An Automatic Calculation Method for Node Centers of Steel Structure Box Girder Based on Point Cloud Model
By using a point cloud model-based method, the center of the steel structure box girder node is automatically calculated, which solves the problems of high safety risk, low accuracy and low efficiency in the acquisition of node coordinates in the existing technology. It achieves high-precision and high-efficiency node coordinate extraction, supporting the accuracy of curtain wall design and installation.
Patent Information
- Application Number
- CN202310530845.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-11
- Publication Date
- 2026-11-17
- Estimated Expiration
- 2043-05-11
AI Technical Summary
Existing technologies for obtaining node coordinates in steel structure construction suffer from high safety risks, low measurement accuracy, low efficiency, and low automation. This is especially true in the design and installation of curtain walls for large-span buildings, where it is difficult to achieve accurate node position calibration.
The method based on point cloud models includes obtaining local point clouds of box girder nodes from design drawings, performing planar extraction and feature determination of local point clouds of nodes, using the least squares method to obtain the optimal node center coordinates, and combining 3D laser scanning and point cloud processing software for accurate noise reduction and planar fitting to automatically calculate the node center.
It improves the accuracy and efficiency of node coordinate acquisition, enhances the level of automation, and provides accurate coordinate data for subsequent structural construction deviation verification and decorative curtain wall design, ensuring construction quality and installation accuracy.
Smart Images

Figure CN116579060B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of civil engineering construction technology, and in particular to an automatic calculation method for the node center of a steel structure box girder based on a point cloud model. Background Technology
[0002] Steel structures are one of the main types of building structures today. Due to their light weight, high strength, good rigidity, and strong resistance to deformation, they are widely used in large-span public buildings and landmark buildings such as stadiums and convention centers. However, due to construction deviations caused by pre-assembly, welding, and hoisting processes, as well as structural deformation and natural settlement after the removal of temporary supports, there are significant differences between the actual structure after construction and the design structural model. This renders the design scheme for the decorative curtain wall unusable, requiring adjustments and modifications based on the actual structure. Therefore, it is necessary to select appropriate surveying methods to collect the coordinates of the component nodes of the actual structure to provide data support for the design and installation of the curtain wall.
[0003] Currently, two main solutions are commonly used on construction sites: First, workers use elevated equipment to approach or climb onto the surface of components, measuring their actual dimensions and calculating node coordinates with tools like tape measures. This method is high in safety, has low measurement accuracy, and is inefficient, only allowing verification of some node positions within a limited area. Second, workers pre-place numerous target points and then use a total station to measure the 3D coordinates of characteristic nodes. This method offers high measurement accuracy but is less efficient and can only sample some key nodes, not verify the coordinates of all component nodes. In addition, some projects attempt to use 3D laser scanning technology to collect 3D coordinate data of the new structure's surface, reconstruct a 3D point cloud model, and use this to calibrate and optimize the design model. The curtain wall system is then adjusted or redesigned based on the corrected model. Alternatively, the center of the upper surface of nodes is manually selected in point cloud model visualization software to obtain node coordinates for detailed curtain wall design reference. While these methods address the shortcomings of low efficiency and small coverage in field data collection, the processing flow in the office stage is cumbersome, with low accuracy, significant manual intervention, and low automation. Summary of the Invention
[0004] This invention provides an automatic calculation method for the node center of a steel structure box girder based on a point cloud model, in order to solve the above-mentioned technical problems.
[0005] To address the aforementioned technical problems, this invention provides an automatic calculation method for the node center of a steel structure box girder based on a point cloud model, comprising the following steps:
[0006] Step 1: Obtain local point cloud of box girder nodes based on design drawings;
[0007] Step 2: Local point cloud plane extraction of nodes;
[0008] Step 3: Determine the planar features of the node point cloud;
[0009] Step 4: Find the optimal node center coordinates using the least squares method.
[0010] Preferably, step 1 includes:
[0011] Step 11: Batch export the set of steel structure box girder node points M = {M1, M2, ..., M} from the design drawings. i M N} and each node M i The corresponding three-dimensional coordinates;
[0012] Step 12: Based on the 3D coordinate comparison, determine the nodes M in the point set M exported in Step 11. i If (i = 1, 2, ..., N) is an existing point in the steel structure point cloud data P after accurate denoising, then record the index value of the node in P; otherwise, add the node to the point cloud data P and record its index value.
