A method for positioning foundation bolts in steel structure construction hoisting

Abstract

The present application relates to the technical field of three-dimensional point cloud positioning, and particularly relates to a method for positioning anchor bolts used in hoisting of steel structure construction. The method analyzes three-dimensional point clouds in a graph structure. A comprehensive edge weight value of each edge in the graph structure is defined, and the information reference strength of each edge is evaluated using the comprehensive edge weight value. Further, a center vertex of an anchor bolt point cloud region is taken as a reference point, and the longitudinal alignment degree of each cross section is analyzed based on the reference point. Through a plurality of removal strategies, the comprehensive edge weight value and the longitudinal alignment degree generated by each group removal strategy are analyzed, and the optimal graph structure is selected in combination with the removal information amount. The analysis method based on the graph structure quantifies the information reference strength of each edge through hierarchical detail processing, and further selects the optimal removal strategy to obtain the optimal graph structure for accurate positioning of anchor bolts.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional point cloud positioning technology, specifically to a method for positioning anchor bolts used in steel structure construction hoisting. Background Technology

[0002] In the process of steel structure hoisting and construction, the positioning of anchor bolts is a crucial step to ensure the safe, stable, and precise installation of the steel structure. As the link between the steel structure and the concrete foundation, the accurate position and stability of the anchor bolts directly affect the load-bearing capacity and safety of the entire building structure. There are many requirements for anchor bolt positioning in steel structure construction, including requirements for deviation from the axis, elevation, verticality, and spacing, each requiring corresponding measuring equipment. With the advancement of surveying technology and materials science, total stations have become commonly used to measure the positioning distance of anchor bolts and the parameters of the steel structure itself. Determining the position of anchor bolts using high-precision measuring equipment offers advantages over traditional anchor bolt positioning methods, such as high positioning accuracy, speed, and ease of adjustment.

[0003] Anchor bolt positioning is typically achieved through total station measurements, followed by laser rangefinder correction. However, environmental factors at the construction site and the smoothness of the measurement plane can cause interference, leading to scattering and other issues during 3D scanning. This results in incorrect coordinate points in the obtained 3D point cloud, causing significant errors in the positioning process. While linear error correction methods like Kalman filtering can be effective, these errors are random and cannot be eliminated. Using deep learning or multi-view, multi-scan comparison methods would incur excessive computational load, impacting positioning efficiency. Summary of the Invention

[0004] To address the technical problem that existing technologies cannot quickly and effectively eliminate error point clouds in the three-dimensional point cloud of anchor bolts, the present invention aims to provide a method for positioning anchor bolts during steel structure construction and hoisting. The specific technical solution adopted is as follows:

[0005] This invention proposes a method for positioning anchor bolts during steel structure construction and hoisting, the method comprising:

[0006] Obtain a 3D point cloud of anchor bolts collected by a laser scanner; identify all bolt point cloud regions in the 3D point cloud and use the center vertex of the bolt point cloud region as a reference point;

[0007] Using the reference point as the core, a graph structure is constructed with each bolt point cloud in the bolt point cloud region as a node; within the bolt point cloud region, multiple cross-sections are divided based on the height coordinate axis; for each edge in the graph structure, a first edge weight is obtained based on the edge length and the uniformity of node distribution in the cross-section; a second edge weight is obtained based on the consistency of node distribution in adjacent cross-sections; and a combined edge weight is obtained based on the first edge weight and the second edge weight.

[0008] For each cross section, the longitudinal alignment of each cross section is obtained based on the distance distribution between the reference point and each node in the cross section in the transverse plane;

[0009] Nodes and cross sections in the graph structure are removed. An objective function for each removal strategy is constructed based on the combined edge weights, vertical alignment, and removal information in the graph structure after removal. The optimal removal strategy and its corresponding optimal graph structure are selected based on the objective function. The optimal graph structure is then used for localization.

[0010] Furthermore, the method for identifying the bolt point cloud region includes:

[0011] In a 3D point cloud, convex hull detection is performed on the data points on the horizontal plane, and the point cloud contained in the obtained convex hull region constitutes the bolt point cloud region.

[0012] Furthermore, the method for identifying the reference point includes:

[0013] In each bolt point cloud region, any bolt point cloud in the top region is taken as the target point cloud to obtain the target point cloud set of all bolt point cloud regions;

[0014] In each bolt point cloud region, the maximum height difference between the target point cloud and other bolt point clouds is obtained; the average lateral difference and average height difference between the target point cloud and other bolt point clouds are obtained; the distance between the target point cloud and the centroid point of the convex hull region is negatively correlated and mapped to obtain the centrality of the target point cloud.

