Three-dimensional reconstruction analysis method for hydraulic concrete sampling data

Abstract

The application provides a hydraulic concrete sampling data three-dimensional reconstruction analysis method, relates to the technical field of concrete construction quality data monitoring of hydraulic engineering, and comprises the following steps: acquiring a three-dimensional laser scanning point cloud coordinate set of a surface of a hydraulic concrete core sample, performing uniform voxel grid division processing operation on the coordinate set along three orthogonal directions in space to obtain an initial voxel grid model; according to the initial voxel grid model, extracting an interface point cloud sequence at a junction of aggregate particles and a cement matrix to obtain an interface point cloud coordinate data set; and according to the interface point cloud coordinate data set, solving partial derivatives of each sampling point coordinate in the interface point cloud sequence to obtain an interface curvature distribution field. The application realizes integrated three-dimensional reconstruction of the geometric morphology of the aggregate-cement matrix interface and the internal pore volume characteristics.

Description

Technical Field

[0001] This invention relates to the field of monitoring data on the construction quality of concrete in water conservancy projects, and in particular to a three-dimensional reconstruction and analysis method for hydraulic concrete sampling data. Background Technology

[0002] In the long-term operation of hydraulic structures such as sluice gates, pumping stations, and water conveyance channels, concrete durability assessment is a core task for the safe operation and maintenance of these projects. The three-dimensional morphology of the interface transition zone between aggregates and the cement matrix within the hydraulic concrete, as well as the volume and connectivity of harmful pores, are key characterizing parameters for evaluating the structure's impermeability, resistance to freeze-thaw damage, and degree of degradation due to water seepage and erosion. To obtain these parameters, high-precision three-dimensional reconstruction of concrete core samples drilled on-site is required to obtain accurate microstructural data that can support subsequent hydraulic engineering simulation analyses.

[0003] Currently, the 3D reconstruction of concrete core samples generally adopts a technical approach that separates interface morphology analysis from pore volume calculation. In terms of interface characterization, it often relies on 2D slice edge detection to extract the aggregate-matrix boundary and calculate interface curvature, resulting in dimensionality reduction of the 3D interface information and making it difficult to reflect the true spatial distribution of the interface. In terms of pore characterization, indirect experimental methods such as mercury intrusion porosimetry and gas adsorption are often used to infer the total pore volume or statistically analyze pore size distribution, failing to obtain the spatial location, connectivity properties, and correspondence between pores and interface regions. More critically, there is no spatial dimensional correlation between interface curvature data and pore volume data, causing the reconstructed 3D model to fail to describe the spatial correspondence between weak areas of the interface and various harmful pores such as connected capillaries, freeze-thaw expansion pores, and seepage erosion pores. Therefore, the resulting 3D model cannot be directly used for hydraulic engineering simulation assessments such as pump station seepage path simulation and freeze-thaw damage evolution analysis, severely restricting the reliability of accurate evaluation of the durability of concrete construction in hydraulic engineering projects.

[0004] Therefore, there is an urgent need for a reconstruction analysis method that can uniformly represent the interface geometry and internal pore volume characteristics of hydraulic concrete core samples in three-dimensional space, so as to establish the spatial relationship between the weak interface area and different types of pores, and provide a refined three-dimensional model that can be directly applied for special simulations of seepage, freeze-thaw and other water conservancy projects. Summary of the Invention

[0005] This invention provides a three-dimensional reconstruction and analysis method for hydraulic concrete sampling data, which realizes integrated three-dimensional reconstruction of the geometric morphology of the aggregate-cement matrix interface and the internal pore volume characteristics.

[0006] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows: Firstly, a three-dimensional reconstruction and analysis method for hydraulic concrete sampling data, the method comprising: Step 1: Obtain the coordinate set of the three-dimensional laser scanning point cloud on the surface of the hydraulic concrete core sample, and perform uniform voxel meshing on the coordinate set along three orthogonal directions in space to obtain the initial voxel mesh model. Step 2: Based on the initial voxel mesh model, extract the interface point cloud sequence at the interface between aggregate particles and cement matrix to obtain the interface point cloud coordinate data set. Step 3: Based on the interface point cloud coordinate data set, solve the partial derivatives of the coordinates of each sampling point in the interface point cloud sequence to obtain the interface curvature distribution field. Step 4: Based on the interface curvature distribution field, calculate the shortest spatial path length of adjacent sampling points along the aggregate surface in the interface point cloud sequence to obtain the geodesic distance sequence of the pore boundary. Step 5: Based on the geodesic distance sequence of pore boundaries, identify candidate boundary points for different types of pores through hierarchical mutation determination, and extract an independent pore boundary point cloud set with type and connectivity attributes through connected clustering; construct an appropriate spatial convex hull volume integral domain for different types of pores, perform three-dimensional volume integral operation and classify and statistically analyze pore volumes to obtain a set of pore parameters containing hydraulic engineering evaluation indicators. Step 6: Based on the actual pore volume parameters, fuse the interface curvature distribution field to obtain a three-dimensional reconstructed mesh model that includes the interface geometry and internal pore volume characteristics. Step 7: Based on the three-dimensional reconstruction mesh model, obtain the three-dimensional reconstruction model of the concrete sampling data.

[0007] The above-described solution of the present invention has at least the following beneficial effects: A high-precision point cloud coordinate set of concrete core sample surface is obtained by 3D laser scanning. After voxel meshing, the interface point cloud sequence at the interface between aggregate particles and cement matrix is ​​extracted, and the interface curvature distribution field is constructed based on partial derivatives. By calculating the geodesic distance of adjacent sampling points in the interface point cloud sequence along the aggregate surface, the boundary point cloud set of micropores inside the concrete is accurately extracted. The true volume parameters of the pores are obtained by volume integral operation. The interface curvature distribution field and the true volume parameters of the pores are fused to generate a 3D reconstructed mesh model that includes the interface geometry and the internal pore volume characteristics. This achieves integrated 3D reconstruction of the aggregate-cement matrix interface geometry and the internal pore volume characteristics, overcoming the technical defect of the deviation between the geometric accuracy and physical reality of the reconstruction model caused by the traditional interface-pore separation analysis. Attached Figure Description

[0008] Figure 1 This is a flowchart illustrating the three-dimensional reconstruction analysis method for hydraulic concrete sampling data provided in an embodiment of the present invention.

[0009] Figure 2This is a schematic diagram of a three-dimensional reconstruction and analysis system for hydraulic concrete sampling data provided in an embodiment of the present invention.

[0010] Figure 3 The present invention provides a schematic diagram of a three-dimensional reconstruction analysis method for hydraulic concrete sampling data, which calculates the shortest spatial path length of adjacent sampling points along the aggregate surface in the interface point cloud sequence based on the interface curvature distribution field, and obtains the geodesic distance sequence of pore boundaries. Detailed Implementation

[0011] 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 so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0012] like Figure 1 As shown, embodiments of the present invention propose a three-dimensional reconstruction analysis method for hydraulic concrete sampling data, the method comprising the following steps: Step 1: Obtain the coordinate set of the three-dimensional laser scanning point cloud on the surface of the hydraulic concrete core sample, and perform uniform voxel meshing on the coordinate set along three orthogonal directions in space to obtain the initial voxel mesh model. Step 2: Based on the initial voxel mesh model, extract the interface point cloud sequence at the interface between aggregate particles and cement matrix to obtain the interface point cloud coordinate data set. Step 3: Based on the interface point cloud coordinate data set, solve the partial derivatives of the coordinates of each sampling point in the interface point cloud sequence to obtain the interface curvature distribution field. Step 4: Based on the interface curvature distribution field, calculate the shortest spatial path length of adjacent sampling points along the aggregate surface in the interface point cloud sequence to obtain the geodesic distance sequence of the pore boundary. Step 5: Based on the geodesic distance sequence of pore boundaries, identify candidate boundary points for different types of pores through hierarchical mutation determination, and extract an independent pore boundary point cloud set with type and connectivity attributes through connected clustering; construct an appropriate spatial convex hull volume integral domain for different types of pores, perform three-dimensional volume integral operation and classify and statistically analyze pore volumes to obtain a set of pore parameters containing hydraulic engineering evaluation indicators. Step 6: Based on the actual pore volume parameters, fuse the interface curvature distribution field to obtain a three-dimensional reconstructed mesh model that includes the interface geometry and internal pore volume characteristics. Step 7: Based on the three-dimensional reconstruction mesh model, obtain the three-dimensional reconstruction model of the concrete sampling data.