[0013] Step 13: Construct a KD-tree for the point cloud data P. Based on the index values of each node in Step 12, search for each node M in the point cloud data P. i And using it as the center, extract points within a search radius R to form a point cloud subset p. i As the corresponding steel structure node M i Local point cloud.
[0014] Preferably, the search radius R satisfies 0.25L-σ<R<0.25L+σ, where σ is the expected deformation degree of the steel structure and L is the length of the steel structure node member.
[0015] Preferably, in step 1, the steel structure point cloud data P that has been accurately denoised is obtained by scanning the substations of a 3D laser scanner and preprocessed in point cloud processing software. The preprocessing includes at least: point cloud registration, deletion of non-object point clouds, deletion of non-connected items of object point clouds, and deletion of outliers.
[0016] Preferably, step 2 includes:
[0017] Step 21: Calculate the local point cloud p of the steel structure box girder node described in Step 13 based on principal component analysis. i The normal vectors of each point in the middle;
[0018] Step 22: Fit the node plane based on the random sampling consensus algorithm;
[0019] Step 23: Repeat step 22 for the remaining external points in step 22 to extract the local point cloud p of the nodes. iThe loop continues until the ratio of the number of remaining external points to the number of original point clouds is less than the set minimum ratio threshold p.
[0020] Preferably, step 22 includes:
[0021] Step 221: Randomly select a local point cloud p node i For three points with similar interior normal vectors, calculate the equation of the plane corresponding to them:
[0022] Ax + By + Cz + D = 0;
[0023] Step 222: Calculate the local point cloud p i All points N inside j (x j y j , z j The three-dimensional distance from the plane obtained in step 221:
[0024]
[0025] Step 223: Set the local point cloud p i The three-dimensional distance threshold from the interior point to the plane obtained in step 221 [d] N ];
[0026] Step 224: Compare the distances from each point to the plane calculated in step 222 with the distance thresholds set in step 223. If d Nj <[d N If the value is less than or equal to 0, then the point is considered an interior point; otherwise, it is considered an exterior point, and the current number of interior points N is recorded.
[0027] Step 225: Repeat steps 221 to 224 until the number of iterations reaches the set maximum number of iterations N. max ;
[0028] Step 226: Compare the number of inliers after each iteration, select the fitting plane with the most inliers as the best fitting plane, and consider the parameters corresponding to this plane as the best model parameters. Simultaneously, move the inliers from the origin cloud subset p... i Remove from the middle;
[0029] Step 227: Store the optimal model parameters (A,B,C,D) and the corresponding number of interior points N obtained in step 226 as the attribute values of the plane.
[0030] Preferably, step 3 includes:
[0031] Step 31: Read and count the normal vector and quantity characteristics of the plane obtained in Step 2, where the normal vector is the coefficient (A,B,C) of the equation expression of the plane, and the quantity is the number of interior points N of the plane;
[0032] Step 32: Calculate the angle between the normal vectors of any two planes, and count the frequency of the angle between the normal vector of each plane and the normal vector of the other plane within the range of 90°±θ1. Take the plane that is orthogonal to the other plane the most times as the outer surface. The planes with the same angle as this plane that is less than θ2 or greater than θ3 are also outer surfaces. The remaining planes are the side surfaces S. i ;
[0033] Step 33: Compare the quantitative characteristics of several outer surfaces, and mark the plane with the most internal points as the true outer surface S0 of the steel structure box girder node. The remaining outer surfaces are regarded as redundant planes or erroneous planes and removed.
[0034] Preferably, in step 32, θ1, θ2, and θ3 are taken as 10°, 10°, and 170°, respectively.
[0035] Preferably, step 4 includes:
[0036] Step 41: Set the center point O(x0, y0, z0) to be determined as a point on the outer surface S0 described in step 33. Then we have A0x0+B0y0+C0z0+D0=0, that is, z0=(-D0-A0x0-B0y0) / C0;
[0037] Step 42: Export the design node M from step 11 i coordinates (x) i y i , z i Substitute the side view S i The plane expression and the determination of A i x i +B i y i +C i z i +D i The sign of the value; if it is negative, then the value of the value on side S is positive or negative. i Find the opposite of the coefficients (A, B, C, D) of the plane equation and use them as the coefficients of the new expression for the plane; otherwise, leave them unchanged.