[0015] The uniformity of different target point cloud positions in the target point cloud set is obtained between different bolt point cloud regions;

[0016] The reference level of the target point cloud set is obtained based on the maximum height difference, average lateral difference, average height difference, centrality, and uniformity.

[0017] Multiple sets of target point clouds are selected, and the optimal set of target point clouds is selected based on the degree of reference. Each target point cloud in the optimal set of target point clouds is the reference point, and each reference point corresponds to a bolt point cloud region.

[0018] Furthermore, the method for obtaining the reference level includes:

[0019] For each target point cloud, the relative difference between the average height difference and the average lateral difference is obtained; the vertex feature degree of the target point cloud in the bolt point cloud region is obtained based on the relative difference, the centrality, and the maximum height difference.

[0020] The average distance between the target point cloud and other target point clouds in the target point cloud set is used as the position feature, and the standard deviation of the position feature is negatively correlated to obtain the uniformity.

[0021] The uniformity and vertex feature degree are positively fused to obtain the reference degree.

[0022] Furthermore, the method for obtaining the first edge weight includes:

[0023] For any cross section, obtain the standard deviation of the lengths of all sides contained in the cross section, perform negative correlation mapping on the standard deviations of the lengths and normalize them to obtain the uniformity of the distribution;

[0024] For each edge, the edge length is negatively correlated and normalized, then multiplied by the distribution uniformity to obtain the first edge weight.

[0025] Furthermore, the method for obtaining the second edge weight includes:

[0026] Set the weight of the second side of each edge in the topmost cross section to 1;

[0027] For each edge in each cross-section except the top-level cross-section, the nearest vertex of the two vertices corresponding to each edge in the upper-level cross-section is taken as the reference vertex; the edge between the vertex and the reference vertex is taken as the comparison edge; the included angle between the two comparison edges is obtained; the slope difference between the average slope of all edges in the cross-section and the average slope of all edges in the upper-level cross-section is obtained; the slope difference is used as the weight of the included angle to obtain the longitudinal distribution inconsistency of each edge; the longitudinal distribution inconsistency is negatively correlated and normalized to obtain the second edge weight of each edge.

[0028] Furthermore, the combined edge weight is the product of the first edge weight and the second edge weight.

[0029] Furthermore, the method for obtaining the vertical alignment includes:

[0030] For each cross section, the standard deviation of the distance between each node in the cross section and the reference point on the horizontal plane is obtained. The standard deviation of the distance is negatively correlated and normalized to obtain the vertical alignment.

[0031] Furthermore, the method for obtaining the objective function includes:

[0032] In the graph structure after the removal operation, the average comprehensive edge weight of all edges and the average longitudinal alignment of all cross-sections are calculated. The number of point clouds to be removed is negatively correlated and normalized to obtain a suppression term. The sum of the suppression term, the average comprehensive edge weight, and the average longitudinal alignment is used as the objective function value of the objective function.

[0033] Furthermore, the optimal removal strategy is the removal strategy corresponding to the maximum objective function value.

[0034] The present invention has the following beneficial effects:

[0035] To ensure the rapid and effective removal of error point clouds from the 3D point cloud of anchor bolts, this invention employs a graph structure approach for 3D point cloud analysis. Error point clouds are often isolated, leading to deformation in the corresponding local areas of the detection results. This deformation causes the bolt point cloud region to deviate from a uniform cylindrical structure. Therefore, this invention uses cross-sectional analysis to perform detailed analysis of the bolt point cloud region. In a cross-section of a normal point cloud, the connecting edges between point clouds should be short and uniformly distributed, and the distribution of edges between adjacent cross-sections should also be consistent. Based on this, this invention defines a comprehensive edge weight for each edge in the graph structure and uses this comprehensive edge weight to evaluate the information reference strength of each edge. Furthermore, the central vertex of the bolt point cloud region is used as a reference point to analyze the longitudinal alignment of each cross-section. A higher longitudinal alignment indicates more complete information and fewer error points within the cross-section. Therefore, this invention uses multiple removal strategies, analyzes the comprehensive edge weight and longitudinal alignment generated by each strategy, and combines this with the amount of removed information to select the optimal graph structure. In other words, the optimal graph structure represents the result that removes the most information, eliminates the most error points, and retains the most valid points. Therefore, the optimal graph structure can be used to accurately locate anchor bolts. This invention, based on a graph structure analysis method, quantifies the information reference strength of each edge through layered detail processing, thereby selecting the optimal removal strategy and obtaining the optimal graph structure for accurate anchor bolt positioning. Attached Figure Description

[0036] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the 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.