[0013] In this embodiment of the invention, by constructing an initial voxel mesh model and accurately extracting the interface point cloud between aggregate and cement matrix, the interface curvature distribution field is obtained by solving the partial derivatives to accurately quantify the three-dimensional spatial morphology of the interface transition zone, effectively overcoming the geometric distortion problem at the interface junction in traditional reconstruction methods. The geodesic distance between adjacent sampling points along the aggregate surface is calculated by combining the interface curvature distribution field to accurately define the boundary of micropores, and the volume integral operation is performed on it to obtain the real pore volume parameters, eliminating the cumulative error in the pore volume calculation process. The interface geometry and internal pore volume characteristics are deeply integrated to generate a three-dimensional reconstruction mesh model, improving the geometric fidelity and volume calculation accuracy of the three-dimensional reconstruction of the internal microstructure of concrete.

[0014] In a preferred embodiment of the present invention, step 1 above may include: Step 11: Based on the obtained three-dimensional laser scanning point cloud coordinate set of the hydraulic concrete core sample surface, perform statistical outlier filtering and spatial registration and alignment processing on each sampling point in the point cloud coordinate set to obtain a normalized point cloud coordinate set. Specifically, this includes: for each sampling point in the point cloud coordinate set, searching for the spatially closest neighboring sampling points, calculating the arithmetic mean of the spatial Euclidean distances from the sampling point to all neighboring sampling points, and obtaining the neighboring points of each sampling point. The average distance value of the neighborhood is calculated. Based on the average distance value of the neighborhood of each sampling point, the overall arithmetic mean and standard deviation of the average distance values ​​of the neighborhood of all sampling points are calculated to obtain the neighborhood distance statistical parameters. Based on the neighborhood distance statistical parameters, an upper limit for outlier determination distance is set, which is composed of several multiples of the overall arithmetic mean and standard deviation. Sampling points with a neighborhood average distance value greater than the upper limit for outlier determination distance are identified as statistical outliers and removed. Sampling points with a neighborhood average distance value less than or equal to the upper limit for outlier determination distance are retained to obtain the set of point cloud coordinates after outlier removal. Based on the point cloud coordinate set after outlier filtering, corresponding feature points are extracted from the point cloud data collected from different scanning stations within the spatial overlap area of ​​adjacent station point clouds. A corresponding matching relationship is established between the corresponding feature points of adjacent station point clouds, resulting in point cloud matching point pairs between stations. Based on the point cloud matching point pairs between stations, with the goal of minimizing the sum of spatial Euclidean distances between matching point pairs, the spatial rigid body rotation transformation and translation transformation amounts between adjacent station point clouds are iteratively solved to obtain the spatial registration transformation parameters between stations. Based on the spatial registration transformation parameters between stations, the point cloud coordinates of each scanning station are uniformly transformed to the same spatial reference coordinate system, completing the spatial registration and alignment of multi-station scanning data, and obtaining a normalized point cloud coordinate set.

[0015] Step 12: Based on the normalized point cloud coordinate set, calculate the difference between the maximum and minimum coordinate values ​​of the point cloud in the X-axis, Y-axis, and Z-axis directions to obtain the spatial span range of the point cloud in three orthogonal directions. Divide the spatial span range of the point cloud in each direction by the preset voxel element side length value of the corresponding hydraulic component and round up to obtain the number of voxel meshes in each direction. This determines the voxel meshing parameters. Specifically, based on the obtained normalized point cloud coordinate set, traverse the spatial coordinates of all sampling points in the point cloud coordinate set, and extract the maximum and minimum coordinate values ​​of all sampling points in the X-axis direction, the Y-axis direction, and the Z-axis direction to obtain the coordinate extreme values ​​in the three orthogonal directions. Based on the coordinate extreme values ​​in the three orthogonal directions, calculate the difference between the maximum and minimum coordinate values ​​in the X-axis direction, the Y-axis direction, and the Z-axis direction to obtain the spatial span range of the point cloud in the X-axis, Y-axis, and Z-axis directions. Based on the spatial span of the point cloud in three orthogonal directions, the pre-defined voxel element side length value for the component type of the hydraulic concrete core sample is selected to obtain the voxel element side length parameter. Based on the spatial span of the point cloud and the voxel element side length parameter, the spatial span of the point cloud in the X-axis, Y-axis and Z-axis directions is divided by the voxel element side length parameter and the division result is rounded up to obtain the number of voxel meshes in the X-axis, Y-axis and Z-axis directions respectively. Based on the number of voxel meshes in the three orthogonal directions, the number of voxel meshes in the three directions is combined to form the voxel meshing parameters along the three orthogonal directions, thus obtaining the voxel meshing parameters.

[0016] Step 13: Based on the voxel grid partitioning parameters, for each sampling point in the normalized point cloud coordinate set along the three orthogonal directions of the X-axis, Y-axis, and Z-axis, voxel units are assigned according to the grid index number of the sampling point's spatial coordinates to obtain the mapping relationship data between voxel units and the point cloud. Specifically, this includes: based on the obtained voxel grid partitioning parameters, using the minimum coordinate value of all sampling points in the normalized point cloud coordinate set in the X-axis, Y-axis, and Z-axis directions as the spatial origin of the voxel grid, and using the voxel unit side length parameter as the partitioning step size of the voxel grid in each direction, a uniform voxel grid spatial partitioning framework is established along the three orthogonal directions to obtain the voxel grid spatial partitioning framework; based on the voxel grid spatial partitioning framework, for each sampling point in the normalized point cloud coordinate set, the difference between the coordinate value of the sampling point in the X-axis, Y-axis, and Z-axis directions and the corresponding coordinate value of the voxel grid spatial origin is calculated, and the coordinate difference in each direction is divided by the voxel unit side length parameter and the result is rounded down to obtain the grid index number of the sampling point in the three orthogonal directions; Based on the grid index numbers of each sampling point in three orthogonal directions, the voxel unit to which the sampling point belongs is determined by the grid index numbers in the three directions. The sampling point is then assigned to the voxel unit corresponding to its grid index number, thus obtaining the assignment relationship between each sampling point and the voxel unit. Based on the assignment relationship between each sampling point and the voxel unit, after traversing all sampling points in the normalized point cloud coordinate set and completing the voxel unit assignment, an index association is established between each voxel unit and all sampling points falling within that voxel unit, thus obtaining the mapping relationship data between the voxel unit and the point cloud.

[0017] In a preferred embodiment of the present invention, step 2 above may include: Step 21: Based on the initial voxel grid model, perform global statistical analysis on the attribute parameter values ​​of each voxel unit to determine the attribute classification threshold used to distinguish aggregate particle voxels from cement matrix voxels, thus obtaining the voxel classification threshold parameter. Specifically, this includes: extracting the attribute parameter values ​​of all non-empty voxel units in the initial voxel grid model to form a global sample set of attribute parameters; calculating a gray-level histogram of this global sample set; performing global statistical analysis based on the gray-level histogram using the maximum inter-class variance method; traversing all possible values ​​within the attribute parameter value range as candidate classification thresholds; for each candidate classification threshold, dividing the global sample set of attribute parameters into subsets where the attribute parameter value is greater than or equal to the candidate threshold and subsets where the attribute parameter value is less than the candidate threshold; calculating the inter-class variance of the two subsets; and selecting the candidate classification threshold that maximizes the inter-class variance as the attribute classification threshold used to distinguish aggregate particle voxels from cement matrix voxels, thus obtaining the voxel classification threshold parameter, denoted as . Wherein, the voxel classification threshold parameter It is a dimensionless scalar, with the same physical meaning as the attribute parameters, both representing the arithmetic mean of the laser reflection intensity values.

[0018] Step 22: Based on the voxel classification threshold parameter, compare the attribute parameter values ​​of each voxel unit in the initial voxel mesh model with the voxel classification threshold parameter one by one. Voxel units with attribute parameter values ​​greater than or equal to the threshold are marked as aggregate particle voxels, and voxel units with values ​​less than the threshold are marked as cement-based voxels, thus obtaining the voxel binary classification labeling result. Specifically, this includes: based on the obtained voxel classification threshold parameter... Iterate through all non-empty voxel cells in the obtained initial voxel mesh model, and compare the attribute parameter values ​​of each voxel cell with the voxel classification threshold parameter one by one. Comparison and judgment; for attribute parameter values ​​greater than or equal to The voxel unit is assigned a classification label of 1, and is labeled as an aggregate particle voxel; for attribute parameter values ​​less than The voxel unit is assigned a classification label of 0, and is labeled as a cement matrix voxel, thus obtaining the voxel binary classification label result.