[0038] Step 43: Assume point O extends to any side surface S i The three-dimensional distance is half the design width of the steel structure box girder, denoted as d0. Then the following relationship is established:
[0039]
[0040] Substituting the relation described in step 41 into the above equation, we get:
[0041]
[0042] Step 44: Combining the equations from Step 43, we obtain the following system of linear equations:
[0043]
[0044] Simplifying the above system of linear equations and rewriting it in matrix form, we get:
[0045] AX0 = K;
[0046] in:
[0047]
[0048] Step 45: Find the least squares solution to the overdetermined system of equations listed in Step 44, and define the error function:
[0049] L(x0, y0) = ||AX0 - K||2;
[0050] Step 46: Differentiate with respect to X0, and solve for the X0 that minimizes the above error function, i.e.:
[0051] X0=(A T A) -1 A T K;
[0052] Substitute the obtained (x0, y0) into the relationship described in step 41 to obtain z0, and get the coordinates (x0, y0, z0) of point O. These coordinates are the optimal center point coordinates of the steel structure box girder node.
[0053] Compared with existing technologies, the automatic calculation method for the node center of steel box girder based on point cloud model provided by this invention obtains the precise position of the node center based on the planar features of the point cloud of the steel box girder node, thereby improving the accuracy, efficiency and automation of the extraction of the node position coordinates of the steel box girder, and providing accurate coordinate data for subsequent structural construction deviation verification, decorative curtain wall design and installation. Attached Figure Description
[0054] Figure 1 This is a flowchart of an automatic calculation method for the node center of a steel structure box girder based on a point cloud model in a specific embodiment of the present invention;
[0055] Figure 2 This is a schematic diagram of a steel structure point cloud after precise noise reduction in a specific embodiment of the present invention;
[0056] Figure 3 This is a local point cloud simulation diagram of a steel structure node in a specific embodiment of the present invention. Detailed Implementation
[0057] To illustrate the technical solutions of the invention in more detail, specific embodiments are listed below to demonstrate the technical effects; it should be emphasized that these embodiments are used to illustrate the invention and not to limit the scope of the invention.
[0058] The present invention provides an automatic calculation method for the node center of a steel structure box girder based on a point cloud model, such as... Figure 1 As shown, it includes the following steps:
[0059] Step 1: Obtain local point cloud of box girder nodes based on design drawings;
[0060] Step 2: Local point cloud plane extraction of nodes;
[0061] Step 3: Determine the planar features of the node point cloud;
[0062] Step 4: Find the optimal node center coordinates using the least squares method.
[0063] Specifically, taking a construction site as an example, the dome of the roof structure of this project is a single-layer reticulated steel structure, with nodes welded from box-section steel beams. Before implementation, technicians used a 3D laser scanner to collect 3D laser scanning point clouds of the completed reticulated steel structure at various stations according to relevant requirements. They then used JRC 3D Constructor and Z+FLaser Control software to complete point cloud registration, and used Geomagic software to perform preprocessing work such as deleting non-object point clouds, deleting non-connected items from object point clouds, and deleting outliers, obtaining a precisely denoised steel structure point cloud P, such as... Figure 2 As shown.
[0064] After completing the above work, proceed to step 1: obtain the local point cloud of the box girder nodes based on the design drawings.
[0065] Specifically, it includes:
[0066] Step 11: Batch export the set of steel structure box girder node points M = {M1, M2, ..., M} from the design drawings. i M N} and each node M i The corresponding three-dimensional coordinates.
[0067] In this embodiment, some nodes M i The corresponding three-dimensional coordinates are as follows:
[0068] M1 -76.278038.128526.7635
[0069] M2 -74.050238.251927.3942
[0070] M3 -74.743036.206526.4615
[0071] M4 -76.874036.144526.2665
[0072] ...
[0073] M 80 -92.455847.008324.7588
[0074] ...
