i,j,k) belongs to region r, then let Vr(i,j,k)=1, otherwise let V_r(i,j,k)=0, so as to construct the binary volume data field of the region. Traverse each cube cell in Vr, generate an 8-bit index code based on the binary state combination of its 8 vertices, and use the index code to look up the edgeTable of the MarchingCubes algorithm to determine the intersection edge between the isosurface and the cube cell; For each edge that is determined to intersect, the coordinates of the intersection point between the isosurface and the edge are obtained according to the linear interpolation formula; Generate the corresponding triangular facets by searching the triTable based on the determined sequence of intersections; Merge the triangular patches generated from all the cubic units to construct local isosurfaces for each cluster region, and then construct a three-dimensional surface model of the water-rich surrounding rock.

9. The method for constructing a three-dimensional model of the water-rich nature of the surrounding rock ahead of the tunnel face in railway tunnel construction as described in claim 3, characterized in that, The transformation from the local coordinate system to the global coordinate system is based on the global mileage of the tunnel face center point and the horizontal and vertical curve parameters of the tunnel, which are used to construct a rotation and translation matrix.

10. The method for constructing a three-dimensional model of the water-rich nature of the surrounding rock ahead of the tunnel face in railway tunnel construction as described in claim 1, characterized in that, Also includes: The generated 3D model is spatially compared with the actual excavation and exposure area, where: Based on spatial overlap analysis, Boolean intersection operation is performed on the regions in the 3D surface model with apparent resistivity below 200 Ω·m and the actual excavation and exposure area to obtain the product overlap rate and root mean square of boundary deviation between the 3D surface model and the actual excavation and exposure area. The 3D surface model is validated based on the product overlap rate and the root mean square of the boundary deviation. When the product overlap rate is greater than or equal to the preset product overlap rate and the root mean square of the boundary deviation is less than the preset root mean square of the boundary deviation, the 3D surface model is determined to be free of adjustment.

Images