[0019] Step 23: Based on the voxel binary classification label results, traverse all voxel elements in the initial voxel mesh model, and detect spatially adjacent voxel element pairs with different classification label values ​​one by one. Determine the interface between voxel element pairs with different classification label values ​​as the location of the interface between aggregate particles and cement matrix, thus obtaining a candidate interface voxel pair set. Specifically, this includes: based on the obtained voxel binary classification label results, for each non-empty voxel element in the initial voxel mesh model, detect its adjacent voxel elements in the X-axis, Y-axis, and Z-axis directions, respectively. The classification label values ​​of the 6 adjacent voxel units are determined; if the classification label of a voxel unit is different from the classification label value of one of its adjacent voxel units, then the voxel unit and the adjacent voxel unit form a pair of adjacent voxel units with different classification label values; the interface of the adjacent voxel unit pairs with different classification label values ​​is determined as the location of the interface between the aggregate particles and the cement matrix; after traversing all non-empty voxel units and completing the above adjacency detection, all detected voxel unit pairs with different classification label values ​​are summarized to obtain a set of candidate interface voxel pairs.

[0020] Step 24: Based on the candidate interface voxel pair set, extract the spatial coordinates of all sampling points on the interface of adjacent voxel units with different classification labels for each pair, and summarize and merge the spatial coordinates of sampling points on the interface of all candidate interface voxel pairs to obtain the interface point cloud coordinate data set. Specifically, this includes: based on the obtained candidate interface voxel pair set, for each pair of adjacent voxel units with different classification label values, extract the spatial coordinates of all original sampling points falling into the interface region within the spatial range involved in the interface region of the two voxel units; summarize and merge the spatial coordinates of all sampling points extracted on the interface region of all candidate interface voxel pairs. Since these sampling points are all located in the boundary region between aggregate particle voxels and cement matrix voxels, their spatial positions constitute a discrete expression of the geometric morphology of the interface between aggregate particles and cement matrix, thereby obtaining the interface point cloud coordinate data set.

[0021] In a preferred embodiment of the present invention, step 3 above may include: Step 31: Based on the interface point cloud coordinate data set, for each sampling point in the interface point cloud sequence, search for several neighboring interface sampling points within a preset spatial neighborhood radius centered on the sampling point. Fit a local quadratic surface function using the spatial coordinates of the neighboring interface sampling points to obtain the local surface fitting function at each sampling point. Specifically, this includes: based on the obtained interface point cloud coordinate data set, for each sampling point in the interface point cloud sequence... ,by Using the spatial coordinates as the center, set a preset spatial neighborhood radius. Search the point cloud coordinate data set of the interface and find the corresponding data. Spatial Euclidean distance less than or equal to All neighboring interface sampling points; if the number of neighboring interface sampling points found reaches the preset minimum fitting point requirement, then... The origin of the local coordinate system is the sampling point. The direction of the normal vector of the aggregate surface at that location is in the local coordinate system. Axis, establishing mutually orthogonal axes within the tangential plane of the aggregate surface. shaft and Axis; In the local coordinate system, the spatial coordinates of the sampling points of the adjacent interfaces are used to fit the local quadratic surface function. The fitting method is to substitute the coordinates of each sampling point of the adjacent interface in the local coordinate system into the quadratic surface equation, and solve the coefficients of the quadratic surface equation by the least squares method to obtain the local surface fitting function at each sampling point.

[0022] The local surface fitting function is expressed as: ; In the formula: Represents the coordinate values ​​along the direction of the normal vector on the aggregate surface, in millimeters; and These represent the coordinate values ​​of two mutually orthogonal directions within the tangential plane of the aggregate surface, in millimeters; , , The coefficient of the quadratic term, , The coefficient of the linear term, The coefficients are constant terms, and all of the above coefficients are obtained by the least squares method. , , The dimension is the reciprocal of the millimeter, and the coefficient is... , For dimensionless quantities, constant term The unit is millimeters.

[0023] Step 32: Based on the local surface fitting function at each sampling point, solve for the first and second partial derivatives of the local surface fitting function along two mutually orthogonal directions within the tangent plane of the aggregate surface, respectively, to obtain the values ​​of the first and second partial derivatives of the surface at each sampling point. Specifically, this includes: based on the obtained local surface fitting function at each sampling point... Two mutually orthogonal directions within the tangential plane of the aggregate surface and Solve for the first and second partial derivatives of the local surface fitting function, respectively. Then, along the local surface fitting function... Taking the first-order partial derivative along the direction, we obtain the local surface fitting function along... First-order partial derivatives in the direction ;along Taking the first-order partial derivative along the direction, we get... First-order partial derivatives in the direction ;along Taking the second-order partial derivative along the direction, we get... Second-order partial derivatives of the direction ;along Taking the second-order partial derivative along the direction, we get... Second-order partial derivatives of the direction First along direction along By taking the mixed second-order partial derivatives in the direction of the direction, we obtain the mixed second-order partial derivatives. The partial derivative values ​​mentioned above are all directly calculated from the coefficients of the obtained quadratic surface fitting function, thus obtaining the first-order and second-order partial derivative values ​​of the surface at each sampling point.

[0024] Step 33: Based on the first and second partial derivative values ​​of the surface at each sampling point, calculate the Gaussian curvature value and average curvature value at each interface sampling point location by combining the first and second partial derivative values, thus obtaining the interface point cloud curvature data. Specifically, this includes: calculating the Gaussian curvature value and average curvature value at each interface sampling point location based on the first and second partial derivative values ​​of the surface obtained in step 32. The Gaussian curvature value... The calculation relationship is obtained by combining the numerical values ​​of the second-order partial derivatives: ; The average curvature value The calculation relationship is obtained by combining the values ​​of the first-order partial derivative and the second-order partial derivative: ; in: and These are the values ​​of the first-order partial derivatives obtained, and the dimensionless quantities. , and These are the values ​​of the obtained second-order partial derivatives, which are the reciprocals of the units of millimeters. This represents the Gaussian curvature value, with the dimension of negative squared in millimeters; This represents the average curvature value, with dimensions in millimeters to the power of negative one, representing the Gaussian curvature value corresponding to each interface sampling point. and mean curvature value The point cloud curvature data of the interface is then obtained by summarizing the data.

[0025] Step 34: Based on the interface point cloud curvature data, spatially correlate and map the three-dimensional spatial coordinates of each interface sampling point with the corresponding Gaussian curvature value and average curvature value. Construct a curvature scalar field within the spatial region covered by the interface between aggregate particles and cement matrix to obtain the interface curvature distribution field. Specifically, this includes: establishing the three-dimensional spatial coordinates of each interface sampling point based on the obtained interface point cloud curvature data. Its corresponding Gaussian curvature value and mean curvature value The spatial correlation mapping relationship between them; the interface sampling points carrying curvature attribute information are used to construct a spatially continuous curvature scalar field with the curvature value at the spatial location of each sampling point as the scalar field value in the three-dimensional spatial region covered by the interface between aggregate particles and cement matrix. The curvature value at any spatial location in the curvature scalar field is determined by spatial interpolation of the curvature values ​​of its neighboring sampling points to obtain the interface curvature distribution field.

[0026] In a preferred embodiment of the present invention, step 4 above may include: Step 41: Based on the interface curvature distribution field, for each sampling point in the interface point cloud sequence, search along the direction of the tangent plane of the aggregate surface for several adjacent interface sampling points with the closest spatial distance to it, establish the local adjacency relationship between each sampling point, and obtain the interface point cloud adjacency topology map. Specifically, this includes: based on the obtained interface curvature distribution field, for each sampling point in the interface point cloud sequence, search along the direction of the tangent plane of the aggregate surface at that sampling point for the interface point cloud sequence for the nearest spatial Euclidean distance to that sampling point. There are adjacent interface sampling points, among which The preset adjacency number parameter; the searched... An undirected connection edge is established between each adjacent interface sampling point and the sampling point. The initial weight of each connection edge is the spatial Euclidean distance between the two sampling points. After performing the above adjacency search and connection operation on all sampling points in the interface point cloud sequence, an interface point cloud adjacency topology graph with interface sampling points as nodes and the connection edges between adjacent sampling points as edges is obtained.

[0027] Step 42: Based on the interface point cloud adjacency topology graph, using the curvature values ​​of each sampling point in the interface curvature distribution field as path traversal cost weighting factors, calculate the weighted distance value for each edge connecting adjacent sampling points in the interface point cloud adjacency topology graph to obtain a weighted adjacency topology graph. Specifically, this includes: based on the obtained interface point cloud adjacency topology graph and the obtained interface curvature distribution field, using the average curvature value at each sampling point in the interface curvature distribution field... As a path travel cost weighting factor, each adjacent sampling point in the interface point cloud adjacency topology graph is connected. and Calculate the weighted distance value of the edge. The calculation relationship is as follows: ; in: Indicates sampling point and The spatial Euclidean distance between them, in millimeters; and The obtained sampling points are respectively represented as follows. and The absolute value of the average curvature at that point; The preset curvature weight adjustment coefficient, with units in millimeters, is used to make the average curvature value dimensionless. A larger value indicates a greater contribution of curvature to path cost. Five sets of standard surfaces with known curvature distributions are selected, and each is weighted with different values. The values ​​(0.1, 0.5, 1.0, 2.0, 5.0) are used to calculate the geodesic distance, with the goal of minimizing the error from the theoretical geodesic distance to determine the final distance. value; This represents the weighted distance value, in millimeters.

