Water wall hole area sealing method and system using 3D scanning

Abstract

This invention relates to the field of computer-aided design technology, specifically to a method and system for sealing water-cooled wall hole areas using 3D scanning. The method includes the following steps: extracting point cloud normals to solve partial differential equations to generate a macroscopic base surface; calculating distances and combining spatial gradients to generate and superimpose microscopic defect surfaces; decomposing shape operators to extract curvature and construct a non-uniform node network; translating along the normal to generate a three-dimensional model and extracting a set of reference points; orthogonally decomposing to obtain rigid body parameters to calculate distances and generate a deviation field. In this invention, the macroscopic morphology is reconstructed by constructing partial differential equations, vector superposition is completed by combining microscopic surface compensation, a non-uniform network is adaptively constructed based on curvature and stitched together to form a three-dimensional model, transformation parameters are obtained to eliminate residuals, avoiding the loss of microscopic morphology defects caused by manual surveying, replacing the traditional manual cutting and grinding process with precise microscopic compensation, and reducing leakage risks caused by mechanical fastening by relying on rigid body registration to analyze gaps.

Description

Technical Field

[0001] This invention relates to the field of computer-aided design technology, and in particular to a method and system for sealing water-cooled wall hole areas using 3D scanning. Background Technology

[0002] Computer-aided design technology encompasses a complete technology system, from physical models to digital 3D models, and further to customized product design and manufacturing. Its core aspects mainly include high-precision 3D data acquisition, comparison and deviation analysis between measured models and theoretical models, parameterized generation of customized structural shapes based on geometric deviation distribution, and data-driven CNC machining and assembly guidance. This field combines optical measurement equipment and computer software to transform complex irregular curved surfaces or 3D spatial coordinates deformed by heat into digital spatial parameters, thereby replacing the manual drawing and mapping process and directly outputting structural design parameters for physical machining. The traditional water-cooled wall hole sealing method and system refers to the reconstruction of the hole sealing surface caused by thermal stress in the thermally affected area of ​​high-temperature equipment such as boilers, which leads to changes in the original size. It usually relies on maintenance personnel to manually measure and record the deformed contour around the water-cooled wall opening using vernier calipers and arc templates. Then, based on the dimensional parameters measured on site, the standard-formed flexible sealing gasket or metal-clad part is manually cut and the edges are ground. Finally, in the equipment assembly area, the trimmed sealing part is pressed onto the hole contour with surface defects by repeated adjustments and trial assembly and mechanical fastening force.

[0003] Existing methods for reconstructing holes deformed by heat rely on manual measurement and recording using calipers and templates. Based on the on-site measurements, the molded gaskets or metal-clad parts are manually cut and polished. In the assembly area, the parts are pressed onto the defective hole contour by repeated adjustments and trial assembly and the application of tightening force. Manual operation results in low measurement accuracy and inability to capture microscopic morphology. Repeated trial assembly and cutting reduce overall assembly efficiency. Relying on external force for forced tightening poses the risk of not being able to completely fit the surface defects and may cause leakage hazards. Summary of the Invention

[0004] To achieve the above objectives, the present invention adopts the following technical solution: a water-cooled wall hole area sealing method using 3D scanning, comprising the following steps: S1: Obtain the set of spatial point coordinates of the water-cooled wall hole area to generate a three-dimensional point cloud of the water-cooled wall hole area, downsample to obtain a sparse three-dimensional point cloud, extract the spatial normal vector field and solve the partial differential equation of the implicit indicator function, and extract the isosurface to generate the macroscopic base surface. S2: Calculate the spatial two-point straight line deviation between the three-dimensional coordinate node data of the three-dimensional point cloud of the water-cooled wall hole area and the nearest neighbor mapping base point of the macroscopic base surface, generate the residual distance scalar field, calculate the spatial gradient to obtain the residual scalar field gradient field, solve the spatial partial differential equation to generate the microscopic defect compensation surface, and superimpose it with the macroscopic base surface to generate the water-cooled wall base composite surface. S3: Calculate and decompose the shape operator for the composite surface of the water-cooled wall substrate, extract the maximum principal curvature value, the minimum principal curvature value and the principal curvature direction vector, calculate the Gaussian curvature value and the average curvature value, trace along the principal curvature direction vector to generate a positive AC line network and adjust the sampling density to generate a non-uniform node network. S4: Translate along the three-dimensional normal vector of the non-uniform node network to generate a top-level control node set, merge and splice them to generate a three-dimensional digital model of the seal, extract the outer shell contour nodes and inner layer contact nodes, and generate a set of seal installation reference points. S5: Perform orthogonal decomposition on the outer layer point cloud data and the outer shell contour nodes in the set of installation reference points of the sealing component to obtain rotation matrix parameters and translation vector parameters, transform the inner layer contact nodes to generate actual installation space coordinates, calculate the spatial distance deviation between the two points of the composite surface of the water-cooled wall substrate, and generate the hidden gap deviation field.

[0005] As a further embodiment of the present invention, the macroscopic substrate surface specifically comprises an overall contour patch, a basic skeleton mesh, and a global topological boundary; the composite surface of the water-cooled wall substrate includes residual compensation patches, microscopic raised patches, and a comprehensive surface morphology; the non-uniform node network specifically refers to high-density clusters, sparse connecting lines, and adaptive mesh pores; the combination of the sealing component installation reference point set specifically comprises outer assembly anchor points, inner wall fitting reference points, and boundary alignment markers; and the hidden gap deviation field includes a positive interference zone, a negative gap band, and a discrete error matrix.

[0006] As a further aspect of the present invention, the step of obtaining the macroscopic base surface specifically includes: S101: Obtain the set of spatial point coordinates of the water-cooled wall hole area collected by the 3D scanning device, extract the coordinate arrays of each item in the spatial point coordinate set, merge the topological node sequence of each coordinate array, and generate a three-dimensional point cloud of the water-cooled wall hole area. S102: Based on the three-dimensional point cloud of the water-cooled wall hole area, extract the spatial coordinates of the octree nodes of the three-dimensional point cloud of the water-cooled wall hole area, calculate the size of the voxel bounding box around the spatial coordinates of the octree nodes, compare with the preset voxel downsampling benchmark value to filter the associated coordinate points inside the bounding box of the corresponding node, and obtain a sparse three-dimensional point cloud. S103: Call the sparse 3D point cloud, extract the spatial normal vector field associated with the node coordinates of the sparse 3D point cloud, construct the implicit indicator function partial differential equation based on the parameter terms of the spatial normal vector field, solve the implicit indicator function partial differential equation by numerical method to obtain the implicit indicator function scalar field, extract the corresponding isosurface network of the implicit indicator function scalar field, and generate the macroscopic base surface.

[0007] As a further aspect of the present invention, the process of constructing an implicit indicator function partial differential equation based on the spatial normal vector field parameter terms, solving the implicit indicator function partial differential equation numerically to obtain the implicit indicator function scalar field, and extracting the corresponding isosurface network of the implicit indicator function scalar field specifically includes: Extract the normal component values ​​at each node of the spatial normal vector field, calculate the divergence scalar values ​​of the spatial normal vector field in each dimension of three-dimensional space, and use the divergence scalar values ​​as the right-hand source terms of the implicit indicator function partial differential equation to establish the implicit indicator function partial differential equation. Extract the vertex coordinates of the convex hull boundary of the sparse 3D point cloud, configure the boundary condition parameters of the implicit indicator function partial differential equation based on the vertex coordinates of the convex hull boundary, perform iterative numerical discretization solution operation on the implicit indicator function partial differential equation according to the boundary condition parameters, and combine the implicit indicator function values ​​at each spatial node to generate an implicit indicator function scalar field. The distribution interval of the implicit indicator function scalar field is statistically analyzed, and the median of the distribution interval is taken as the isosurface threshold parameter. The set of spatial node coordinates in the implicit indicator function scalar field that the function values ​​cross the isosurface threshold parameter is extracted. The node connection relationship is established based on the topological connectivity of the spatial node coordinate set, and the corresponding isosurface network is extracted.