[0037] Figure 1This is a flowchart illustrating a method for positioning anchor bolts during steel structure construction hoisting, provided in one embodiment of the present invention.

[0038] Figure 2 This is a schematic diagram illustrating the process of acquiring three-dimensional point clouds of anchor bolts according to an embodiment of the present invention.

[0039] Figure 3 This is a schematic diagram of a cross-sectional division provided in one embodiment of the present invention. Detailed Implementation

[0040] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a method for positioning anchor bolts for steel structure construction hoisting according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0041] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0042] The following describes in detail, with reference to the accompanying drawings, a specific scheme for positioning anchor bolts for steel structure construction hoisting provided by the present invention.

[0043] Please see Figure 1 The diagram illustrates a flowchart of a method for positioning anchor bolts during steel structure construction hoisting, according to an embodiment of the present invention. The method includes:

[0044] Step S1: Obtain the three-dimensional point cloud of the anchor bolts collected by the laser scanner; identify all bolt point cloud regions in the three-dimensional point cloud, and use the center vertex of the bolt point cloud region as the reference point.

[0045] The positioning method for anchor bolts is primarily to ensure that steel structure components (such as columns and beams) can be accurately and stably installed in predetermined positions. Laser positioning can determine the center position of bolts with very high precision, typically down to the millimeter level. However, because multiple anchor bolt positions need to be located, total stations use laser scanning positioning instead of single-laser-beam positioning. Please refer to [link to relevant documentation]. Figure 2 This illustrates a schematic diagram of a process for acquiring a three-dimensional point cloud of anchor bolts according to an embodiment of the present invention. Figure 2As shown, a reference point (such as a pre-marked control point or baseline) is determined using a total station. The laser equipment is then set on the reference point, and a stable location with a large line-of-sight is selected for installation to ensure that the laser scanning beam can cover the entire installation area. The three-dimensional coordinate data (X, Y, Z) of the installation area can be obtained using the total station. These data represent the specific spatial positions of all objects, including the bolt. In this embodiment of the invention, the height axis is the Z-axis, and the transverse plane is the plane formed by the X and Y axes. During scanning, it is necessary to ensure that the Z-axis is parallel to the bolt axis, and the X and Y plane is parallel to the bolt cross-section. It should be noted that after obtaining the three-dimensional point cloud through scanning, the subsequent point cloud distance calculation, projection calculation, and other methods are all well-known techniques to those skilled in the art and will not be elaborated upon here.

[0046] It should be noted that, as Figure 2 As shown, anchor bolts are typically located in specific areas or grid points, such as on the surface of concrete embedded parts. During calibration, the approximate location range of the anchor bolts is predetermined. Therefore, the laser ranging device sets multiple measurement points and re-calibration point coordinates during measurement, filtering out positioning points within the preset range. This preset range constitutes the 3D point cloud analyzed in this embodiment of the invention, eliminating interference from background information or other irrelevant data. It should be noted that the preset range can be comprehensively set based on parameters such as the anchor bolt size, image field of view parameters, and the distance between the camera and the anchor bolt. For example, during scanning, the target area can be placed at the center of the scanner's scanning range; in subsequent analysis, only the point cloud at the center position needs to be extracted. This is a technique well-known to those skilled in the art and will not be elaborated upon here.

[0047] In the 3D point cloud, all bolt point cloud regions can be further identified, such as... Figure 2 As shown, in the scenarios described in this embodiment of the invention, there are often multiple anchor bolts, thus each bolt point cloud region can be identified for targeted analysis. Furthermore, this embodiment of the invention uses the central vertex of the bolt point cloud region as a reference point for subsequent graph structure construction and vertical analysis.

[0048] Preferably, in this embodiment of the invention, since the bolts protrude in the Z-axis direction on the concrete embedded parts, even if there are errors in the coordinate point array measured by the laser, the approximate position of all anchor bolts can be obtained. Furthermore, since the bolts have obvious shape characteristics, this embodiment of the invention performs convex hull detection on the projection results of all point clouds on the horizontal plane of the three-dimensional point cloud. Since the top view of the bolt is obviously a circular or symmetrical polygon, the result of the point cloud contained in the obtained convex hull area in the three-dimensional coordinate system can be directly used as the bolt point cloud area.