[0075] Step 12: Based on the 3D coordinate comparison, determine the nodes M in the point set M exported in Step 11. i If (i = 1, 2, ..., N) is an existing point in point cloud data P, then record the index value of the node in P; otherwise, add the node to point cloud data P and record its index value.
[0076] Step 13: Construct a KD-tree for the point cloud data P. Based on the index values of each node in Step 12, search for each node M in the point cloud data P. i And using it as the center, extract points within a search radius R to form a point cloud subset p. i As the corresponding steel structure box girder node M i Local point cloud.
[0077] In some embodiments, the value of the search radius R depends on the expected deformation σ of the steel structure and the length L of the box girder member. In some embodiments, the search radius R satisfies 0.25L-σ<R<0.25L+σ, so that the search radius R can better meet the actual engineering requirements.
[0078] Taking node M3 (-74.7430, 36.2065, 26.4615) from step 11 as an example, since the length L of each box girder component is approximately 2.4m and the expected deformation σ is approximately 20-40mm, the search radius R is set to 0.6m. After constructing a KD tree on the point cloud data P, all points within a radius of 0.6m are extracted with M3 as the center, thus obtaining the local point cloud corresponding to node M3. The obtained local point cloud is shown below. Figure 3 As shown.
[0079] In some embodiments, step 2 is used for local point cloud plane extraction of nodes, specifically including:
[0080] Step 21: Calculate the local point cloud p of the steel structure box girder node described in Step 13 based on principal component analysis. i The normal vectors of each point in the vector.
[0081] Step 22: Fit the node plane based on the Random Sample Consensus Algorithm (RANSAC).
[0082] Specifically, it includes:
[0083] Step 221: Randomly select a local point cloud p node i For three points with similar interior normal vectors, calculate the equation of the plane corresponding to them:
[0084] Ax + By + Cz + D = 0:
[0085] Step 222: Calculate the local point cloud p i All points N inside j (x j y j , z j The three-dimensional distance from the plane obtained in step 221:
[0086]
[0087] Step 223: Set the local point cloud p i The three-dimensional distance threshold from the interior point to the plane obtained in step 221 [d] N In this embodiment, [d] is set. N ] = 0.01;
[0088] Step 224: Compare the distances from each point to the plane calculated in step 222 with the distance thresholds set in step 223. If d Nj <[d N If the value is less than or equal to 0, then the point is considered an interior point; otherwise, it is considered an exterior point, and the current number of interior points N is recorded.
[0089] Step 225: Repeat steps 221 to 224 until the number of iterations reaches the set maximum number of iterations N. max In this embodiment, N is set. max =1000;
[0090] Step 226: Compare the number of inliers after each iteration, select the fitting plane with the most inliers as the best fitting plane, and consider the parameters corresponding to this plane as the best model parameters. Simultaneously, move the inliers from the origin cloud subset p... i Remove from the middle;
[0091] Step 227: Store the optimal model parameters (A, B, C, D) obtained in step 226 and the corresponding number of interior points N as the attribute values of the plane.
[0092] Step 23: Repeat step 22 for the remaining external points in step 22 to extract the local point cloud p of the nodes. i The remaining plane in the loop continues until the ratio of the number of remaining external points to the number of original point clouds is less than the set minimum ratio threshold p. In this embodiment, p = 8%.
[0093] Preferably, step 3 is used for the determination of planar features of the node point cloud, specifically including:
[0094] Step 31: Read and count the normal vector and quantity characteristics of the plane (plane equation is Ax+By+Cz+D=0) obtained in Step 2. The normal vector is the coefficient (A, B, C) of the equation expression of the plane, and the quantity is the number of interior points N of the plane.
[0095] For a node (82.7612, 37.9064, 23.6725), the plane equation within its local point cloud can be obtained based on step 2 as follows (the number of interior points N is in parentheses):
[0096]
[0097] The characteristics and magnitude of the normal vectors of each plane can be obtained as shown in Table 1:
[0098] Table 1. Planar features of a point cloud at a certain node in this embodiment.