[0028] The original weights of each edge in the interface point cloud adjacency topology graph are replaced with the corresponding weighted distance values ​​to obtain a weighted adjacency topology graph. Since areas with greater curvature on the aggregate surface usually correspond to the edges or depressions of aggregate particles, assigning higher passage cost weights to the connecting edges in areas with greater curvature can make subsequent path searches tend to select paths with gentler curvature changes, thus more accurately reflecting the real spatial path extending along the aggregate surface.

[0029] Step 43: Based on the weighted adjacency topology graph, for each pair of sequentially arranged adjacent sampling points in the interface point cloud sequence, under the condition that the search path does not detach from the aggregate surface, find the path with the smallest cumulative weighted distance connecting the two adjacent sampling points in the weighted adjacency topology graph, and obtain the shortest spatial path extending along the aggregate surface between adjacent sampling points. Specifically, this includes: based on the obtained weighted adjacency topology graph, for each pair of spatially arranged adjacent sampling points in the interface point cloud sequence, using the two adjacent sampling points as the starting point and ending point of the path search respectively, using the shortest path search algorithm to find the node sequence path with the smallest sum of cumulative weighted distances connecting the starting point and the ending point in the weighted adjacency topology graph; during the path search process, by constraining each node on the search path to belong to the interface sampling point in the interface point cloud adjacency topology graph, ensuring that the search path does not detach from the aggregate surface; and determining the path with the smallest sum of cumulative weighted distances as the shortest spatial path extending along the aggregate surface between adjacent sampling points.

[0030] Step 44: Based on the shortest spatial path extending along the aggregate surface between adjacent sampling points, record the length values ​​of the shortest spatial path corresponding to each pair of adjacent sampling points in the order of the sampling points in the interface point cloud sequence to obtain the geodesic distance sequence of the pore boundary. Specifically, this includes: based on the obtained shortest spatial path extending along the aggregate surface between adjacent sampling points, calculating the sum of the weighted distance values ​​of all connecting edges on the shortest spatial path corresponding to each pair of adjacent sampling points, and using this sum as the length of the shortest spatial path corresponding to that pair of adjacent sampling points; recording the length values ​​of the shortest spatial path of all pairs of adjacent sampling points in the order of the spatial arrangement of the sampling points in the interface point cloud sequence to form a one-dimensional ordered sequence, thus obtaining the geodesic distance sequence of the pore boundary.

[0031] In a preferred embodiment of the present invention, step 5 above may include: Step 51: Based on the geodesic distance sequence of pore boundaries, set two levels of geodesic distance abrupt change judgment thresholds. Identify the sampling point positions corresponding to different threshold levels for the changes between adjacent distance values ​​in the geodesic distance sequence as candidate pore boundaries of the corresponding type, obtaining a set of candidate pore boundaries with classification labels. Specifically, this includes: based on the obtained geodesic distance sequence of pore boundaries, setting a first-level abrupt change judgment threshold and a second-level abrupt change judgment threshold. Both thresholds are dimensionless ratio parameters. The preferred value for the first-level abrupt change judgment threshold is 0.15, corresponding to the abrupt change characteristics of micro-gel pore boundaries. The preferred value for the second-level abrupt change judgment threshold is 0.35, corresponding to the abrupt change characteristics of harmful pore boundaries. Harmful pores include three types of typical harmful pores in hydraulic concrete: capillary pores, freeze-thaw expansion macropores, and water seepage and erosion pores. Starting from the second distance value in the geodesic distance sequence, calculate the change between each distance value and its previous distance value in the sequence. The calculation relationship for the change is as follows: ; in: It represents the relative change of the i-th distance value with respect to the previous distance value, and is a dimensionless ratio. This represents the i-th distance value, in millimeters. Let represent the (i-1)th distance value, in millimeters. Each change is compared to two levels of mutation detection thresholds: if the change is greater than the second-level threshold, the location between the i-th and (i-1)-th sampling points is identified as having a harmful pore boundary geodesic distance mutation, and these two sampling points are marked as harmful pore boundary candidate points; if the change is greater than the first-level threshold and less than or equal to the second-level threshold, this location is identified as having a gel pore boundary geodesic distance mutation, and these two sampling points are marked as gel pore boundary candidate points; if the change is less than or equal to the first-level threshold, it is determined that there is no pore boundary mutation at this location. After traversing all distance values ​​in the geodesic distance sequence and completing the above mutation detection, all marked pore boundary candidate points are summarized to obtain a set of pore boundary candidate points labeled with pore type.

[0032] Step 52: Based on the set of candidate points for pore boundaries with classification labels, perform connectivity clustering on spatially adjacent and continuously distributed candidate points, extract a subset of candidate points forming a spatially closed contour as the boundary point cloud of a single pore, and simultaneously label the type attribute and connectivity attribute of each pore to obtain a subset of attributed boundary point clouds corresponding to each independent pore. Specifically, this includes: performing spatial connectivity clustering on the three-dimensional spatial coordinates of each candidate point in the obtained set of candidate points for pore boundaries with classification labels; specifically, any two candidate points in the set of candidate points for pore boundaries that meet a preset spatial proximity threshold are considered to be connected to each other, and the set of candidate points for pore boundaries is divided into several sub-sets based on spatial connectivity. The clustering process involves selecting a subset of candidate points that are not connected to each other. During the clustering process, the original pore type label of each candidate point is preserved simultaneously. For each candidate point subset, it is checked whether the candidate points within it form a closed contour in space. If a closed contour is formed, the candidate point subset is extracted as the boundary point cloud of a single independent pore. The pore type with the highest proportion in the subset is labeled as the type attribute of the corresponding pore, and the independent or connected attribute of the pore is also labeled. If a closed contour is not formed, it is regarded as an incomplete boundary and discarded or merged with the adjacent candidate point subset and then judged. The boundary point clouds of all extracted single independent pores are recorded separately to obtain the attributed boundary point cloud subset corresponding to each independent pore. Each boundary point cloud subset corresponds to an independent pore entity inside the hydraulic concrete.

[0033] Step 53: Based on the attributed boundary point cloud subsets corresponding to each independent pore, construct a spatial convex hull as the envelope geometry of the pore for each pore's boundary point cloud subset. Use the internal spatial region of the envelope geometry as the volume integration domain of the pore, and simultaneously retain the pore type attribute label to obtain the classification integration calculation region corresponding to each pore. Specifically, this includes: based on the obtained attributed boundary point cloud subsets corresponding to each independent pore, for each pore's boundary point cloud subset, selecting the sampling points with the smallest X-axis coordinate value, the largest X-axis coordinate value, the smallest Y-axis coordinate value, the largest Y-axis coordinate value, the smallest Z-axis coordinate value, and the largest Z-axis coordinate value as initial convex hull vertices. Construct an initial spatial convex hull polyhedron from the initial convex hull vertices through spatial triangulation to obtain the initial convex hull geometry; based on the initial convex hull geometry, traverse the remaining points in the pore boundary point cloud subset located outside the initial convex hull geometry one by one. For sampling points, an iterative expansion strategy adapted to the narrow and irregular contours is adopted for pores marked as dissolution cavities. For other types of pores, a standard farthest point iterative strategy is adopted. Each time, an external sampling point that meets the requirements of the corresponding strategy is selected as a new vertex. The current convex hull geometry is expanded with triangular facets and reconnected with vertex topology using the new vertex. The boundary range of the convex hull geometry is updated. The above traversal and expansion operations are repeated until all sampling points in the pore boundary point cloud subset are located inside or on the surface of the convex hull geometry, thus obtaining a complete spatial convex hull polyhedron. Based on the complete spatial convex hull polyhedron, the internal closed space region enclosed by all the triangular facets of the complete spatial convex hull polyhedron is used as the three-dimensional volume integral calculation region of the pore. The type attribute and connectivity attribute of the pore are bound simultaneously. The above construction process is performed on all the obtained independent pores and then summarized to obtain the classification integral calculation region corresponding to each pore.