[0049] It should be noted that although the bolt point cloud region can be determined, the hoisting of steel structure components generally occurs in open or semi-open environments. Laser positioning methods are easily affected by environmental factors, such as light interference, dust or smoke at the construction site, wind speed changes, and irregular, rough, or obstructed surfaces of the anchor bolts, which can cause unstable laser beam reflection. These factors can all lead to deviations in the measurement results. Directly using the center point of the top horizontal region of the obtained bolt point cloud region as a reference point may cause errors. Therefore, in a preferred embodiment of the present invention, the center point of the top horizontal region of the bolt point cloud region is not directly used as the central vertex. Instead, a more accurate central vertex is determined as the reference point by analyzing the positional relationship between each point cloud on the top horizontal region and other bolt point clouds. Specifically, this includes:

[0050] (1) In each bolt point cloud region, any bolt point cloud in the top region is taken as the target point cloud to obtain the target point cloud set of all bolt point cloud regions. For example... Figure 2 As shown, if there are four anchor bolts, four bolt point cloud regions will be obtained, resulting in four target point cloud sets. It should be noted that, to avoid errors, the top region here is not directly set as the top horizontal cross-section of the bolt point cloud region. Instead, a local region corresponding to the top layer in the 3D point cloud coordinate system is obtained with a width of 1 cm in the real world. That is, the position after coordinate transformation is determined on the Z-axis, and the XY plane corresponding to this position is used as the dividing plane to determine the top region.

[0051] (2) In each bolt point cloud region, obtain the maximum height difference between the target point cloud and other bolt point clouds; obtain the average lateral difference and average height difference between the target point cloud and other bolt point clouds; perform a negative correlation mapping on the distance between the target point cloud and the centroid of the convex hull region to obtain the centrality of the target point cloud. Among them, in a single bolt point cloud region, the larger the maximum height difference, the more likely the target point cloud is to be the top layer of the bolt; the larger the average height difference is relative to the average lateral difference, the more significant the target point cloud is in height compared to other bolt point clouds, and the more likely it is to be at the center of the overall bolt point cloud region in the lateral direction; the greater the centrality, the closer the target point cloud is to the center of the bolt point cloud data section, and the more likely the target point cloud is to belong to the central vertex.

[0052] It should be noted that, in this embodiment of the invention, the negative correlation mapping method adopts the exponential function mapping method, which is expressed by the formula: ; where exp represents an exponential function with base to the natural constant. Let be the distance between the p-th target point cloud and the centroid of the convex hull region. This negative correlation mapping limits the result value range to between 0 and 1, facilitating computation.

[0053] (3) Obtain the uniformity of different target point cloud positions in the target point cloud set between different bolt point cloud regions. For example... Figure 2 As shown, the distribution of different bolts should exhibit regularity. That is, the greater the uniformity, the more suitable the selected target point cloud in the current target point cloud set is as a whole, which conforms to the laws of the real scene.

[0054] (4) The reference degree of the target point cloud set is obtained based on the maximum height difference, average lateral difference, average height difference, centrality, and uniformity. In summary, the reference degree can be constructed based on the meaning of each feature. The reference degree corresponds to a target point cloud set and represents the correctness of the selection of the target point cloud.

[0055] (5) Select multiple sets of target point clouds, and filter out the optimal set of target point clouds according to the degree of reference. Each target point cloud in the optimal set of target point clouds is the reference point, and each reference point corresponds to a bolt point cloud region.

[0056] Furthermore, in a specific implementation of this invention, the method for obtaining the aforementioned level of reference may include:

[0057] For each target point cloud, the relative difference between the average height difference and the average lateral difference is obtained. Based on the relative difference, the centrality, and the maximum height difference, the vertex feature degree of the target point cloud within its bolt point cloud region is obtained. A larger relative difference indicates that the target point cloud's vertical difference from other bolt point clouds within its bolt point cloud region is greater than its horizontal difference, indicating that the target point cloud is closer to the top. A larger centrality indicates that the target point cloud is closer to the top centroid. Similarly, a larger maximum height difference also indicates that the target point cloud is closer to the top. Combining these three features, the vertex feature degree of the target point cloud within its bolt point cloud region can be obtained; that is, a larger vertex feature degree indicates that the target point cloud is closer to the central vertex.

[0058] The average distance between the target point cloud and other target point clouds in the target point cloud set is used as the positional feature. The standard deviation of the positional feature is negatively correlated to obtain the uniformity. That is, the larger the standard deviation, the more uneven the distribution of the selected target point clouds. Therefore, after negative correlation mapping, uniformity can be obtained. The greater the uniformity, the more the selected target point cloud meets the actual needs.

[0059] The uniformity and vertex feature degree are positively fused to obtain the reference degree.