[0008] As a further aspect of the present invention, the step of obtaining the composite curved surface of the water-cooled wall substrate specifically includes: S201: Extract the three-dimensional coordinate node data of the three-dimensional point cloud of the water-cooled wall hole area and the surface mesh data of the macroscopic base surface, search for the nearest neighbor mapping base point position, calculate the straight line deviation between two points in space, use the straight line deviation between two points in space as the signed residual distance value, assign it to the corresponding node attribute item, and establish a residual distance scalar field. S202: Call the residual distance scalar field, calculate the partial derivative values ​​of each dimension of the residual distance scalar field to extract the gradient field of the residual scalar field, set the coefficients of the constant term of the spatial partial differential equation according to the values ​​of the gradient field of the residual scalar field, calculate the coordinates of the nodes of the topological isosurface region connected by the boundary values ​​of the spatial partial differential equation, and obtain the micro-defect compensation surface. S203: Map the micro-defect compensation surface and the macro-base surface to a unified three-dimensional spatial coordinate system, search for the nearest neighbor corresponding node pairs between the two surfaces, establish a node correspondence matrix, extract the spatial coordinate vectors of the corresponding nodes of the macro-base surface and the residual offset vectors of the corresponding nodes of the micro-defect compensation surface based on the node correspondence matrix, apply the residual offset vectors to the spatial coordinate vectors to perform the three-dimensional vector superposition operation of the corresponding nodes and extract the corresponding topology network to generate the composite surface of the water-cooled wall base.

[0009] As a further aspect of the present invention, the process of setting the coefficients of the constant term of the spatial partial differential equation according to the numerical values ​​of the residual scalar field gradient field, and calculating the coordinates of the nodes in the topological isosurface region connecting the edge boundary values ​​of the spatial partial differential equation is specifically as follows: Extract the magnitude values ​​of the gradient vectors at each node of the residual scalar field gradient field, perform global normalization on the magnitude values ​​of the gradient vectors, and map the normalized magnitude values ​​of the gradient vectors at each node to the constant term coefficients of the corresponding spatial nodes of the spatial partial differential equation, thereby establishing a spatial partial differential equation containing spatially distributed constant term coefficients. Extract the upper and lower bounds of the global numerical distribution interval of the residual distance scalar field, take the mean of the upper and lower bounds of the global numerical distribution interval as the edge boundary value of the spatial partial differential equation, perform iterative numerical discretization operation on the spatial partial differential equation, obtain the scalar value of the spatial partial differential equation solution at each spatial node, extract the set of spatial node coordinates in the distribution field of the scalar value of the spatial partial differential equation solution where the function value crosses the edge boundary value of the spatial partial differential equation, establish node connection relationship based on the topological connectivity of the spatial node coordinate set, and connect to form a topological isosurface region node coordinate network.

[0010] As a further aspect of the present invention, the step of obtaining the non-uniform node network specifically includes: S301: For the synthetic surface of the water-cooled wall substrate, extract the node distribution parameters corresponding to the three-dimensional topological features, calculate the partial derivative differential matrix of the surface shape operator, perform eigenvalue decomposition operation, extract the corresponding orthogonal eigenvectors as principal curvature direction vectors, obtain the maximum principal curvature value, the minimum principal curvature value and the principal curvature direction vector, and output the curvature analysis results. S302: Based on the curvature analysis results, calculate the Gaussian curvature value and the average curvature value according to the algebraic product and arithmetic mean of the maximum principal curvature value and the minimum principal curvature value, perform tangential streamline tracing operation along the principal curvature direction vector on the synthetic surface of the water-cooled wall substrate, connect adjacent nodes, and establish a positive AC line network. S303: Based on the positive AC line network, read the parameters of the absolute value range corresponding to the maximum principal curvature value, the minimum principal curvature value, and the principal curvature direction vector. Compare the absolute values ​​of the maximum and minimum principal curvature values ​​to allocate the grid node spacing ratio in each region. Perform density adjustment operation on the sampling interval of the positive AC line network according to the grid node spacing ratio in each region to generate a non-uniform node network.

[0011] As a further aspect of the present invention, the process of allocating the grid node spacing ratio of each region by comparing the absolute values ​​of the maximum and minimum principal curvature values ​​is specifically as follows: Extract the maximum and minimum principal curvature values ​​at each node from the curvature analysis results. Calculate the absolute values ​​of the maximum and minimum principal curvature values ​​at each node. Merge the two sets of absolute value data and calculate the combined maximum value across all nodes. Take the combined maximum value as the curvature absolute value normalization benchmark. Divide the absolute values ​​of the maximum and minimum principal curvature values ​​at each node by the curvature absolute value normalization benchmark to obtain the maximum and minimum principal curvature normalization coefficients at each node. The larger of the normalized coefficients of the maximum and minimum principal curvatures at each node is taken as the curvature dominance coefficient at that node. The average of the node spacing across the entire positive AC line network is used as the baseline spacing. A preset small positive real number is added to the curvature dominance coefficient to prevent overflow due to division by zero. The product of the reciprocal of the compensated curvature dominance coefficient and the baseline spacing is taken as the grid node spacing value at that node. Based on the ratio of the grid node spacing value of each node to the baseline spacing, the grid node spacing ratio of each region is output.

[0012] As a further aspect of the present invention, the step of obtaining the set of reference points for sealing component installation specifically includes: S401: Extract the three-dimensional normal vector coordinate parameters of the non-uniform node network, read the preset fixed thickness offset value along the direction of the three-dimensional normal vector, set the corresponding spatial position increment parameter, and perform a spatial coordinate translation operation according to the spatial position increment parameter to obtain the top-level control node set. S402: Call the non-uniform node network and the top-level control node set, read the feature sequence of the shared connection control point of the three-dimensional coordinates of the two node sets, perform spline patch splicing operation according to the feature sequence of the shared connection control point to connect the mesh gap positions, and obtain the three-dimensional digital model of the seal. S403: Extract the connectivity feature parameters of the mesh topology corresponding to the three-dimensional digital model of the seal and perform inner and outer surface separation analysis. Extract the three-dimensional coordinate matrix of the outer shell contour node and inner contact node from the mesh topology separation result and perform combination packaging and encapsulation operation to generate a set of seal installation reference points.

[0013] As a further aspect of the present invention, the step of obtaining the concealed gap deviation field specifically includes: S501: Collect measured spatial point data of the metal sealing component installation surface, perform three-dimensional coordinate system mapping and projection operation to generate installation outer layer point cloud data, separate and extract the three-dimensional coordinate space sequence of the shell contour node from the sealing component installation reference point set combination, calculate the covariance matrix distribution parameters of the installation outer layer point cloud data and the shell contour node, perform eigenvector orthogonal decomposition on the covariance matrix distribution parameters to extract the corresponding rigid body motion transformation relationship variables, and obtain the rotation matrix parameters and translation vector parameters; S502: Call the rotation matrix parameters and translation vector parameters to extract the inner layer contact node coordinate array distribution parameters from the set of sealing installation reference points. Apply the rotation matrix parameters and translation vector parameters to the inner layer contact node coordinate array distribution parameters to perform an affine spatial coordinate product transformation operation to obtain the actual installation spatial coordinates. S503: Based on the actual installation spatial coordinates and the composite surface of the water-cooled wall base, extract the nearest neighbor surface node coordinate sequence of the actual installation spatial coordinates and the corresponding composite surface of the water-cooled wall base within a unified three-dimensional reference coordinate system, calculate the distance deviation value between the two points in space, combine the scalar distribution characteristics of the distance deviation value between the two points in space according to the topological network connection sequence, and generate a hidden gap deviation field.