[0034] Step 54: Based on the categorized integral calculation region corresponding to each pore, perform a three-dimensional spatial volume integral operation within the integral calculation region of each pore to obtain the integral operation result; use the integral operation result as the true volume value of the corresponding pore, classify and summarize the volume data according to pore type, and calculate the hydraulic engineering-specific pore evaluation index to obtain the set of pore parameters for hydraulic concrete. Specifically, this includes: based on the obtained categorized integral calculation region corresponding to each pore, performing a three-dimensional spatial volume integral operation within the three-dimensional volume integral calculation region of each pore, with the calculation relationship as follows: ; in: Indicates the first k The three-dimensional volume integral calculation region enclosed by the spatial convex hull polyhedron corresponding to each pore; dV Represents a volume element; Indicates the first kThe true volume value of each pore is expressed in cubic millimeters. Since the spatial convex hull polyhedron is formed by triangular facets, the volume integral operation decomposes the convex hull polyhedron into several spatial tetrahedrons with triangular facets as bases and the origin as vertices during actual execution. The directed volume of each tetrahedron is calculated and summed to obtain the true volume value of the pore. The true volume values ​​corresponding to all independent pores are classified and summarized by type to obtain the total volume of gel pores, total volume of capillary pores, total volume of freeze-thaw macropores, and total volume of dissolution pores. Further calculations are made to obtain the total pore volume, total porosity, proportion of harmful pores, and total volume of connected pores, which are core evaluation indicators for hydraulic engineering. All indicators match the testing requirements of the hydraulic concrete test procedure. All parameters are summarized to obtain the set of pore parameters for hydraulic concrete.

[0035] In a preferred embodiment of the present invention, step 53 above may include: Step 531: Based on the attributed boundary point cloud subsets corresponding to each independent pore, for each pore's boundary point cloud subset, select the sampling points with the smallest and largest spatial X-coordinates, smallest and largest Y-coordinates, smallest and largest Z-coordinates, and largest Z-coordinates respectively as initial convex hull vertices. Construct an initial spatial convex hull polyhedron from the initial convex hull vertices through spatial triangulation to obtain the initial convex hull geometry. Specifically, this includes: based on the obtained attributed boundary point cloud subsets corresponding to each independent pore, simultaneously retaining the class of the pore for each pore's boundary point cloud subset. Type attribute labeling: Among all the sampling points in this subset, find the sampling point with the smallest X-axis coordinate value, the sampling point with the largest X-axis coordinate value, the sampling point with the smallest Y-axis coordinate value, the sampling point with the largest Y-axis coordinate value, the sampling point with the smallest Z-axis coordinate value, and the sampling point with the largest Z-axis coordinate value. Take these 6 extreme sampling points as the initial convex hull vertices. Using the initial convex hull vertices as nodes, connect the nearest initial convex hull vertices to form triangular patches through spatial triangulation, construct an initial spatial convex hull polyhedron that surrounds the initial convex hull vertices, and obtain the initial convex hull geometry.

[0036] Step 532: Based on the initial convex hull geometry, iterate through the remaining sampling points in the pore boundary point cloud subset located outside the initial convex hull geometry. For different types of pores, adopt corresponding iterative expansion strategies. Each time, select an external sampling point that meets the strategy requirements. Use this sampling point as a new vertex to expand the triangular facets and reconnect the vertex topology of the current convex hull geometry, and update the boundary range of the convex hull geometry until all sampling points in the pore boundary point cloud subset are located inside the convex hull geometry, thus obtaining a complete spatial convex hull polyhedron. Specifically, this includes: based on the obtained initial convex hull geometry, iterate through the remaining sampling points in the current pore boundary point cloud subset located outside the initial convex hull geometry. If the current pore type is a dissolution cavity, a narrow contour adaptation iterative strategy is adopted. For each remaining sampling point located outside the convex hull geometry, calculate the spatial perpendicular distance between the sampling point and each surface triangular facet of the current convex hull geometry, and simultaneously superimpose the weighting coefficient of the long axis extension direction of the cavity to obtain a weighted distance. The maximum weighted distance value is selected from all external sampling points, and the sampling point corresponding to the maximum weighted distance value is selected as the new vertex for this iteration. If the current aperture is of other types, the standard farthest point iteration strategy is adopted. For each remaining sampling point located outside the convex hull geometry, the spatial vertical distance between the sampling point and each surface triangle of the current convex hull geometry is calculated, and the maximum spatial vertical distance value is selected from all external sampling points as the new vertex for this iteration. A new triangle is constructed between the new vertex and the visible triangle on the current convex hull geometry, and the topological connection between the new vertex and the original triangle occluded by the new triangle is reconnected so that the expanded convex hull geometry includes the new vertex inside it. The above traversal and expansion operations are repeated until all sampling points in the boundary point cloud subset of the current aperture are located inside or on the surface of the convex hull geometry, thus obtaining a complete spatial convex hull polyhedron.

[0037] Step 533: Based on the complete spatial convex hull polyhedron, the internal closed space region enclosed by all the triangular facets of the complete spatial convex hull polyhedron is used as the three-dimensional volume integral calculation region for the pore. The type attribute and connectivity attribute of the pore are simultaneously bound to obtain the classification integral calculation region corresponding to each pore. Specifically, this includes: based on the obtained complete spatial convex hull polyhedron, the internal closed space region enclosed by all the triangular facets of the complete spatial convex hull polyhedron is used as the three-dimensional volume integral calculation region for the pore, and the type attribute label and connectivity attribute label corresponding to the pore are simultaneously bound to it; for the boundary point cloud subsets corresponding to each independent pore, after constructing their respective complete spatial convex hull polyhedrons, the three-dimensional volume integral calculation region for each pore is determined according to this step, and the classification integral calculation regions corresponding to each pore are summarized.

[0038] In a preferred embodiment of the present invention, step 6 above may include: Step 61: Based on the true pore volume parameters, the true volume value of each pore is used as the geometric attribute parameter of the corresponding pore. A pore geometric marker corresponding to the true volume value is generated at the spatial location of the integral calculation region of each pore. A mapping relationship is established between the spatial coordinates of each pore geometric marker and the volume attribute parameter to obtain pore volume geometric marker data. Specifically, this includes: based on the obtained true pore volume parameters, the true volume value of each pore... As the geometric property parameter of the corresponding pore; for the first For each pore, at the spatial location of the integral calculation region corresponding to that pore, a pore geometric marker is generated that matches the outer contour of the integral calculation region and is filled with the volume attribute information of the pore. A mapping relationship is established between the spatial coordinates of each pore geometric marker and its volume attribute parameters, where the spatial position of each pore geometric marker is represented by its spatial coordinate range, and the corresponding... The value represents its volume attribute parameter, and the pore volume geometric mark data is obtained.

[0039] Step 62: Based on the interface curvature distribution field, along the interface region between aggregate particles and cement matrix, use the curve formed by connecting the sampling points with equal curvature values ​​in the interface curvature distribution field as the constraint edge for mesh generation. Generate a surface triangular mesh within the interface region using a surface triangular meshing method, and attach the curvature value at the spatial location of each mesh vertex as a vertex scalar attribute value to the corresponding mesh vertex to obtain the interface curvature attribute mesh. Specifically, this includes: based on the obtained interface curvature distribution field, extracting sampling points with equal Gaussian curvature values ​​or average curvature values ​​along the interface region between aggregate particles and cement matrix in the interface curvature distribution field; connecting these sampling points with equal curvature values ​​sequentially along the tangent plane of the aggregate surface to form a spatial curve; and using the spatial curve as the constraint edge for mesh generation. Constraint edges are formed; within the interface region between two adjacent constraint edges, a curved triangular mesh generation method is adopted, using sampling points on the constraint edges as mesh boundary nodes and interface sampling points inside the interface region as internal mesh nodes to generate a curved triangular mesh covering the interface region; each mesh vertex in the curved triangular mesh corresponds to the three-dimensional spatial coordinates of an interface sampling point, and each mesh vertex is attached with the curvature value corresponding to that spatial position as the vertex scalar attribute value; all curved triangular meshes at the interface between aggregate particles and cement matrix are summarized to obtain the interface curvature attribute mesh. The curved triangular mesh generation method, when generating triangular patches, uses the maximization of the minimum interior angle of the triangle as the criterion to avoid generating elongated triangles and ensure the shape quality of the mesh units.