[0060] As a specific example, in one implementation of this invention, relative differences are implemented using ratios, the negative correlation mapping method for centrality is in reciprocal form, and uniformity is set as the negative of the standard deviation of location features. The final reference level is expressed by the formula:

[0061] ;in For the first The reference level of a set of target point clouds, where N is the number of target point clouds. Let p be the average height difference of the target point cloud. Let p be the average lateral difference of the target point cloud. The maximum height difference between the p-th target point cloud and other bolt point clouds; For the first The positional characteristics of each target point cloud in a set of target point clouds. is the standard deviation of the location feature. Indicating the uniformity, This indicates the relative difference. Indicates centrality, Indicates the degree of vertex feature.

[0062] In the above formula, the averaging method is first used to fuse the vertex feature levels corresponding to the target point clouds of all bolt point cloud regions. Then, a summation method is used to achieve a positive fusion of uniformity and average vertex feature levels. The purpose of adding a positive integer 1 to the denominator is to prevent the denominator from being 0, that is, if If the value is 0, then the negative correlation mapping result is... The maximum value is 1. It should be noted that in the above formula, the parameters involved only involve numerical values ​​in the calculation, and the final result does not need to be standardized. This is because the higher the reference level, the more suitable the target point cloud set is. Therefore, it is only necessary to select the reference level with the maximum value from multiple target point cloud sets as the optimal target point cloud set.

[0063] Step S2: Construct a graph structure with the reference point as the core and each bolt point cloud in the bolt point cloud region as a node; divide the bolt point cloud region into multiple cross sections with the height coordinate axis as the reference; for each edge in the graph structure, obtain the first edge weight based on the edge length and the uniformity of node distribution in the cross section; obtain the second edge weight based on the consistency of node distribution in adjacent cross sections; obtain the combined edge weight based on the first edge weight and the second edge weight.

[0064] To enable effective detailed analysis of the bolt point cloud region, this embodiment of the invention constructs a graph structure using a reference point and each bolt point cloud in the region as a node. Adjacent bolt point clouds in the graph structure form edges, and the distance and distribution of these edges reflect the overall uniformity of the bolt point cloud distribution. A standard bolt point cloud region should exhibit a uniform distribution of bolt point clouds, a regular edge distribution, and relatively short edge lengths. However, bolt point cloud regions affected by the construction site environment may experience structural deformation or missing information, resulting in irregular edge distributions or some edges being excessively long.

[0065] To further effectively analyze the information in the graph structure, this embodiment of the invention divides the graph structure into multiple cross-sections along the height axis. If the current bolt point cloud region is a normal region without deformation, short and evenly distributed edges will be generated in the cross-section, and the distribution between adjacent upper and lower cross-sections will be relatively neat. The entire structure will be aligned along the axis with the reference point as the vertex, and the distribution of edges on adjacent cross-sections will also remain consistent. Conversely, if the current bolt point cloud region is deformed due to noise, isolated edges will be generated, the number of points in the cross-section will decrease, the distribution of edges will become chaotic, and the distribution of edges will not be consistent between the deformed cross-section and adjacent cross-sections due to the influence of random noise. Based on this, this embodiment of the invention analyzes each edge and each cross-section in the graph structure. For each edge in the graph structure, a first edge weight is obtained based on the edge length and the uniformity of node distribution in the cross-section. That is, the smaller the edge length and the greater the uniformity of node distribution in the cross-section, the more the edge and cross-section belong to a normal point cloud region, and the larger the first edge weight. Further, the second edge weight of each edge is obtained based on the consistency of node distribution in adjacent cross-sections. For an edge, if there is a consistent distribution with the corresponding edge on the adjacent cross-section, it indicates that the bolt point cloud region is locally uniformly distributed and belongs to a normal local region of the bolt point cloud, thus the second edge weight is larger. The final comprehensive edge weight is obtained by fusing the first and second edge weights. The larger the comprehensive edge weight, the more the corresponding edge belongs to a normal bolt point cloud region.

[0066] In this embodiment of the invention, each adjacent node can be extended from the reference point to obtain a graph structure. Specifically, a tree structure or other common graph structure construction form can be adopted. These are technical means well known to those skilled in the art and will not be described in detail here.

[0067] It should be noted that you should refer to [link / reference]. Figure 3This illustration shows a schematic diagram of cross-sectional segmentation provided by an embodiment of the present invention. The present invention sets the height range of the cross-section on the Z-axis to 1 cm in the real world. Starting from the top layer of the reference point, the entire bolt point cloud region is progressively segmented to obtain the final cross-sectional segmentation result. Each cross-section is parallel to the XY plane in the three-dimensional coordinate system. That is, the obtained cross-section can be regarded as a three-dimensional structure with a relatively small height. To avoid interference from noise points, this embodiment of the present invention removes cross-sections containing only one point cloud. The final obtained cross-section includes more than two point clouds, and the number of point clouds contained in each cross-section may differ. It should be noted that... Figure 3 For ease of illustration, a small number of point clouds are presented to show the details of the cross-sectional division. In actual practice, a cross-section may include more point clouds, which will not be elaborated here.