[0099] Plane number normal vector quantity 1 0.2653,-0.3457,0.9001 15072 2 0.7049,-0.5045,-0.4986 28462 3 0.8995,0.3766,-0.2217 21240 4 0.1317,0.9065,0.4011 10674 5 0.7162,-0.4792,-0.5074 9152 6 0.8948,0.3918,-0.2141 18276 7 -0.1480,-0.8981,-0.4142 9268 8 -0.9494,-0.3120,0.0375 16559 9 0.0972,0.9662,0.2389 8692 10 0.2502,-0.3464,0.9041 4270
[0100] Step 32: Calculate the angle between the normal vectors of any two planes, and count the frequency of the angle between the normal vector of each plane and the normal vector of the other plane being within 90°±θ1 (considered as orthogonal). Take the plane that is orthogonal to the other plane the most times as the outer surface. The planes with the same angle as this plane that is less than θ2 or greater than θ3 are also outer surfaces. The remaining planes are the side surfaces S. i In this embodiment, θ1, θ2, and θ3 are taken as 10°, 10°, and 170°, respectively.
[0101] Calculate the angle between each pair of plane normal vectors obtained in step 31, and count the frequency of each plane normal vector being orthogonal to the normal vectors of other planes, which are 8, 2, 2, 2, 2, 2, 2, 2, 8 respectively. Based on this, it is determined that plane 1 and plane 10 are outer surfaces, and the rest are side surfaces.
[0102] Step 33: Compare the quantitative characteristics of several outer surfaces, and mark the plane with the most internal points as the true outer surface S0 of the steel structure node. The remaining outer surfaces are regarded as redundant planes or erroneous planes and removed, thereby avoiding interference from connectors installed on the steel structure node or miscellaneous objects placed there.
[0103] For the node described in step 31, planes 1 and 10 are both outer surfaces. As shown in Table 1, the number of inner points are 15072 and 4270 respectively. Therefore, plane 1 is taken as the true outer surface S0, and plane 10 is regarded as a redundant or erroneous plane and removed.
[0104] Preferably, step 4 is used to obtain the optimal node center coordinates based on the least squares method, specifically including:
[0105] Step 41: Set the center point O(x0, y0, z0) to be determined as a point on the outer surface S0 described in step 33. Then we have A0x0+B0y0+C0z0+D0=0, that is, z0=(-D0-A0x0-B0y0) / C0.
[0106] Substituting point O into the equation of plane 1 described in step 3, we get...
[0107] 0.2653x0 - 0.3457y0 + 0.9001z0 - 30.1724 = 0, that is...
[0108] z0=(30.1724-0.2653x0+0.3457y0) / 0.9001.
[0109] Step 42: Export the design node M from step 11 i coordinates (x) i y i , z i Substitute the side view S i The plane expression and the determination of A i x i +B i y i +C i z i +D i The sign of the value; if it is negative, then the value of the value on side S is positive or negative. i Find the opposite of the coefficients (A, B, C, D) of the plane equation and use them as the coefficients of the new expression for the plane; otherwise, leave them unchanged.
[0110] Substitute the node coordinates (82.7612, 37.9064, 23.6725) from step 3 into the planar expression of the side surface (i.e., planes 2-9) and determine A. i x i +B i y i +C i z i +D i The sign of the point was calculated to be negative when substituted into the expressions for planes 2-3 and 5-9. Therefore, by taking the opposite of the coefficients of the plane equations for these seven planes, the final side equations are obtained as follows:
[0111]
[0112] Step 43: Assume point O extends to any side surface S i (The equation of the plane is A) i x+B i y+C i z+D i=0, where i = 1, 2, ..., n) The three-dimensional distance is 1 / 2 of the design width of the steel structure box girder, denoted as d0. In this embodiment, d0 = 0.05, and the following relationship can be listed:
[0113]
[0114] Substituting the relation described in step 41 into the above equation, we get:
[0115]
[0116] Step 44: Combining the equations from Step 43, we obtain the following system of linear equations:
[0117]
[0118] Rewriting the above system of linear equations into matrix form and simplifying it, we can obtain:
[0119] AX0 = K
[0120] in,
[0121]
[0122] Step 45: Since the number of unknowns in the equations listed in Step 44 is less than the number of equations, the system of equations is overdetermined. Therefore, we need to find its least squares solution and define the error function:
[0123] L(x0, y0) = ||AX0 - K||2;
[0124] Step 46: By taking the derivative with respect to X0, we can find the X0 that minimizes the above error function, i.e.:
[0125] X0=(A T A) -1 A T K = (82.7877, 37.9121) T ;
[0126] Substituting the obtained (x0, y0) = (82.7877, 37.9121) into the relationship described in step 41, we obtain z0 = 23.6807, and the coordinates (x0, y0, z0) of point O are (82.7877, 37.9121, 23.6807). These coordinates are the optimal center point coordinates of the steel structure box girder node.