[0040] Step 63: Based on the pore volume geometric marker data and the interface curvature attribute mesh, embed each pore geometric marker into the corresponding spatial region of the interface curvature attribute mesh according to the spatial position coordinates recorded in the pore volume geometric marker data. Perform mesh node fusion processing and mesh patch re-subdivision processing on the boundary region between the pore geometric marker surface mesh and the interface curvature attribute mesh to obtain a three-dimensional reconstructed mesh model. Specifically, this includes: based on the obtained pore volume geometric marker data and the obtained interface curvature attribute mesh, embed each pore geometric marker into the spatial region of the interface curvature attribute mesh corresponding to the spatial coordinate range recorded in the pore volume geometric marker data in three-dimensional space; perform mesh node fusion processing and mesh patch re-subdivision processing on the boundary region between the pore geometric marker surface mesh and the interface curvature attribute mesh to obtain a three-dimensional reconstructed mesh model. In the boundary region where the mesh and the interface curvature attribute mesh spatially overlap, mesh nodes belonging to the pore marker mesh and the interface curvature attribute mesh on both sides of the interface are identified. Pairs of mesh nodes with a spatial distance less than a preset tolerance threshold are merged into shared mesh nodes, and the corresponding nodes in the two original meshes are replaced with the merged shared mesh nodes. At the same time, for the mesh patches in the boundary region whose shape has changed due to node merging, the mesh patches are re-subdivided, and the topological connection relationship of the triangular patches in this region is regenerated according to the distribution of the merged nodes. After node fusion and patch re-subdivision, the interface curvature attribute mesh and the pore geometric marker mesh have shared mesh nodes and continuous mesh patch topological connection relationships at the boundary, resulting in a three-dimensional reconstructed mesh model.

[0041] In a preferred embodiment of the present invention, step 7 above may include: Step 71: Based on the three-dimensional reconstructed mesh model, select the corresponding smoothing parameters according to the component type of the hydraulic concrete core sample, and perform mesh smoothing processing on the spatial coordinates of all mesh vertices in the three-dimensional reconstructed mesh model to obtain a smoothed three-dimensional mesh model. Specifically, this includes: performing mesh smoothing processing on the spatial coordinates of all mesh vertices in the obtained three-dimensional reconstructed mesh model. Specifically, for each grid vertex in the three-dimensional reconstructed mesh model, all adjacent grid vertices directly connected to that vertex through grid edges are searched to form a spatial neighborhood vertex set. The smoothing relaxation coefficient is selected according to the component type of the hydraulic concrete core sample. For the overflow surface anti-erosion region core sample, a low value of 0.2 to 0.3 is selected to preserve the micro-uneven features of the surface erosion pit. For the seepage prevention structure layer and the normal concrete core sample inside the pump station, a high value of 0.6 to 0.8 is selected to fully smooth the discrete noise and ensure the accuracy of subsequent seepage calculations. For the roller-compacted concrete core sample, a locally low value is used for the layer interface region to preserve the geometric features of the interlayer interface. The spatial coordinate value of the vertex is adjusted to the weighted average of the original spatial coordinate of the vertex and the spatial coordinates of all adjacent grid vertices in its spatial neighborhood vertex set. The smoothing calculation relationship is as follows: ; in: Indicates the first step before smoothing. i The spatial coordinate vectors of each grid vertex are in millimeters. Indicates the first step after smoothing. i The spatial coordinate vectors of each grid vertex are in millimeters. This represents the set of all adjacent grid vertices that are directly connected to a vertex via grid edges; This indicates the number of adjacent grid vertices contained in the set; The selected smoothing relaxation coefficient is a dimensionless positive number between 0 and 1. After performing the above smoothing process on all mesh vertices in the three-dimensional reconstructed mesh model, the local jagged undulations and geometric distortions on the mesh surface caused by laser scanning sampling noise and discretization processing are eliminated, while the microscopic geometric features of the key hydraulic analysis area are preserved, resulting in a smoothed three-dimensional mesh model.

[0042] Step 72: Based on the smoothed 3D mesh model, and according to the curvature attribute values ​​of each mesh vertex in the interface curvature attribute mesh, the volume attribute parameters of each region in the pore volume geometric marker data, and the compaction layer interface feature data, the material type to which each mesh patch in the smoothed 3D mesh model belongs is distinguished and determined, resulting in a 3D mesh model with material attribute markings, specifically including: Based on the obtained smoothed 3D mesh model, for each mesh patch in the smoothed 3D mesh model, all the mesh vertices constituting that mesh patch are determined. The curvature attribute values ​​attached to each mesh vertex of that mesh patch are read from the interface curvature attribute mesh. The arithmetic mean of the curvature attribute values ​​of all mesh vertices of that mesh patch is calculated to obtain the curvature characteristic value of each mesh patch. Based on the curvature characteristic value of each mesh patch, the curvature characteristic value of each mesh patch is compared with a preset interface curvature judgment threshold one by one. Mesh patches with curvature characteristic values ​​greater than or equal to the interface curvature judgment threshold are marked as candidate patches for the interface transition zone, and mesh patches with curvature characteristic values ​​less than the interface curvature judgment threshold are marked as candidate patches for the interface transition zone. The grid patches are marked as candidate patches for non-interface regions, and the curvature assignment characteristics of the grid patches are obtained. Based on the obtained curvature assignment characteristics of the grid patches and the pore volume geometric marker data, for each grid patch, it is detected whether the spatial position of the grid patch falls within the spatial region occupied by the pore geometric marker body recorded in the pore volume geometric marker data. For grid patches that fall within the spatial region of the pore geometric marker body, the volume attribute parameters of the region corresponding to the pore geometric marker body are read, and the grid patch is marked as a candidate patch for pore defect region. For grid patches that do not fall within the spatial region of any pore geometric marker body, their original curvature assignment characteristics are retained, and the pore assignment characteristics of the grid patches are obtained.

[0043] Based on the obtained pore location characteristics and compaction interface characteristics of the grid patches, for each grid patch, it is determined whether the spatial position of the grid patch falls within the spatial region of the compaction interface as indicated by the compaction interface characteristic data. For grid patches falling within the compaction interface region, these patches are marked as candidate patches for weak areas of the compaction interface. For grid patches not falling within the compaction interface region, their existing location characteristics are retained to obtain the grid patch layer interface location characteristics. Based on these obtained grid patch layer interface location characteristics, the curvature location characteristics, pore location characteristics, and layer interface location characteristics of each grid patch are combined. The material type of each mesh patch is distinguished and determined. Mesh patches marked as candidate areas of pore defects are determined to be materials with pore defects. Mesh patches marked as candidate areas of weak areas of compacted layer interface are determined to be materials with weak areas of compacted layer interface. Mesh patches marked as candidate areas of interface transition zone but not determined to be materials with pore defects or materials with weak areas of compacted layer interface are determined to be materials of transition between aggregate and cement matrix. The remaining mesh patches are determined to be materials of concrete matrix. The material type determination result of each mesh patch is used as the material property label of the mesh patch and attached to the corresponding mesh patch to obtain a three-dimensional mesh model with material property labels.

[0044] Step 73: Based on the 3D mesh model with material property labels, integrate the set of spatial coordinates of mesh vertices, the set of topological connections of mesh patches, and the set of material property labels of mesh patches into a structured 3D reconstruction data file adapted to the format of hydraulic engineering simulation software. Simultaneously output a structured report for hydraulic engineering inspection. The structured 3D reconstruction data file and the accompanying report together constitute a 3D reconstruction model of hydraulic concrete sampling data, specifically including: Based on the obtained 3D mesh model with material property labels, traverse all mesh vertices in the 3D mesh model with material property labels, extract the 3D spatial coordinates of each mesh vertex, and record the 3D spatial coordinates of all mesh vertices in order of mesh vertex number to obtain a set of mesh vertex spatial coordinate data; based on the obtained set of mesh vertex spatial coordinate data, traverse all mesh patches in the 3D mesh model with material property labels, extract the mesh vertex numbers constituting each mesh patch and their connection relationships, and record the corresponding connection relationships between each mesh patch and the mesh vertex numbers it contains in order to obtain a set of mesh patch topology connection relationship data; based on the obtained mesh... The topological connection relationship data set of the grid patches is obtained by traversing all grid patches in the 3D mesh model with material property labels, extracting the material property labels of each grid patch, and recording the correspondence between each grid patch number and its material property label in sequence. Based on the obtained grid vertex spatial coordinate data set, grid patch topological connection relationship data set, and grid patch material property label data set, the three data sets are integrated and packaged into a structured data file containing node coordinate information, element topology information, and material partition information according to the data organization format required by the hydraulic simulation software, thus obtaining a structured 3D reconstruction data file.