[0014] A water-cooled wall perforation sealing system utilizing 3D scanning includes: The base surface reconstruction module obtains the set of spatial point coordinates of the water-cooled wall hole area and combines them to generate a three-dimensional point cloud of the water-cooled wall hole area. It then downsamples to obtain a sparse three-dimensional point cloud, extracts the spatial normal vector field and solves the partial differential equation of the implicit indicator function, and extracts the isosurface to generate a macroscopic base surface. The defect compensation fusion module calculates the spatial two-point straight line deviation between the three-dimensional coordinate node data of the three-dimensional point cloud of the water-cooled wall hole area and the nearest neighbor mapping base point of the macroscopic base surface, generates a residual distance scalar field, calculates the spatial gradient to obtain the gradient field of the residual scalar field, solves the spatial partial differential equation to generate a microscopic defect compensation surface, and superimposes it with the macroscopic base surface to generate a composite surface of the water-cooled wall base. The curvature adaptive point layout module calculates and decomposes the shape operator for the synthetic surface of the water-cooled wall substrate, extracts the maximum principal curvature value, the minimum principal curvature value and the principal curvature direction vector, calculates the Gaussian curvature value and the average curvature value, traces along the principal curvature direction vector to generate a positive AC line network and adjusts the sampling density to generate a non-uniform node network. The reference point set generation module translates along the three-dimensional normal vector of the non-uniform node network to generate a top-level control node set, merges and splices them to generate a three-dimensional digital model of the seal, extracts the outer shell contour nodes and inner layer contact nodes, and generates a combination of seal installation reference point sets. The installation deviation analysis module performs orthogonal decomposition on the outer layer point cloud data and the outer shell contour nodes in the set of installation reference points of the sealing component to obtain rotation matrix parameters and translation vector parameters, transforms the inner layer contact nodes to generate actual installation spatial coordinates, calculates the spatial distance deviation between the composite surface of the water-cooled wall substrate and the outer shell contour nodes, and generates a hidden gap deviation field.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, partial differential equations are solved for 3D point clouds to extract isosurfaces and generate macroscopic base surfaces. The deviation of the mapped base point straight line is calculated and combined with spatial gradient to generate microscopic defect compensation surfaces to complete vector superposition. The shape operator of the synthesized surface is decomposed to extract curvature features. A non-uniform node network is constructed by tracking and adjusting the sampling density along the principal curvature direction. The network is translated along the normal direction and stitched together to generate a 3D digital model. Orthogonal decomposition is performed to obtain transformation parameters and distance deviations, avoiding the loss of microscopic morphology defects caused by manual mapping. Based on curvature adaptive network and precise compensation of microscopic residuals, the traditional manual process of repeated cutting, polishing and trial assembly is completely replaced. Rigid body motion registration is used to eliminate the risk of poor fit and operation leakage caused by forced mechanical fastening. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a schematic diagram of the steps of the present invention; Figure 2 This is a detailed schematic diagram of S1 of the present invention; Figure 3 This is a detailed schematic diagram of S2 of the present invention; Figure 4 This is a detailed schematic diagram of S3 of the present invention; Figure 5 This is a detailed schematic diagram of S4 of the present invention; Figure 6 This is a detailed schematic diagram of S5 of the present invention; Figure 7 This is a system module diagram of the present invention. Detailed Implementation

[0018] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0019] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0020] Please see Figure 1 This invention provides a method for sealing water-cooled wall perforation areas using 3D scanning, comprising the following steps: S1: Obtain the set of spatial point coordinates of the water-cooled wall hole area to generate a three-dimensional point cloud of the water-cooled wall hole area, downsample to obtain a sparse three-dimensional point cloud, extract the spatial normal vector field and solve the partial differential equation of the implicit indicator function, and extract the isosurface to generate the macroscopic base surface. S2: Calculate the spatial two-point straight line deviation between the three-dimensional coordinate node data of the three-dimensional point cloud of the water-cooled wall hole area and the nearest neighbor mapping base point of the macroscopic base surface, generate the residual distance scalar field, calculate the spatial gradient to obtain the gradient field of the residual scalar field, solve the spatial partial differential equation to generate the microscopic defect compensation surface, and superimpose it with the macroscopic base surface to generate the composite surface of the water-cooled wall base. S3: Calculate and decompose the shape operator for the composite surface of the water-cooled wall substrate, extract the maximum principal curvature value, the minimum principal curvature value and the principal curvature direction vector, calculate the Gaussian curvature value and the average curvature value, trace along the principal curvature direction vector to generate a positive AC line network and adjust the sampling density to generate a non-uniform node network. S4: Translate along the three-dimensional normal vector of the non-uniform node network to generate the top-level control node set, merge and splice to generate the three-dimensional digital model of the seal, extract the outer shell contour nodes and inner layer contact nodes, and generate the seal installation reference point set combination. S5: Perform orthogonal decomposition on the outer shell contour nodes in the combination of the outer layer point cloud data and the sealing component installation reference point set to obtain rotation matrix parameters and translation vector parameters, transform the inner layer contact nodes to generate the actual spatial coordinates of the installation, calculate the spatial distance deviation between the composite surface of the water-cooled wall and the substrate, and generate the hidden gap deviation field.

[0021] The macroscopic base surface specifically includes the overall contour patch, the basic skeleton mesh, and the global topological boundary. The composite surface of the water-cooled wall base includes residual compensation patches, microscopic raised patches, and comprehensive surface morphology. The non-uniform node network specifically refers to high-density clusters, sparse connecting lines, and adaptive mesh pores. The combination of sealing component installation reference point sets specifically includes outer assembly anchor points, inner wall fitting reference points, and boundary alignment markers. The hidden gap deviation field includes the positive interference zone, the negative gap band, and the discrete error matrix.

[0022] Please see Figure 2 The specific steps for obtaining the macroscopic base surface are as follows: S101: Obtain the set of spatial point coordinates of the water-cooled wall hole area collected by the 3D scanning device, extract the coordinate arrays of each item in the spatial point coordinate set, merge the topological node sequence of each coordinate array, and generate a three-dimensional point cloud of the water-cooled wall hole area. The system retrieves the raw 3D coordinate dataset acquired by the laser scanner. It reads the horizontal, vertical, and depth coordinates within the dataset line by line, storing each value into a pre-initialized one-dimensional floating-point array. It then iterates through these arrays, extracting the corresponding horizontal, vertical, and depth coordinates at the same index position, and combining them into a single 3D vector containing three floating-point numbers. An empty list of topology nodes is created, and the combined 3D vector is appended to this list sequentially. Each new vector is assigned a unique integer identifier as its topology node number. A hash table is built, using the topology node number as the key and the 3D vector as the value, ensuring adjacent scan points are stored contiguously in memory. The hash table is verified to be free of null values ​​or abnormal coordinates. The topology node sequence is then merged, outputting a 3D point cloud of the water-cooled wall pore area containing 150,000 discrete spatial point coordinates.

[0023] S102: Based on the 3D point cloud of the water-cooled wall hole area, extract the spatial coordinates of the octree nodes of the 3D point cloud of the water-cooled wall hole area, calculate the size of the bounding box of the voxels around the spatial coordinates of the octree nodes, compare with the preset voxel downsampling benchmark value to filter the associated coordinate points inside the bounding box of the corresponding node, and obtain the sparse 3D point cloud. Obtain the 3D point cloud of the water-cooled wall hole area output from the previous step. Calculate the maximum and minimum values ​​of the global coordinates in the horizontal, vertical, and depth directions, and subtract them to obtain the physical size of the global bounding box. Set the initial octree data structure to a level depth of 8. Recursively divide the global bounding box into equal parts in the three dimensions at a 50% ratio until the 8th level is reached, generating multiple local voxel bounding boxes. Extract the 3D spatial center point coordinates of the local voxel bounding boxes, and set the voxel downsampling benchmark value to 2 mm. Traverse all local voxel bounding boxes and determine whether their 3D side length is greater than the preset voxel downsampling benchmark value of 2 mm. For voxel bounding boxes that meet the size condition, scan and record the set of 3D point cloud spatial coordinates of all points falling inside the bounding box. Calculate the arithmetic mean of the associated coordinate points inside the bounding box in the horizontal, vertical, and depth directions to generate a representative 3D coordinate point. Discard all scattered coordinate points inside the original bounding box, and only retain the coordinate point of the arithmetic mean as the downsampled node data. After processing all voxel bounding boxes in sequence, a sparse 3D point cloud with voxelization downsampling is output, reducing the data volume from 150,000 nodes to approximately 35,000 nodes, effectively reducing subsequent memory usage.