[0068] Preferably, in this embodiment of the invention, the method for obtaining the first edge weight includes:

[0069] For any cross-section, the standard deviation of the lengths of all sides contained in the cross-section is obtained. This standard deviation is then negatively correlated and normalized to obtain the distribution uniformity. A larger standard deviation indicates a more uneven distribution of side lengths in the cross-section, with significantly longer sides. These sides may correspond to locations where deformation has occurred; therefore, negative correlation mapping and normalization are performed to obtain the distribution uniformity. In this embodiment of the invention, an exponential function mapping method with the natural constant as the base can also be used to achieve negative correlation mapping and normalization, as described in the above embodiments and will not be repeated here.

[0070] For each edge, the edge length is negatively correlated and normalized, then multiplied by the distribution uniformity to obtain the first edge weight. That is, the edge length distribution within the corresponding cross-section is used as the weight; the smaller the edge length, the more normal the cross-section, and therefore the larger its corresponding first edge weight should be. It should be noted that the negative correlation mapping and normalization here can still be performed using the exponential function mapping method described above, and the final first edge weight will also be data with values ​​between 0 and 1.

[0071] Preferably, in this embodiment of the invention, the method for obtaining the second edge weight includes:

[0072] Since the second side weight represents the consistency of the side distribution between adjacent cross sections, this embodiment of the invention analyzes the side distribution relationship between the upper cross section and the current cross section for a cross section. However, the top cross section does not have an upper cross section, so the second side weight of each side in the top cross section can be set to 1.

[0073] For each edge in each cross-section except the top-level cross-section, the nearest vertex of the two vertices corresponding to each edge in the upper-level cross-section is taken as the reference vertex; the edge between the vertex and the reference vertex is taken as the comparison edge; the included angle between the two comparison edges is obtained. The smaller the included angle, the stronger the consistency of the distribution of the edge to be analyzed between the upper and lower cross-sections, indicating that the edge to be analyzed has a one-to-one vertex distribution in the upper-level cross-section.

[0074] The average slope of all edges in the cross-section is obtained, and the slope difference between it and the average slope of all edges in the upper cross-section is calculated. Similar to the distribution uniformity in the first edge weight, the slope difference in the second edge weight is also used as a confidence weight. The slope difference reflects the difference in the overall trend of the edges between the two cross-sections. The larger the slope difference, the greater the distribution difference of the edges between the upper and lower cross-sections, and the smaller the corresponding second edge weight should be. It should be noted that the slope is a concept in two-dimensional space. In the embodiments of this invention, the slopes obtained are all considered to be the slopes of the projection obtained by mapping the edges of the three-dimensional point cloud in the XY plane.

[0075] The slope difference is used as a weight for the included angle to obtain the longitudinal distribution inconsistency of each side. In this embodiment of the invention, the slope difference is the absolute value of the difference between two average slopes. This slope difference is directly multiplied by the value of the included angle to obtain a dimensionless value as the longitudinal distribution inconsistency. The greater the longitudinal distribution inconsistency, the more inconsistent the distribution of the sides between the cross section where the corresponding side is located and the adjacent cross section above, and the smaller the weight of the second side should be.

[0076] The inconsistencies in the vertical distribution are negatively correlated and normalized to obtain the weight of the second side of each edge. Similarly, in this embodiment of the invention, the negative correlation mapping and normalization can be implemented using the aforementioned exponential function. In other specific implementations of this invention, based on statistical data, the inconsistencies in the vertical distribution can be normalized first using range standardization, and then the negative correlation mapping can be achieved by subtracting the normalized result from the positive integer 1. The specific processing methods are basic mathematical techniques well known to those skilled in the art, and will not be elaborated or limited here.

[0077] In this embodiment of the invention, since both the first edge weight and the second edge weight represent that the larger the value, the more the corresponding edge belongs to the normal bolt area, the product of the first edge weight and the second edge weight can be directly used as the comprehensive edge weight.

[0078] Step S3: For each cross section, obtain the longitudinal alignment of each cross section based on the distance distribution between the reference point and each node in the cross section on the transverse plane.