[0045] Based on the obtained structured 3D reconstruction data file, the number and volume of pores corresponding to pore defects in the material are statistically analyzed according to pore type. The interface curvature distribution characteristics of the transition material between aggregate and cement matrix are statistically analyzed. The distribution range of the layered interface corresponding to the weak material of the compacted layered interface is statistically analyzed. A hydraulic engineering inspection structured report containing pore parameters, interface curvature parameters, and layered interface parameters is generated, resulting in a hydraulic engineering inspection structured report. Based on the obtained structured 3D reconstruction data file and hydraulic engineering inspection structured report, a refined 3D model that can be directly imported into hydraulic engineering simulation software is provided by the structured 3D reconstruction data file. The hydraulic engineering inspection structured report provides quantitative indicators for concrete durability evaluation. The structured 3D reconstruction data file and the accompanying hydraulic engineering inspection structured report together constitute a 3D reconstruction model of hydraulic concrete sampling data.

[0046] like Figure 2 As shown, embodiments of the present invention also provide a three-dimensional reconstruction and analysis system for hydraulic concrete sampling data, including: The point cloud acquisition and voxelization module is used to acquire the three-dimensional laser scanning point cloud coordinate set of the surface of the hydraulic concrete core sample, and to perform uniform voxel mesh division processing on the coordinate set along three orthogonal directions in space to obtain the initial voxel mesh model. The interface point cloud extraction module is used to extract the interface point cloud sequence at the interface between aggregate particles and cement matrix based on the initial voxel mesh model, and obtain the interface point cloud coordinate data set. The interface curvature calculation module is used to solve the partial derivatives of the coordinates of each sampling point in the interface point cloud sequence based on the interface point cloud coordinate data set, so as to obtain the interface curvature distribution field. The geodesic distance calculation module is used to calculate the shortest spatial path length of adjacent sampling points along the aggregate surface in the interface point cloud sequence based on the interface curvature distribution field, and obtain the geodesic distance sequence of the pore boundary. The pore volume calculation module is used to identify candidate boundary points of different types of pores based on the geodesic distance sequence of pore boundaries through hierarchical mutation judgment, and extract an independent pore boundary point cloud set with type and connectivity attributes through connected clustering; construct an appropriate spatial convex hull volume integration domain for different types of pores, perform three-dimensional volume integration operation and classify and statistically analyze pore volumes to obtain a set of pore parameters including hydraulic engineering evaluation indicators; The 3D reconstruction and fusion module is used to fuse the interface curvature distribution field based on the actual pore volume parameters to obtain a 3D reconstructed mesh model that includes the interface geometry and internal pore volume characteristics. The model output module is used to obtain a 3D reconstruction model of concrete sampling data based on the 3D reconstructed mesh model.

[0047] It should be noted that this system is a system corresponding to the above method. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0048] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A three-dimensional reconstruction analysis method for hydraulic concrete sampling data, characterized in that, The method includes: Step 1: Obtain the coordinate set of the three-dimensional laser scanning point cloud on the surface of the hydraulic concrete core sample. Perform uniform voxel meshing on the coordinate set along three orthogonal directions in space to obtain the initial voxel mesh model. Step 2: Based on the initial voxel mesh model, extract the interface point cloud sequence at the interface between aggregate particles and cement matrix to obtain the interface point cloud coordinate data set. Step 3: Based on the interface point cloud coordinate data set, solve the partial derivatives of the coordinates of each sampling point in the interface point cloud sequence to obtain the interface curvature distribution field. Step 4: Based on the interface curvature distribution field, calculate the shortest spatial path length of adjacent sampling points along the aggregate surface in the interface point cloud sequence to obtain the geodesic distance sequence of the pore boundary, including: Based on the interface curvature distribution field, for each sampling point in the interface point cloud sequence, search along the direction of the tangent plane of the aggregate surface for several adjacent interface sampling points that are closest to it in space, establish the local adjacency relationship between each sampling point, and obtain the interface point cloud adjacency topology map. Based on the interface point cloud adjacency topology graph, the curvature value of each sampling point in the interface curvature distribution field is used as the path passage cost weight factor. The weighted distance value of each edge connecting adjacent sampling points in the interface point cloud adjacency topology graph is calculated to obtain the weighted adjacency topology graph. Based on the weighted adjacency topology graph, for each pair of sequentially arranged adjacent sampling points in the interface point cloud sequence, under the condition that the search path does not leave the aggregate surface, the path with the minimum cumulative weighted distance connecting the two adjacent sampling points in the weighted adjacency topology graph is found, and the shortest spatial path extending along the aggregate surface between adjacent sampling points is obtained. Based on the shortest spatial path extending along the aggregate surface between adjacent sampling points, the length values ​​of the shortest spatial path corresponding to each pair of adjacent sampling points are recorded sequentially according to the arrangement order of the sampling points in the interface point cloud sequence to obtain the geodesic distance sequence of the pore boundary. Step 5: Based on the geodesic distance sequence of pore boundaries, identify the candidate boundary points of different types of pores through hierarchical mutation judgment, and extract the set of independent pore boundary point clouds with type and connectivity attributes through connected clustering; construct an appropriate spatial convex hull volume integral domain for different types of pores, perform three-dimensional volume integral operation and classify and statistically analyze pore volume to obtain a set of pore parameters containing hydraulic engineering evaluation indicators. Step 6: Based on the actual pore volume parameters, fuse the interface curvature distribution field to obtain a three-dimensional reconstructed mesh model that includes the interface geometry and internal pore volume characteristics. Step 7: Based on the three-dimensional reconstruction mesh model, obtain the three-dimensional reconstruction model of the concrete sampling data.

2. The method for three-dimensional reconstruction analysis of hydraulic concrete sampling data according to claim 1, characterized in that, Step 1 includes: Based on the three-dimensional laser scanning point cloud coordinate set of the obtained hydraulic concrete core sample surface, statistical outlier filtering and spatial registration and alignment of multi-station scanning data are performed on each sampling point in the point cloud coordinate set to obtain a normalized point cloud coordinate set. Based on the normalized point cloud coordinate set, the difference between the maximum and minimum coordinate values ​​of the point cloud in the X-axis, Y-axis and Z-axis directions is calculated to obtain the spatial span range of the point cloud in the three orthogonal directions. The spatial span range of the point cloud in each direction is divided by the preset voxel element side length value of the corresponding hydraulic component and rounded up to obtain the number of voxel meshes in each direction, thereby determining the voxel meshing parameters. Based on the voxel grid division parameters, each sampling point in the normalized point cloud coordinate set is assigned a voxel unit according to the grid index number where the sampling point's spatial coordinates are located, along the three orthogonal directions of the spatial X-axis, Y-axis and Z-axis, to obtain the mapping relationship data between the voxel unit and the point cloud; Based on the mapping relationship data between voxel units and point clouds, the mean values ​​of X, Y, and Z coordinates of all sampling points contained in each voxel unit are calculated. The mean values ​​of the three coordinates are used as the spatial position coordinates of the corresponding voxel unit. The arithmetic mean of the laser reflection intensity of all sampling points in each voxel unit is calculated as the attribute parameter of the voxel unit, thus obtaining the initial voxel mesh model.

3. The three-dimensional reconstruction analysis method for hydraulic concrete sampling data according to claim 2, characterized in that, Based on the initial voxel mesh model, the interface point cloud sequence at the junction of aggregate particles and cement matrix is ​​extracted to obtain the interface point cloud coordinate data set, including: Based on the initial voxel grid model, a global statistical analysis is performed on the attribute parameter values ​​of each voxel unit to determine the attribute classification threshold used to distinguish aggregate particle voxels from cement matrix voxels, thus obtaining the voxel classification threshold parameter. Based on the voxel classification threshold parameter, the attribute parameter values ​​of each voxel unit in the initial voxel mesh model are compared with the voxel classification threshold parameter one by one. Voxel units with attribute parameter values ​​greater than or equal to the threshold are marked as aggregate particle voxels, and voxel units with attribute parameter values ​​less than the threshold are marked as cement matrix voxels, thus obtaining the voxel binary classification labeling result. Based on the results of the binary classification of voxels, all voxel elements in the initial voxel mesh model are traversed, and pairs of voxel elements that are spatially adjacent and have different classification values ​​are detected one by one. The interface of the voxel element pairs with different classification values ​​is determined as the location of the interface between the aggregate particles and the cement matrix, and a set of candidate interface voxel pairs is obtained. Based on the candidate interface voxel pair set, extract all sampling point spatial coordinates on the interface of adjacent voxel units with different classification labels, and summarize and merge the sampling point spatial coordinates on the interface of all candidate interface voxel pairs to obtain the interface point cloud coordinate data set.