[0127] In summary, the automatic calculation method for the node center of steel box girder based on point cloud model provided by this invention can automatically obtain the local point cloud of the steel box girder node according to the design drawings, segment and extract the plane of the local point cloud of the node, determine the plane features of the node and remove redundant planes, and finally, based on the node plane, use the least squares method to realize the accurate position of the box girder node center. This solves the problem of difficulty in obtaining the node position of existing completed structural components, greatly improves the accuracy, efficiency and automation of node coordinate acquisition, and is beneficial to subsequent engineering applications such as curtain wall detailed design and construction quality management.
[0128] Obviously, those skilled in the art can make various modifications and variations to the invention without departing from the spirit and scope of the invention. Therefore, if these modifications and variations fall within the scope of the claims of the invention and their equivalents, the invention is also intended to include these modifications and variations.
Claims
1. A method for automatically calculating the center of a steel structure box girder joint based on a point cloud model, characterized in that, Includes the following steps: Step 1: Collect 3D laser scanning point cloud data of the completed steel structure to obtain accurately denoised steel structure point cloud data P; export the node point set of the steel structure box girder in batches based on the design drawings. M and point set M Internal nodes M i The corresponding three-dimensional coordinates; the nodes in the design drawings M i Compare with the measured steel structure point cloud data P, and extract the nodes. M i Search radius centered R Within the range of points, a local point cloud of the steel structure box girder nodes is obtained; the search radius R satisfy ,in, σ The expected deformation of the steel structure, L The length of the steel structure node component; Step 2: For the local point cloud of the node, first calculate the normal vector of each point in the local point cloud, then use the random sampling consensus algorithm to fit the node plane. After each round of fitting, remove the interior points corresponding to the plane. Repeat the fitting until the ratio of the number of remaining exterior points to the number of original point clouds is less than the set minimum ratio threshold. Extract all the fitted planes of the local point cloud of the node, and store the plane equation parameters for each fitted plane. A, B, C, D ) and the corresponding number of interior points N; Step 3: Node point cloud planar feature determination: Read the normal vectors of each fitted plane ( A,B,C ) and the number of interior points N; calculate the angle between the normal vectors of any two planes, and count the angles between the normal vectors of each plane and the normal vectors of other planes within 90° ± Within the range of frequencies, the plane that has the highest frequency of orthogonalities to other planes is taken as the outer surface, and the angle between the plane and this plane is less than [value missing]. or greater than The plane is also the outer surface, and the other planes are marked as side surfaces. S i Compare the quantitative characteristics of several outer surfaces, and take the plane with the most interior points as the true outer surface. S 0; Step 4: Determine the optimal node center coordinates using the least squares method; Step 4 includes: Step 41: Set the center point O to be determined For steel structure nodes M i The true outer surface S At a point on 0, there is ,Right now ; Step 42: Export the nodes from Step 1 3D coordinates Substitute side S i The plane expression and judgment The positive or negative value depends on the outcome; if it is negative, then the outcome depends on the side. S i plane equation coefficients ( A, B, C, D Find the opposite number and use it as the coefficient of the new expression for the plane; otherwise, leave it unchanged. Step 43: Assume point O extends to any side. S i The three-dimensional distance is half the design width of the steel structure box girder, denoted as . d If 0, then the following relationships are listed: ; Substituting the relation from step 41 into the above equation, we get: ; Step 44: Combining the equations from Step 43, we obtain the following system of linear equations: ; Simplifying the above system of linear equations and rewriting it in matrix form, we get: ; in: , , ; Step 45: Find the least squares solution to the overdetermined system of equations listed in Step 44, and define the error function: ; Step 46: [Regarding...] X 0 Find the derivative and solve for the one that minimizes the above error function. X 0 ,Right now: ; The obtained Substituting the relational expression described in step 41, we obtain z 0, obtain the coordinates of point O. These coordinates are the optimal center point coordinates of the steel structure box girder node.