[0079] The embodiments of the present invention further consider that, in step S2, the bolt point cloud region has been divided into multiple cross-sections, such as... Figure 3 As shown, if the bolt point cloud region is deformed, the cross-section will exhibit significant distortion in the longitudinal region, demonstrating an inability to align along the vertical axis corresponding to the reference point. Therefore, to further quantify the normality of each cross-section, this embodiment of the invention obtains the longitudinal alignment of each cross-section based on the distance distribution between the reference point and each node in the cross-section on the transverse plane. If the cross-section is a normal cross-section that can be aligned and does not suffer from missing or deformed information, the distance distribution should be uniform without significant chaotic distance changes, resulting in a higher longitudinal alignment. Conversely, if the cross-section is deformed or has missing information, the distance distribution will be chaotic, leading to a lower longitudinal alignment.

[0080] Preferably, in this embodiment of the invention, the method for obtaining the longitudinal alignment includes:

[0081] For each cross section, the standard deviation of the distance between each node in the cross section and the reference point on the transverse plane is obtained. This standard deviation is then negatively correlated and normalized to obtain the longitudinal alignment. The transverse plane is the XY plane that has already been aligned during the scanning process.

[0082] Step S4: Remove nodes and cross sections from the graph structure. Construct an objective function for each removal strategy based on the combined edge weights, vertical alignment, and removal information in the graph structure after removal. Select the optimal removal strategy and its corresponding optimal graph structure based on the objective function. Use the optimal graph structure for localization.

[0083] This invention aims to eliminate noise in 3D point clouds to avoid errors in bolt positioning. Therefore, for the scanned bolt point cloud region, this invention removes nodes and cross-sections from the resulting graph structure. The goal is to remove more noise and retain more effective point cloud information with less material removal. Multiple removal strategies can be randomly generated, and an objective function for each strategy is constructed based on the combined edge weights, vertical alignment, and amount of removed information in the resulting graph structure. Specifically, the larger the combined edge weights of the remaining edges, the greater the vertical alignment of the remaining cross-sections, and the less information removed, the better the removal strategy, and the larger the corresponding objective function should be. Therefore, the optimal removal strategy and its corresponding optimal graph structure can be selected through the objective function, allowing for accurate bolt positioning.

[0084] Preferably, in this embodiment of the invention, the method for obtaining the objective function includes:

[0085] In the graph structure after the removal operation, the average comprehensive edge weight of all edges and the average longitudinal alignment of all cross-sections are calculated. The number of removed point clouds is negatively correlated and normalized to obtain a suppression term. The sum of the suppression term, the average comprehensive edge weight, and the average longitudinal alignment is used as the objective function value of the objective function. The suppression term aims to avoid information distortion due to excessive removal. Therefore, the number of removed point clouds is used as the amount of removed information for negative correlation mapping and normalization. It should be noted that the removal of cross-sections can also be regarded as the removal of a portion of the point cloud. Finally, the total number of removed point clouds is calculated to obtain the suppression term. The method for negative correlation mapping and normalization of the suppression term has been described in the above embodiment and will not be repeated here.

[0086] The optimal removal strategy in this embodiment of the invention is the removal strategy corresponding to the maximum objective function value. After obtaining the optimal graph structure, re-interpolation and fitting can be performed based on the remaining point cloud with strong reference information to finally obtain the complete bolt point cloud region, thereby realizing the process of bolt edge localization and center localization. The specific method is a well-known technique to those skilled in the art and will not be described in detail here.

[0087] In summary, this invention employs a graph structure approach to analyze 3D point clouds. A comprehensive edge weight is defined for each edge in the graph structure, and this weight is used to evaluate the information reference strength of each edge. Furthermore, the central vertex of the bolt point cloud region is used as a reference point, and the longitudinal alignment of each cross-section is analyzed based on this reference point. Multiple removal strategies are employed, and the comprehensive edge weight and longitudinal alignment generated by each strategy are analyzed. Combining the amount of removed information, the optimal graph structure is selected. This invention's graph structure-based analysis method quantifies the information reference strength of each edge through layered detail processing, thereby selecting the optimal removal strategy and obtaining the optimal graph structure for accurate anchor bolt positioning.

[0088] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0089] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for positioning anchor bolts during steel structure construction and hoisting, characterized in that, The method includes: Obtain a 3D point cloud of anchor bolts collected by a laser scanner; identify all bolt point cloud regions in the 3D point cloud and use the center vertex of the bolt point cloud region as a reference point; Using the reference point as the core, a graph structure is constructed with each bolt point cloud in the bolt point cloud region as a node; within the bolt point cloud region, multiple cross-sections are divided based on the height coordinate axis; for each edge in the graph structure, a first edge weight is obtained based on the edge length and the uniformity of node distribution in the cross-section; a second edge weight is obtained based on the consistency of node distribution in adjacent cross-sections; and a combined edge weight is obtained based on the first edge weight and the second edge weight. For each cross section, the longitudinal alignment of each cross section is obtained based on the distance distribution between the reference point and each node in the cross section in the transverse plane; Nodes and cross sections in the graph structure are removed. An objective function for each removal strategy is constructed based on the combined edge weights, vertical alignment, and removal information in the graph structure after removal. The optimal removal strategy and its corresponding optimal graph structure are selected based on the objective function. The optimal graph structure is then used for localization.