4. The three-dimensional reconstruction analysis method for hydraulic concrete sampling data according to claim 3, characterized in that, Based on the interface point cloud coordinate data set, partial derivatives are calculated for the coordinates of each sampling point in the interface point cloud sequence to obtain the interface curvature distribution field, including: Based on the interface point cloud coordinate data set, for each sampling point in the interface point cloud sequence, several neighboring interface sampling points are searched within a preset spatial neighborhood radius with the sampling point as the center. The local quadratic surface function is fitted by the spatial coordinates of the neighboring interface sampling points to obtain the local surface fitting function at each sampling point. Based on the local surface fitting function at each sampling point, the first and second partial derivatives of the local surface fitting function are solved along two mutually orthogonal directions in the tangent plane of the aggregate surface, respectively, to obtain the values ​​of the first and second partial derivatives of the surface at each sampling point. Based on the first and second partial derivative values ​​of the surface at each sampling point, the Gaussian curvature value and average curvature value at each interface sampling point are calculated by combining the first and second partial derivative values ​​to obtain the interface point cloud curvature data. Based on the curvature data of the interface point cloud, the three-dimensional spatial coordinates of each interface sampling point are spatially correlated and mapped with the corresponding Gaussian curvature value and average curvature value. A curvature scalar field is constructed in the spatial region covered by the interface between aggregate particles and cement matrix to obtain the interface curvature distribution field.

5. The three-dimensional reconstruction analysis method for hydraulic concrete sampling data according to claim 4, characterized in that, Step 5 includes: Based on the geodesic distance sequence of pore boundaries, two levels of geodesic distance abrupt change judgment thresholds are set. The sampling point positions corresponding to different threshold levels of the change between adjacent distance values ​​in the geodesic distance sequence are identified as pore boundary candidate points of the corresponding type, resulting in a set of pore boundary candidate points with classification labels. Based on the set of candidate points for pore boundaries with classification labels, the candidate points that are spatially adjacent and continuously distributed are subjected to connectivity clustering. The subset of candidate points that form a spatial closed contour is extracted as the boundary point cloud of a single pore. The type attribute and connectivity attribute of each pore are labeled simultaneously to obtain the attributed boundary point cloud subset corresponding to each independent pore. Based on the attributed boundary point cloud subset corresponding to each independent pore, a spatial convex hull is constructed for the boundary point cloud subset of each pore as the envelope geometry of the pore. The spatial region inside the envelope geometry is used as the volume integration domain of the pore. The type attribute label of the pore is retained simultaneously to obtain the calculation region with classification integration corresponding to each pore. Based on the categorized integral calculation region corresponding to each pore, a three-dimensional spatial volume integral operation is performed within the integral calculation region of each pore to obtain the integral operation result; the integral operation result is used as the true volume value of the corresponding pore, the volume data is summarized according to the pore type, and the hydraulic engineering pore evaluation index is calculated to obtain the set of pore parameters for hydraulic concrete.

6. The three-dimensional reconstruction analysis method for hydraulic concrete sampling data according to claim 5, characterized in that, Based on the categorized integral calculation region corresponding to each pore, a three-dimensional spatial volume integral operation is performed within the integral calculation region of each pore to obtain the integral operation result; the integral operation result is used as the true volume value of the corresponding pore, and the volume data is summarized according to pore type and hydraulic engineering-specific pore evaluation index is calculated to obtain a set of pore parameters for hydraulic concrete, including: Based on the attributed boundary point cloud subset corresponding to each independent pore, for each pore's boundary point cloud subset, the sampling points with the smallest spatial X coordinate, the largest X coordinate, the smallest Y coordinate, the largest Y coordinate, the smallest Z coordinate, and the largest Z coordinate are selected from the boundary point cloud subset as the initial convex hull vertices. The initial spatial convex hull polyhedron is constructed from the initial convex hull vertices through spatial triangulation to obtain the initial convex hull geometry. Based on the initial convex hull geometry, the remaining sampling points located outside the initial convex hull geometry in the subset of pore boundary point clouds are traversed one by one. For different types of pores, corresponding iterative expansion strategies are adopted. Each time, an external sampling point that meets the strategy requirements is selected. The current convex hull geometry is expanded with triangular facets and reconnected with vertex topology using the sampling point as a new vertex. The boundary range of the convex hull geometry is updated until all sampling points in the subset of pore boundary point clouds are located inside the convex hull geometry, thus obtaining a complete spatial convex hull polyhedron. Based on the complete spatial convex hull polyhedron, the internal closed space region enclosed by all the triangular facets of the complete spatial convex hull polyhedron is used as the three-dimensional volume integral calculation region of the pore. The type attribute and connectivity attribute of the pore are bound simultaneously to obtain the integral calculation region with classification for each pore.

7. The three-dimensional reconstruction analysis method for hydraulic concrete sampling data according to claim 6, characterized in that, Based on the actual pore volume parameters and by fusing the interface curvature distribution field, a three-dimensional reconstructed mesh model is obtained, which includes the interface geometry and internal pore volume characteristics, including: Based on the true volume parameters of the pores, the true volume value of each pore is used as the geometric attribute parameter of the corresponding pore. A pore geometric marker body corresponding to the true volume value is generated at the spatial location of the integral calculation region of each pore. The corresponding mapping relationship between the spatial position coordinates of each pore geometric marker body and the volume attribute parameter is established to obtain the pore volume geometric marker data. Based on the interface curvature distribution field, along the interface region between aggregate particles and cement matrix, the curve connecting the sampling points of equal curvature values ​​in the interface curvature distribution field is used as the constraint edge for mesh generation. In the interface region, a surface triangular mesh is generated using the surface triangular mesh generation method, and the curvature value at the spatial position of each mesh vertex is attached to the corresponding mesh vertex as the vertex scalar attribute value to obtain the interface curvature attribute mesh. Based on the pore volume geometric marker data and the interface curvature attribute mesh, each pore geometric marker is embedded into the corresponding spatial region in the interface curvature attribute mesh according to the spatial position coordinates recorded in the pore volume geometric marker data. Mesh node fusion processing and mesh patch re-subdivision processing are performed on the boundary region between the surface mesh of the pore geometric marker and the interface curvature attribute mesh to obtain a three-dimensional reconstructed mesh model.

8. The method for three-dimensional reconstruction analysis of hydraulic concrete sampling data according to claim 7, characterized in that, Based on the 3D reconstructed mesh model, a 3D reconstruction model of the concrete sampling data is obtained, including: Based on the 3D reconstructed mesh model, and combined with the component type of the hydraulic concrete core sample, the corresponding smoothing parameters are selected. Mesh smoothing is then performed on the spatial coordinates of all mesh vertices in the 3D reconstructed mesh model to obtain a smoothed 3D mesh model. Based on the smooth 3D mesh model, the material type of each mesh patch in the smooth 3D mesh model is distinguished and determined according to the curvature attribute value of each mesh vertex in the interface curvature attribute mesh, the volume attribute parameters of each region in the pore volume geometric mark data, and the compaction layer interface feature data, so as to obtain the 3D mesh model with material attribute marking. Based on the three-dimensional mesh model with material property labels, the set of spatial coordinate data of mesh vertices, the set of topological connection data of mesh patches, and the set of material property label data of mesh patches in the three-dimensional mesh model with material property labels are integrated into a structured three-dimensional reconstruction data file adapted to the format of hydraulic engineering simulation software. Simultaneously, a structured report of hydraulic engineering inspection is output. The structured three-dimensional reconstruction data file and the supporting report together constitute a three-dimensional reconstruction model of hydraulic concrete sampling data.

9. The method for three-dimensional reconstruction analysis of hydraulic concrete sampling data according to claim 8, characterized in that, Based on the smoothed 3D mesh model, and according to the curvature attribute values ​​of each mesh vertex in the interface curvature attribute mesh, the volume attribute parameters of each region in the pore volume geometric marker data, and the compaction layer interface feature data, the material type to which each mesh patch in the smoothed 3D mesh model belongs is distinguished and determined, resulting in a 3D mesh model with material attribute markings, including: Based on the smoothed 3D mesh model, the arithmetic mean of all vertex curvature attribute values ​​is extracted for each mesh patch to obtain the curvature feature value of each mesh patch; the curvature feature value of each mesh patch is compared with the preset interface curvature judgment threshold one by one and the category is marked to obtain the curvature classification feature of the mesh patch. Based on the curvature attribution characteristics of the grid patches and the pore volume geometric marker data, it is detected whether each grid patch falls into the spatial region of the pore geometric marker volume, and the pore attribution characteristics of the grid patches are obtained. Based on the pore location characteristics of the grid surface and the compaction layer interface characteristics, it is detected whether each grid surface falls into the compaction layer interface spatial region, and the grid surface layer interface location characteristics are obtained. Based on the layering interface characteristics of the mesh patches, each mesh patch is classified into four categories: porous defect material, compacted layering interface weak material, aggregate-cement matrix interface transition material, and concrete matrix material. The classification results are then used as material property markers and attached to the corresponding mesh patches to obtain a three-dimensional mesh model with material property markers.

Images