2. The method for automatic calculation of node centers of steel structure box girders based on point cloud models as described in claim 1, characterized in that, Step 1 includes: Step 11: Batch export the node set of steel structure box girder from the design drawings and each node M i The corresponding three-dimensional coordinates; Step 12: Based on the 3D coordinate comparison, determine the point set exported in Step 11. M Nodes in If a node is an existing point in the steel structure point cloud data P that has been accurately denoised, then record the index value of the node in P; otherwise, add the node to the point cloud data P and record its index value. Step 13: Construct a KD-tree for the point cloud data P. Based on the index values of each node in Step 12, search for each node in the point cloud data P. M i And using it as the center, extract points within a search radius R to form a subset of the point cloud. p i As the corresponding steel structure node M i Local point cloud.
3. The method for automatic calculation of node centers of steel structure box girders based on point cloud models as described in claim 1, characterized in that, In step 1, the steel structure point cloud data P, which has been accurately denoised, is obtained by scanning the substations of a 3D laser scanner and preprocessed in point cloud processing software. The preprocessing includes at least: point cloud registration, deletion of non-object point clouds, deletion of non-connected items of object point clouds, and deletion of outliers.
4. The method for automatically calculating the node center of a steel structure box girder based on a point cloud model as described in claim 2, characterized in that, Step 2 includes: Step 21: Calculate the local point cloud of the steel structure box girder node described in Step 13 based on principal component analysis. The normal vectors of each point in the middle; Step 22: Fit the node plane based on the random sampling consensus algorithm; Step 23: Repeat step 22 for the remaining external points in step 22 to extract the local point cloud of the nodes. p i The remaining planes in the cloud are counted until the ratio of the number of remaining outliers to the number of original point clouds is less than a set minimum ratio threshold. p Stop the loop when the time is right.
5. The method for automatic calculation of node centers of steel structure box girders based on point cloud models as described in claim 4, characterized in that, Step 22 includes: Step 221: Randomly select a local point cloud node p i For three points with similar interior normal vectors, calculate the equation of the plane corresponding to them: ; Step 222: Calculate the local point cloud p i All internal points The three-dimensional distance to the plane obtained in step 221: ; Step 223: Set up the local point cloud p i Three-dimensional distance threshold from interior points to the plane obtained in step 221 ; Step 224: Compare the distances from each point to the plane calculated in step 222 with the distance thresholds set in step 223. If... If the value is true, then the point is considered an interior point; otherwise, it is considered an exterior point, and the current number of interior points N is recorded. Step 225: Repeat steps 221 to 224 until the set maximum number of iterations is reached. ; Step 226: Compare the number of inliers after each iteration, select the fitting plane with the most inliers as the best fitting plane, and consider the parameters corresponding to this plane as the best model parameters. Simultaneously, move the inliers from the origin cloud subset. p i Remove from the middle; Step 227: Apply the optimal model parameters obtained in step 226 ( A, B, C, D The plane's attribute values are stored as the number of interior points N and the corresponding number of interior points N.
6. The automatic calculation method for the node center of a steel structure box girder based on a point cloud model as described in claim 5, characterized in that, Step 3 includes: Step 31: Read and statistically analyze the normal vector and quantitative characteristics of the plane obtained in Step 2, where the normal vector is the coefficient of the equation expression of the plane. A,B,C The quantity is the number of interior points N in the plane; Step 32: Calculate the angle between the normal vectors of any two planes, and count the angles between the normal vectors of each plane and the normal vectors of other planes within 90° ± Within the range of frequencies, the plane that has the highest frequency of orthogonalities to other planes is taken as the outer surface, and the angle between the plane and this plane is less than [value missing]. or greater than The plane is also the outer surface, and the other planes are the side surfaces. S i ; Step 33: Compare the quantitative characteristics of several outer surfaces, and take the plane with the most inner points as the true outer surface of the steel structure box girder node. S 0, the remaining outer surfaces are treated as redundant or erroneous planes and removed.
7. The method for automatically calculating the node center of a steel structure box girder based on a point cloud model as described in claim 6, characterized in that, In step 32, , , The values are 10°, 10° and 170° respectively.