[0024] S103: Call the sparse 3D point cloud, extract the spatial normal vector field associated with the node coordinates of the sparse 3D point cloud, construct the implicit indicator function partial differential equation based on the parameter terms of the spatial normal vector field, solve the implicit indicator function partial differential equation by numerical method to obtain the implicit indicator function scalar field, extract the corresponding isosurface network of the implicit indicator function scalar field, and generate the macroscopic base surface. For each node in a sparse 3D point cloud, extract its 15 nearest neighbors. Construct a local covariance matrix using principal component analysis (PCA). Perform eigenvalue decomposition to extract the eigenvector corresponding to the smallest eigenvalue, which serves as the spatial normal component of that node. Calculate the partial derivatives of the spatial normal vector field in the three independent dimensions of horizontal, vertical, and depth, and sum them to obtain a divergence scalar value. Configure this divergence scalar value as the source term constant on the right-hand side of the implicit indicator function partial differential equation. Extract the coordinates of the convex hull boundary vertices of the sparse 3D point cloud, and force the implicit indicator function values ​​of these vertices to a constant of 0 as Dirichlet boundary condition parameters. Construct a partial differential equation containing a Laplace operator and convert it into a system of linear algebraic equations. Perform up to 500 discrete-time solutions using the Gauss-Seidel iteration method until the root mean square of the residuals between two consecutive iterations is less than 0.0001. Collect the converged values ​​of each spatial node to generate an implicit indicator function scalar field. The maximum and minimum values ​​of the scalar field are read, and half of their sum is taken as the median, which is then set as the isosurface threshold parameter. A moving cube algorithm is used to traverse the entire space, locking adjacent spatial nodes whose implicit indicator function values ​​abruptly change from less than the median to greater than the median. Linear interpolation is then used to calculate the set of spatial node coordinates that accurately cross the isosurface threshold parameter. Based on the intersection state of the cube mesh edges, the connectivity between adjacent triangular faces is established, generating a continuous isosurface network, which is then output as the macroscopic base surface.

[0025] Please see Figure 3 The specific steps for obtaining the composite surface of the water-cooled wall substrate are as follows: S201: Extract the 3D coordinate node data of the 3D point cloud of the water-cooled wall hole area and the planar mesh data of the macroscopic base surface, search for the nearest neighbor mapping base point position, calculate the straight line deviation between two points in space, use the straight line deviation between two points in space as the signed residual distance value, assign it to the corresponding node attribute item, and establish a residual distance scalar field. A discrete spatial segmentation tree data structure is established for the set of triangular facets on the macroscopic base surface. All 3D coordinate nodes of the 3D point cloud in the water-cooled wall cavity area are traversed, and each 3D coordinate node is input into the discrete spatial segmentation tree for fast retrieval. The Euclidean distance between the input coordinate node and the plane containing all triangular facets on the macroscopic base surface is calculated, and the coordinates of the perpendicular point on the facet with the smallest distance are extracted as the nearest neighbor mapping base point. The 3D coordinate node data and the corresponding nearest neighbor mapping base point coordinates are obtained. The sum of the squares of the coordinate differences in the three dimensions are then squared to obtain the linear deviation between the two points in space. The dot product of the directed vector from the input coordinate node to the mapping base point and the normal vector of the macroscopic base surface at the mapping base point is calculated. If the dot product is positive, the linear deviation between the two points in space is assigned a positive sign, indicating that the point cloud is located outside the surface; if the dot product is negative, it is assigned a negative sign, indicating that the point cloud is located inside the surface, thus generating a signed residual distance value. A set of floating-point attribute storage fields is allocated in memory. According to the correspondence of the topological node sequence, the signed residual distance values ​​are written one by one into the deviation attribute item in the data structure of each node to complete the assignment operation of the global nodes. A residual distance scalar field containing spatial coordinates and corresponding deviation parameters is established for subsequent feature gradient calculation.

[0026] S202: Call the residual distance scalar field, calculate the partial derivative values ​​of each dimension of the residual distance scalar field to extract the gradient field of the residual scalar field, set the coefficients of the constant term of the spatial partial differential equation according to the gradient field values ​​of the residual scalar field, calculate the boundary values ​​of the spatial partial differential equation, connect the node coordinates of the topological isosurface region, and obtain the micro-defect compensation surface. The residual distance scalar field is traversed, and the central difference method is used to extract the numerical differences of the deviations between adjacent nodes in the horizontal, vertical, and depth dimensions for each node. These differences are then divided by the physical distance between nodes to construct a three-dimensional partial derivative vector, i.e., the gradient field of the residual scalar field. The square root of the sum of the squares of the three-dimensional components of the gradient vector at each node is calculated to obtain the magnitude of the gradient vector. The maximum and minimum values ​​of the global gradient vector magnitude are found, and the min-max normalization algorithm is used to linearly map all gradient vector magnitudes to a dimensionless interval of 0 to 1. The normalized values ​​are used as diagonal elements of the coefficient matrix and substituted into a pre-defined three-dimensional spatial partial differential equation, replacing the original constant term coefficients. The highest and lowest bounding values ​​of all deviations in the residual distance scalar field are found, and their sum is divided by 2. The resulting mean is set as the first-type boundary condition value for the edge region of the spatial partial differential equation. The Jacobi iteration method is used to perform up to 1000 numerical discretization operations on the spatial partial differential equation containing spatially distributed constant term coefficients. Extract the scalar values ​​from the converged partial differential equations and locate the set of zero-intersection point coordinates in the discrete 3D mesh space where the scalar values ​​are exactly equal to the edge boundary values. Based on the shared surface properties of adjacent voxels in space, connect these zero-intersection point coordinates sequentially using line segments to construct a closed topological isosurface region node coordinate network, thereby obtaining a microscopic defect compensation surface that accurately reflects the local concavity and convexity features.

[0027] S203: Map the micro-defect compensation surface and the macro-base surface to the interior of a unified three-dimensional spatial coordinate system, search for the nearest neighbor corresponding node pairs between the two surfaces, establish a node correspondence matrix, extract the spatial coordinate vectors of the corresponding nodes of the macro-base surface and the residual offset vectors of the corresponding nodes of the micro-defect compensation surface based on the node correspondence matrix, apply the residual offset vectors to the spatial coordinate vectors to perform the three-dimensional vector superposition operation of the corresponding nodes and extract the corresponding topology network to generate the composite surface of the water-cooled wall base; The local coordinate reference parameters of the micro-defect compensation surface and the global coordinate reference parameters of the macro-base surface are multiplied using matrix multiplication to eliminate relative translation and rotation offsets, ensuring that both surfaces are projected onto the same standard three-dimensional spatial coordinate frame. Based on a search radius of 0.1 mm (nearest neighbor distance threshold), the nodes of the micro-defect compensation surface and the macro-base surface are compared one by one, and pairs of nodes with the smallest spatial distance are selected. The index numbers of the corresponding nodes are recorded in a two-dimensional array, establishing a node correspondence matrix containing 50,000 sets of corresponding indices. The reference spatial coordinate vectors of the corresponding nodes on the macro-base surface are read according to the indices in the two-dimensional array, and the three-dimensional residual offset vectors of the corresponding nodes on the micro-defect compensation surface are also read. The three components of the reference spatial coordinate vector are algebraically added to the three components of the residual offset vector to achieve three-dimensional vector superposition of the corresponding nodes. The original triangular connected index array of the micro-surface is extracted and mapped onto the new coordinate point set after superposition, generating a composite surface of the water-cooled wall base with a surface topology network.