2. The method for positioning anchor bolts for steel structure construction hoisting according to claim 1, characterized in that, The method for identifying the bolt point cloud region includes: In a 3D point cloud, convex hull detection is performed on the data points on the horizontal plane, and the point cloud contained in the obtained convex hull region constitutes the bolt point cloud region.

3. The method for positioning anchor bolts for steel structure construction hoisting according to claim 2, characterized in that, The method for identifying the reference point includes: In each bolt point cloud region, any bolt point cloud in the top region is taken as the target point cloud to obtain the target point cloud set of all bolt point cloud regions; In each bolt point cloud region, the maximum height difference between the target point cloud and other bolt point clouds is obtained; the average lateral difference and average height difference between the target point cloud and other bolt point clouds are obtained; the distance between the target point cloud and the centroid point of the convex hull region is negatively correlated and mapped to obtain the centrality of the target point cloud. The uniformity of different target point cloud positions in the target point cloud set is obtained between different bolt point cloud regions; The reference level of the target point cloud set is obtained based on the maximum height difference, average lateral difference, average height difference, centrality, and uniformity. Multiple sets of target point clouds are selected, and the optimal set of target point clouds is selected based on the degree of reference. Each target point cloud in the optimal set of target point clouds is the reference point, and each reference point corresponds to a bolt point cloud region.

4. The method for positioning anchor bolts for steel structure construction hoisting according to claim 3, characterized in that, The method for obtaining the reference level includes: For each target point cloud, the relative difference between the average height difference and the average lateral difference is obtained; based on the relative difference, the centrality, and the maximum height difference, the vertex feature degree of the target point cloud in the bolt point cloud region is obtained; The average distance between the target point cloud and other target point clouds in the target point cloud set is used as the position feature, and the standard deviation of the position feature is negatively correlated to obtain the uniformity. The uniformity and vertex feature degree are positively fused to obtain the reference degree.

5. The method for positioning anchor bolts for steel structure construction hoisting according to claim 1, characterized in that, The method for obtaining the first edge weight includes: For any cross section, obtain the standard deviation of the lengths of all sides contained in the cross section, perform negative correlation mapping on the standard deviations of the lengths and normalize them to obtain the uniformity of the distribution; For each edge, the edge length is negatively correlated and normalized, then multiplied by the distribution uniformity to obtain the first edge weight.

6. The method for positioning anchor bolts for steel structure construction hoisting according to claim 1, characterized in that, The methods for obtaining the second edge weight include: Set the weight of the second side of each edge in the topmost cross section to 1; For each edge in each cross-section except the top-level cross-section, the nearest vertex of the two vertices corresponding to each edge in the upper-level cross-section is taken as the reference vertex; the edge between the vertex and the reference vertex is taken as the comparison edge; the included angle between the two comparison edges is obtained; the slope difference between the average slope of all edges in the cross-section and the average slope of all edges in the upper-level cross-section is obtained; the slope difference is used as the weight of the included angle to obtain the longitudinal distribution inconsistency of each edge; the longitudinal distribution inconsistency is negatively correlated and normalized to obtain the second edge weight of each edge.

7. The method for positioning anchor bolts for steel structure construction hoisting according to claim 1, characterized in that, The combined edge weight is the product of the first edge weight and the second edge weight.

8. The method for positioning anchor bolts for steel structure construction hoisting according to claim 1, characterized in that, The method for obtaining the vertical alignment includes: For each cross section, the standard deviation of the distance between each node in the cross section and the reference point on the horizontal plane is obtained. The standard deviation of the distance is negatively correlated and normalized to obtain the vertical alignment.

9. A method for positioning anchor bolts for steel structure construction hoisting according to claim 1, characterized in that, The method for obtaining the objective function includes: In the graph structure after the removal operation, the average comprehensive edge weight of all edges and the average longitudinal alignment of all cross-sections are calculated. The number of point clouds to be removed is negatively correlated and normalized to obtain a suppression term. The sum of the suppression term, the average comprehensive edge weight, and the average longitudinal alignment is used as the objective function value of the objective function.

10. A method for positioning anchor bolts for steel structure construction hoisting according to claim 9, characterized in that, The optimal removal strategy is the removal strategy corresponding to the maximum objective function value.

Images