[0028] Table 1. Nodal Residual Vector Compensation Data Table 1 (10.5, 20.1, 5.0) (0.1, -0.2, 0.0) (10.6, 19.9, 5.0) 2 (12.0, 21.0, 5.2) (-0.1, 0.1, 0.1) (11.9, 21.1, 5.3) 3 (15.5, 25.4, 6.0) (0.2, 0.0, -0.1) (15.7, 25.4, 5.9) As shown in Table 1, the vector superposition process was verified by extracting some node data. The results show that the residual offset vector successfully endowed the base node with microscopic morphological features.

[0029] Please see Figure 4 The specific steps for obtaining a non-uniform node network are as follows: S301: For the synthetic surface of water-cooled wall substrate, extract the node distribution parameters corresponding to the three-dimensional topological features, calculate the partial derivative differential matrix of the surface shape operator, perform eigenvalue decomposition operation, extract the corresponding orthogonal eigenvectors as principal curvature direction vectors, obtain the maximum principal curvature value, the minimum principal curvature value and the principal curvature direction vector, and output the curvature analysis results. The node coordinates and their first-order neighborhood node sets of the synthesized surface of the water-cooled wall substrate are retrieved. A quadratic surface fitting algorithm is used to calculate the coefficient matrices of the first and second fundamental forms of the local surface at each node. By solving these coefficient matrices, a partial derivative differential matrix of the surface shape operator, i.e., the Weingarden mapping matrix, describing the surface curvature, is constructed. Singular value decomposition is performed on the Weingarden mapping matrix to extract a diagonal matrix containing two real eigenvalues ​​and an orthogonal matrix containing two corresponding eigenvectors. The eigenvalue with the larger value in the diagonal matrix is ​​defined as the maximum principal curvature value, and the eigenvalue with the smaller value is defined as the minimum principal curvature value. Simultaneously, the orthogonal eigenvectors corresponding to these two eigenvalues ​​are extracted from the orthogonal matrix and assigned as principal curvature direction vectors. The maximum principal curvature value, minimum principal curvature value, and normalized principal curvature direction vector of each synthesized node in the global domain are merged and written into a structure array to form the curvature analysis results. This operation directly quantifies the geometric curvature characteristics of the surface into specific tensor data, providing a reasonable mathematical reference for subsequent division of the surface into flat and sharp regions.

[0030] S302: Based on the curvature analysis results, calculate the Gaussian curvature value and the average curvature value according to the algebraic product and arithmetic mean of the maximum principal curvature value and the minimum principal curvature value. Perform tangential streamline tracing operation on the synthetic surface of the water-cooled wall substrate along the principal curvature direction vector, connect adjacent nodes, and establish a positive AC line network. The curvature analysis result structure array is traversed. For each node, its maximum and minimum principal curvature values ​​are extracted and algebraically multiplied to obtain the Gaussian curvature value of that node, used to determine local convexity / concavity properties or saddle point features. The maximum and minimum principal curvature values ​​are then algebraically added and divided by 2 to obtain the average curvature value of that node, used to characterize the local overall average bending trend. Using the set starting node as the origin, the principal curvature direction vector of that node is read. The step size parameter is set to 0.5 mm, and the coordinates of the next streamline node are calculated using the fourth-order Runge-Kutta numerical integration method along the three-dimensional tangential direction of the principal curvature direction vector. The tangential streamline tracing operation is repeated until the streamline reaches the topological boundary of the synthesized surface. Similarly, secondary streamlines are generated in the transverse tangential direction of the principal curvature direction. The coordinates of the intersection points are extracted at the mesh intersections, and adjacent intersection point coordinates are connected sequentially by straight line segments to construct an initial orthogonal streamline network composed of orthogonal mesh wireframes. The network is distributed along the natural curvature of the surface, which effectively reduces the surface distortion rate in subsequent geometric modeling.

[0031] S303: Based on the positive AC line network, read the parameters of the absolute value range of the maximum principal curvature value, the minimum principal curvature value, and the principal curvature direction vector. Compare the absolute values ​​of the maximum and minimum principal curvature values ​​to allocate the grid node spacing ratio of each region. Perform density adjustment operation on the sampling interval of the positive AC line network according to the grid node spacing ratio of each region to generate a non-uniform node network. The maximum and minimum principal curvature values ​​for each node in the curvature analysis results are read. The absolute value function is used to remove the sign of the values, obtaining the absolute values ​​of the maximum and minimum principal curvatures. These two sets of absolute values ​​are merged into a one-dimensional floating-point array containing the absolute values ​​of all nodes in the entire domain. A traversal comparison method is used to find the maximum value in this array, and this maximum value is set as the curvature absolute value normalization benchmark. The two principal curvature absolute values ​​of each node are divided by the curvature absolute value normalization benchmark to obtain the maximum and minimum principal curvature normalization coefficients, which are between 0 and 1. The maximum value judgment logic is called, and the larger of the two values ​​is extracted and assigned to the node as the dominant curvature coefficient. The lengths of all initial grid segments in the positive AC line network are counted, and the arithmetic mean is calculated as the benchmark spacing value of 1.5 mm. A small positive real number compensation constant of 0.001 is forcibly added to the dominant curvature coefficient to prevent overflow errors where the denominator is zero. Calculate the reciprocal of the compensated curvature dominance coefficient, multiply this reciprocal by the baseline spacing of 1.5 mm, and obtain the adaptive target mesh node spacing value at each node. Calculate the division ratio of the target spacing value to the baseline spacing, and output the mesh node spacing ratio for each region. Based on this ratio, dynamically add or remove interpolation nodes in the positive AC line network, resulting in smaller node spacing in high curvature regions and larger node spacing in low curvature regions, generating a structurally optimized non-uniform node network.

[0032] Please see Figure 5 The specific steps for obtaining the set of reference points for seal installation are as follows: S401: Extract the three-dimensional normal vector coordinate parameters of the non-uniform node network, read the preset fixed thickness offset value along the direction of the three-dimensional normal vector, set the corresponding spatial position increment parameters, and perform spatial coordinate translation operation according to the spatial position increment parameters to obtain the top-level control node set; Extract the spatial 3D coordinates and corresponding unitized 3D normal vector coordinate parameters of each node in the non-uniform node network. A fixed thickness offset value, 3.0 mm, representing the required solid thickness of the metal sealing gasket, is preset. The horizontal, vertical, and depth components of the unitized 3D normal vector are multiplied by the fixed thickness offset value of 3.0 to calculate the spatial position increment parameters of the corresponding node in the three dimensions of the spatial coordinate frame. The original 3D coordinates of the non-uniform node network are read, and the corresponding spatial position increment parameters are added to its horizontal, vertical, and depth coordinates, performing a spatial coordinate translation operation on all nodes. All the newly calculated coordinate points after translation are entered into a new spatial point set cache array, forming a lattice topology structure. This generates a top-level control node set that floats directly above the original non-uniform node network, with a constant relative normal distance of 3.0 mm across the entire network. This calculation process, based on a rigorous vector addition criterion, pushes out the upper boundary of the model, effectively preventing mesh self-interlacing caused by complex surface offsets.

[0033] S402: Call the non-uniform node network and the top-level control node set, read the feature sequence of the shared connection control point of the three-dimensional coordinates of the two node sets, perform spline patch splicing operation according to the feature sequence of the shared connection control point to connect the mesh gap positions, and obtain the three-dimensional digital model of the seal. The non-uniform node network at the bottom and the top-level control node set at the top are read. The 3D coordinates of the outermost ring of each node set are extracted to obtain a shared sequence of control points for closed space. For the bottom network, top set, and side control point sequence, a non-uniform rational B-spline surface interpolation algorithm is used to solve the control polygon equations on the horizontal and vertical parameter space node vectors. Based on the solved surface control point coordinates, closed spline patches covering the top, bottom, and outer side edges are generated. For physical gaps between adjacent spline patches, the normal deviation of edge nodes is extracted. The control point coordinates at the patch edges are fine-tuned using curvature continuity constraints. A spline patch splicing operation is then performed to connect all mesh gap positions. The consistency of the normal vector orientation and topological watertightness of the spliced ​​patches are checked. After confirming the absence of open boundaries, a 3D digital model of the seal, including internal and external surface features and solid volume properties, is exported. This processing mechanism ensures that the digital model can well match the seamless closure requirements of the input source file for industrial 3D printing or high-precision CNC machining.

[0034] S403: Extract the connectivity feature parameters of the mesh topology corresponding to the three-dimensional digital model of the sealing component, perform inner and outer surface separation and analysis operations, extract the three-dimensional coordinate matrix of the outer shell contour node and inner contact node from the mesh topology separation results, perform combination and packaging encapsulation operations, and generate a set of sealing component installation reference points. For the mesh topology structure composed of polygonal facets within the 3D digital model of the seal, the coordinates of the geometric center point of each facet are calculated, and surface normal vectors are generated. The connectivity feature parameters between each facet and its adjacent facets in the mesh topology are extracted, and the size of the dihedral angle formed by shared edges and the 3D spatial orientation of the normal vectors are examined. Based on the outward divergence of the facet normal vectors, an analytical separation operation of the inner and outer curved surfaces is performed along the normal direction. From the mesh topology separation results, the coordinate set of normal vectors that deviate from the assembly contact area and characterize the appearance is extracted as the outer shell contour nodes; simultaneously, the coordinate set of normal vectors that point towards the bottom wall and characterize the assembly contact shape is extracted as the inner contact nodes. A data structure package containing two sets of 3D coordinate matrices is established. The extracted coordinate matrices of the outer shell contour nodes and inner contact nodes are hierarchically categorized, and a combination and packaging encapsulation operation is performed to generate a set of seal installation reference points containing accurate corresponding mapping labels. This set is used as a standard reference for direct import into the subsequent assembly deformation detection and feature comparison process.

[0035] Please see Figure 6 The specific steps for obtaining the hidden gap deviation field are as follows: S501: Collect measured spatial point data of the metal sealing component installation surface, perform three-dimensional coordinate system mapping and projection operation, generate installation outer layer point cloud data, separate and extract the three-dimensional coordinate space sequence of the shell contour node from the sealing component installation reference point set combination, calculate the covariance matrix distribution parameters of the installation outer layer point cloud data and the shell contour node, perform eigenvector orthogonal decomposition on the covariance matrix distribution parameters to extract the corresponding rigid body motion transformation relationship variables, and obtain the rotation matrix parameters and translation vector parameters; The scattered scan point coordinates of the actual mated surface of the physical seal are obtained using an external spatial coordinate measuring device. A homogeneous coordinate transformation matrix is ​​used to convert the local coordinate system data of the scanning device to a unified world coordinate framework. A three-dimensional coordinate architecture mapping and projection operation is performed to generate installation outer layer point cloud data containing the physical installation deviation state. The three-dimensional coordinate space sequence of the shell contour nodes, serving as the theoretical benchmark, is extracted from the preset channels of the data structure. The centroid coordinates of the installation outer layer point cloud data set and the centroid coordinates of the shell contour node set are calculated, and both sets of point clouds are centered by subtracting their respective centroids. The two centered sets of three-dimensional coordinate matrices are multiplied to calculate a 3x3 covariance matrix distribution parameter. Singular value decomposition is performed on the covariance matrix to extract the left singular vector matrix, the singular value diagonal matrix, and the right singular vector matrix. The right singular vector matrix is ​​multiplied by the transpose of the left singular vector matrix to extract the corresponding rigid body motion transformation variables, thereby obtaining a rotation matrix parameter containing 9 elements. Then, by subtracting the product of the rotation matrix and the outer shell contour centroid from the outer point cloud centroid, a translation vector parameter containing 3 components is obtained.

[0036] Table 2 Rigid Body Motion Transformation Parameter Matrix Translation vector 0.50 -0.20 0.15 Rotate the first row 0.9998 -0.0174 0.0000 Rotate the second row 0.0174 0.9998 0.0000 As shown in Table 2, the extracted rotation and translation parameters quantify the degree of positional offset of the solid body caused by assembly stress or machining errors.

[0037] S502: Call the rotation matrix parameters and translation vector parameters to extract the inner layer contact node coordinate array distribution parameters from the seal installation reference point set combination. Apply the rotation matrix parameters and translation vector parameters to the inner layer contact node coordinate array distribution parameters to perform affine space coordinate product transformation operation to obtain the actual installation space coordinates. The system retrieves the 3x3 rotation matrix parameters and the translation vector parameters (containing real components for lateral, longitudinal, and depth) stored in memory. From the structured data package containing the set of seal installation reference points, it fully reads the coordinate array distribution parameters of all inner contact nodes representing the ideal assembly bottom shape. The three-dimensional coordinates of each node are organized into column vectors, constructing a global coordinate matrix of dimension 3xN. The rotation matrix parameters are then multiplied by this global coordinate matrix using standard linear algebra, implementing the rigid body rotation step of the affine space coordinate product transformation operation, calculating the intermediate coordinates of each inner contact node after twisting around a specific axis. Each column vector of these intermediate coordinates is traversed, and the three real components of the translation vector parameters are accumulated in its corresponding three dimensions to complete the parallel displacement offset. After the rigorous composite affine transformation involving left multiplication of the rotation matrix and addition of the translation vector, the ideal inner contact node, which originally existed only in theoretical space, is effectively endowed with the physical offset obtained from actual measurements. This accurately generates the actual spatial coordinates of the installation, which conform to the actual assembly stress state, providing a spatial reference that closely approximates the engineering site for subsequent extraction of surface gaps.

[0038] S503: Based on the actual spatial coordinates of the installation and the composite surface of the water-cooled wall base, the nearest neighbor surface node coordinate sequence of the actual spatial coordinates of the installation and the corresponding composite surface of the water-cooled wall base is extracted within a unified three-dimensional reference coordinate system. The distance deviation between the two points in space is calculated, and the scalar distribution characteristics of the distance deviation between the two points in space are combined according to the topological network connection sequence to generate a hidden gap deviation field. Using the initially obtained and fixed-position water-cooled wall substrate composite surface as a comparison reference, the actual installation spatial coordinates generated in the previous step are imported into the same unified three-dimensional reference coordinate framework. Based on the multi-dimensional spatial segmentation tree nearest neighbor search algorithm, each node in the actual installation spatial coordinates is used as the search starting point to extract the nearest neighbor surface node coordinate sequence with the shortest Euclidean distance on the corresponding water-cooled wall substrate composite surface. The three-dimensional coordinates corresponding to the two sets of nodes are extracted, and the spatial distance deviation between each assembly node and its projection base point on the substrate surface is calculated using the spatial coordinate distance formula. If the assembly point is located outside the substrate, the deviation is recorded as a positive interference value; if there is a gap due to non-fitting, it is recorded as a negative value. The original triangular mesh topology connection relationship of the inner contact point cloud is read, and the spatial distance deviation values ​​calculated independently at each node are assigned as weight scalars to the attributes of each vertex.

[0039] Table 3. Distribution of Concealed Gap Deviations Top edge area (10.0, 20.0, 5.5) (10.0, 20.0, 5.0) 0.50 Side wall central area (15.0, 25.0, 5.0) (15.0, 25.0, 5.0) 0.00 Bottom corner area (20.0, 30.0, 4.8) (20.0, 30.0, 5.1) -0.30 Based on these scalar distribution characteristics, a hidden gap deviation field is generated in the 3D rendering engine, with color numerical gradients representing the magnitude of the error. As shown in Table 3, the distance deviation in different regions is quantified and recorded, visually exposing local non-fitting risk areas caused by the mechanical rigidity limitations of the seal.

[0040] Please see Figure 7 A water-cooled wall perforation sealing system utilizing 3D scanning includes: The base surface reconstruction module obtains the set of spatial point coordinates of the water-cooled wall hole area and combines them to generate a three-dimensional point cloud of the water-cooled wall hole area. It then downsamples to obtain a sparse three-dimensional point cloud, extracts the spatial normal vector field and solves the partial differential equation of the implicit indicator function, and extracts the isosurface to generate a macroscopic base surface. The defect compensation fusion module calculates the spatial two-point straight line deviation between the three-dimensional coordinate node data of the three-dimensional point cloud of the water-cooled wall hole area and the nearest neighbor mapping base point of the macroscopic base surface, generates a residual distance scalar field, calculates the spatial gradient to obtain the gradient field of the residual scalar field, solves the spatial partial differential equation to generate a microscopic defect compensation surface, and superimposes it with the macroscopic base surface to generate a composite surface of the water-cooled wall base. The curvature adaptive point layout module calculates and decomposes the shape operator for the synthetic surface of the water-cooled wall substrate, extracts the maximum principal curvature value, the minimum principal curvature value and the principal curvature direction vector, calculates the Gaussian curvature value and the average curvature value, traces along the principal curvature direction vector to generate a positive AC line network and adjusts the sampling density to generate a non-uniform node network. The reference point set generation module translates along the three-dimensional normal vector of the non-uniform node network to generate a top-level control node set, merges and splices them to generate a three-dimensional digital model of the seal, extracts the outer shell contour nodes and inner layer contact nodes, and generates a combination of seal installation reference point sets. The installation deviation analysis module performs orthogonal decomposition on the outer shell contour nodes in the combination of the outer layer point cloud data and the sealing component installation reference point set to obtain rotation matrix parameters and translation vector parameters. It transforms the inner layer contact nodes to generate the actual spatial coordinates of the installation, calculates the spatial distance deviation between the composite surface of the water-cooled wall and the substrate, and generates the hidden gap deviation field.

[0041] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for sealing water-cooled wall perforation areas using 3D scanning, characterized in that, Includes the following steps: S1: Obtain the set of spatial point coordinates of the water-cooled wall hole area to generate a three-dimensional point cloud of the water-cooled wall hole area, downsample to obtain a sparse three-dimensional point cloud, extract the spatial normal vector field and solve the partial differential equation of the implicit indicator function, and extract the isosurface to generate the macroscopic base surface. S2: Calculate the spatial two-point straight line deviation between the three-dimensional coordinate node data of the three-dimensional point cloud of the water-cooled wall hole area and the nearest neighbor mapping base point of the macroscopic base surface, generate the residual distance scalar field, calculate the spatial gradient to obtain the residual scalar field gradient field, solve the spatial partial differential equation to generate the microscopic defect compensation surface, and superimpose it with the macroscopic base surface to generate the water-cooled wall base composite surface. S3: Calculate and decompose the shape operator for the composite surface of the water-cooled wall substrate, extract the maximum principal curvature value, the minimum principal curvature value and the principal curvature direction vector, calculate the Gaussian curvature value and the average curvature value, trace along the principal curvature direction vector to generate a positive AC line network and adjust the sampling density to generate a non-uniform node network. S4: Translate along the three-dimensional normal vector of the non-uniform node network to generate a top-level control node set, merge and splice them to generate a three-dimensional digital model of the seal, extract the outer shell contour nodes and inner layer contact nodes, and generate a set of seal installation reference points. S5: Perform orthogonal decomposition on the outer layer point cloud data and the outer shell contour nodes in the set of installation reference points of the sealing component to obtain rotation matrix parameters and translation vector parameters, transform the inner layer contact nodes to generate actual installation space coordinates, calculate the spatial distance deviation between the two points of the composite surface of the water-cooled wall substrate, and generate the hidden gap deviation field.

2. The water-cooled wall hole sealing method using 3D scanning according to claim 1, characterized in that, The macroscopic base surface specifically comprises an overall contour patch, a basic skeleton mesh, and a global topological boundary. The composite surface of the water-cooled wall base includes residual compensation patches, microscopic raised patches, and a comprehensive surface morphology. The non-uniform node network specifically refers to high-density clusters, sparse connecting lines, and adaptive mesh pores. The set of sealing component installation reference points specifically comprises outer assembly anchor points, inner wall fitting reference points, and boundary alignment markers. The hidden gap deviation field includes a positive interference zone, a negative gap band, and a discrete error matrix.

3. The water-cooled wall hole sealing method using 3D scanning according to claim 1, characterized in that, The specific steps for obtaining the macroscopic base surface are as follows: S101: Obtain the set of spatial point coordinates of the water-cooled wall hole area collected by the 3D scanning device, extract the coordinate arrays of each item in the spatial point coordinate set, merge the topological node sequence of each coordinate array, and generate a three-dimensional point cloud of the water-cooled wall hole area. S102: Based on the three-dimensional point cloud of the water-cooled wall hole area, extract the spatial coordinates of the octree nodes of the three-dimensional point cloud of the water-cooled wall hole area, calculate the size of the voxel bounding box around the spatial coordinates of the octree nodes, compare with the preset voxel downsampling benchmark value to filter the associated coordinate points inside the bounding box of the corresponding node, and obtain a sparse three-dimensional point cloud. S103: Call the sparse 3D point cloud, extract the spatial normal vector field associated with the node coordinates of the sparse 3D point cloud, construct the implicit indicator function partial differential equation based on the parameter terms of the spatial normal vector field, solve the implicit indicator function partial differential equation by numerical method to obtain the implicit indicator function scalar field, extract the corresponding isosurface network of the implicit indicator function scalar field, and generate the macroscopic base surface.

4. The water-cooled wall hole sealing method using 3D scanning according to claim 3, characterized in that, The process of constructing an implicit indicator function partial differential equation based on the spatial normal vector field parameter terms, solving the implicit indicator function partial differential equation numerically to obtain the implicit indicator function scalar field, and extracting the corresponding isosurface network of the implicit indicator function scalar field is as follows: Extract the normal component values ​​at each node of the spatial normal vector field, calculate the divergence scalar values ​​of the spatial normal vector field in each dimension of three-dimensional space, and use the divergence scalar values ​​as the right-hand source terms of the implicit indicator function partial differential equation to establish the implicit indicator function partial differential equation. Extract the vertex coordinates of the convex hull boundary of the sparse 3D point cloud, configure the boundary condition parameters of the implicit indicator function partial differential equation based on the vertex coordinates of the convex hull boundary, perform iterative numerical discretization solution operation on the implicit indicator function partial differential equation according to the boundary condition parameters, and combine the implicit indicator function values ​​at each spatial node to generate an implicit indicator function scalar field. The distribution interval of the implicit indicator function scalar field is statistically analyzed, and the median of the distribution interval is taken as the isosurface threshold parameter. The set of spatial node coordinates in the implicit indicator function scalar field that the function values ​​cross the isosurface threshold parameter is extracted. The node connection relationship is established based on the topological connectivity of the spatial node coordinate set, and the corresponding isosurface network is extracted.

5. The water-cooled wall hole sealing method using 3D scanning according to claim 3, characterized in that, The specific steps for obtaining the composite surface of the water-cooled wall substrate are as follows: S201: Extract the three-dimensional coordinate node data of the three-dimensional point cloud of the water-cooled wall hole area and the surface mesh data of the macroscopic base surface, search for the nearest neighbor mapping base point position, calculate the straight line deviation between two points in space, use the straight line deviation between two points in space as the signed residual distance value, assign it to the corresponding node attribute item, and establish a residual distance scalar field. S202: Call the residual distance scalar field, calculate the partial derivative values ​​of each dimension of the residual distance scalar field to extract the gradient field of the residual scalar field, set the coefficients of the constant term of the spatial partial differential equation according to the values ​​of the gradient field of the residual scalar field, calculate the coordinates of the nodes of the topological isosurface region connected by the boundary values ​​of the spatial partial differential equation, and obtain the micro-defect compensation surface. S203: Map the micro-defect compensation surface and the macro-base surface to a unified three-dimensional spatial coordinate system, search for the nearest neighbor corresponding node pairs between the two surfaces, establish a node correspondence matrix, extract the spatial coordinate vectors of the corresponding nodes of the macro-base surface and the residual offset vectors of the corresponding nodes of the micro-defect compensation surface based on the node correspondence matrix, apply the residual offset vectors to the spatial coordinate vectors to perform the three-dimensional vector superposition operation of the corresponding nodes and extract the corresponding topology network to generate the composite surface of the water-cooled wall base.

6. The water-cooled wall hole sealing method using 3D scanning according to claim 5, characterized in that, The process of setting the coefficients of the constant term of the spatial partial differential equation according to the numerical values ​​of the residual scalar field gradient field, and calculating the coordinates of the nodes in the topological isosurface region connecting the boundary values ​​of the spatial partial differential equation is as follows: Extract the magnitude values ​​of the gradient vectors at each node of the residual scalar field gradient field, perform global normalization on the magnitude values ​​of the gradient vectors, and map the normalized magnitude values ​​of the gradient vectors at each node to the constant term coefficients of the corresponding spatial nodes of the spatial partial differential equation, thereby establishing a spatial partial differential equation containing spatially distributed constant term coefficients. Extract the upper and lower bounds of the global numerical distribution interval of the residual distance scalar field, take the mean of the upper and lower bounds of the global numerical distribution interval as the edge boundary value of the spatial partial differential equation, perform iterative numerical discretization operation on the spatial partial differential equation, obtain the scalar value of the spatial partial differential equation solution at each spatial node, extract the set of spatial node coordinates in the distribution field of the scalar value of the spatial partial differential equation solution where the function value crosses the edge boundary value of the spatial partial differential equation, establish node connection relationship based on the topological connectivity of the spatial node coordinate set, and connect to form a topological isosurface region node coordinate network.

7. The water-cooled wall hole sealing method using 3D scanning according to claim 5, characterized in that, The specific steps for obtaining the non-uniform node network are as follows: S301: For the synthetic surface of the water-cooled wall substrate, extract the node distribution parameters corresponding to the three-dimensional topological features, calculate the partial derivative differential matrix of the surface shape operator, perform eigenvalue decomposition operation, extract the corresponding orthogonal eigenvectors as principal curvature direction vectors, obtain the maximum principal curvature value, the minimum principal curvature value and the principal curvature direction vector, and output the curvature analysis results. S302: Based on the curvature analysis results, calculate the Gaussian curvature value and the average curvature value according to the algebraic product and arithmetic mean of the maximum principal curvature value and the minimum principal curvature value, perform tangential streamline tracing operation along the principal curvature direction vector on the synthetic surface of the water-cooled wall substrate, connect adjacent nodes, and establish a positive AC line network. S303: Based on the positive AC line network, read the parameters of the absolute value range corresponding to the maximum principal curvature value, the minimum principal curvature value, and the principal curvature direction vector. Compare the absolute values ​​of the maximum and minimum principal curvature values ​​to allocate the grid node spacing ratio in each region. Perform density adjustment operation on the sampling interval of the positive AC line network according to the grid node spacing ratio in each region to generate a non-uniform node network.

8. The water-cooled wall hole sealing method using 3D scanning according to claim 7, characterized in that, The specific steps for obtaining the set of reference points for sealing component installation are as follows: S401: Extract the three-dimensional normal vector coordinate parameters of the non-uniform node network, read the preset fixed thickness offset value along the direction of the three-dimensional normal vector, set the corresponding spatial position increment parameter, and perform a spatial coordinate translation operation according to the spatial position increment parameter to obtain the top-level control node set. S402: Call the non-uniform node network and the top-level control node set, read the feature sequence of the shared connection control point of the three-dimensional coordinates of the two node sets, perform spline patch splicing operation according to the feature sequence of the shared connection control point to connect the mesh gap positions, and obtain the three-dimensional digital model of the seal. S403: Extract the connectivity feature parameters of the mesh topology corresponding to the three-dimensional digital model of the seal and perform inner and outer surface separation analysis. Extract the three-dimensional coordinate matrix of the outer shell contour node and inner contact node from the mesh topology separation result and perform combination packaging and encapsulation operation to generate a set of seal installation reference points.

9. The water-cooled wall hole sealing method using 3D scanning according to claim 8, characterized in that, The specific steps for obtaining the hidden gap deviation field are as follows: S501: Collect measured spatial point data of the metal sealing component installation surface, perform three-dimensional coordinate system mapping and projection operation to generate installation outer layer point cloud data, separate and extract the three-dimensional coordinate space sequence of the shell contour node from the sealing component installation reference point set combination, calculate the covariance matrix distribution parameters of the installation outer layer point cloud data and the shell contour node, perform eigenvector orthogonal decomposition on the covariance matrix distribution parameters to extract the corresponding rigid body motion transformation relationship variables, and obtain the rotation matrix parameters and translation vector parameters; S502: Call the rotation matrix parameters and translation vector parameters to extract the inner layer contact node coordinate array distribution parameters from the set of sealing installation reference points. Apply the rotation matrix parameters and translation vector parameters to the inner layer contact node coordinate array distribution parameters to perform an affine spatial coordinate product transformation operation to obtain the actual installation spatial coordinates. S503: Based on the actual installation spatial coordinates and the composite surface of the water-cooled wall base, extract the nearest neighbor surface node coordinate sequence of the actual installation spatial coordinates and the corresponding composite surface of the water-cooled wall base within a unified three-dimensional reference coordinate system, calculate the distance deviation value between the two points in space, combine the scalar distribution characteristics of the distance deviation value between the two points in space according to the topological network connection sequence, and generate a hidden gap deviation field.

10. A water-cooled wall hole sealing system utilizing 3D scanning, characterized in that, The system is used to implement the water-cooled wall hole sealing method using 3D scanning as described in any one of claims 1-9, the system comprising: The base surface reconstruction module obtains the set of spatial point coordinates of the water-cooled wall hole area and combines them to generate a three-dimensional point cloud of the water-cooled wall hole area. It then downsamples to obtain a sparse three-dimensional point cloud, extracts the spatial normal vector field and solves the partial differential equation of the implicit indicator function, and extracts the isosurface to generate a macroscopic base surface. The defect compensation fusion module calculates the spatial two-point straight line deviation between the three-dimensional coordinate node data of the three-dimensional point cloud of the water-cooled wall hole area and the nearest neighbor mapping base point of the macroscopic base surface, generates a residual distance scalar field, calculates the spatial gradient to obtain the gradient field of the residual scalar field, solves the spatial partial differential equation to generate a microscopic defect compensation surface, and superimposes it with the macroscopic base surface to generate a composite surface of the water-cooled wall base. The curvature adaptive point layout module calculates and decomposes the shape operator for the synthetic surface of the water-cooled wall substrate, extracts the maximum principal curvature value, the minimum principal curvature value and the principal curvature direction vector, calculates the Gaussian curvature value and the average curvature value, traces along the principal curvature direction vector to generate a positive AC line network and adjusts the sampling density to generate a non-uniform node network. The reference point set generation module translates along the three-dimensional normal vector of the non-uniform node network to generate a top-level control node set, merges and splices them to generate a three-dimensional digital model of the seal, extracts the outer shell contour nodes and inner layer contact nodes, and generates a combination of seal installation reference point sets. The installation deviation analysis module performs orthogonal decomposition on the outer layer point cloud data and the outer shell contour nodes in the set of installation reference points of the sealing component to obtain rotation matrix parameters and translation vector parameters, transforms the inner layer contact nodes to generate actual installation spatial coordinates, calculates the spatial distance deviation between the composite surface of the water-cooled wall substrate and the outer shell contour nodes, and generates a hidden gap deviation field.

Images