A quality control method and system for small box beams

Abstract

The present invention provides a quality control method and system for small box girders, comprising the following steps: constructing a BIM model for the design of a small box girder according to the bridge design drawings; converting the BIM design model of the small box girder into a discrete point cloud; obtaining a point cloud of a prefabrication site template and preprocessing the point cloud of the prefabrication site template; aligning the point cloud model with the BIM model; selecting control points on the aligned point cloud model as a benchmark for external mold control layout, and performing calculation and evaluation on the on-site layout data. The present invention provides a quality control method and system for small box girders, which uses BIM model interpolation to generate point clouds, combines a point cloud-point cloud alignment algorithm to achieve high-precision alignment, and evaluates the alignment effect based on the root mean square error. Closed-loop quality control from design to construction reduces manual intervention and can achieve full process digitization. Through algorithm fusion, alignment errors are achieved to meet engineering requirements and effectively improve accuracy. It solves the pain points of low accuracy and poor efficiency of traditional methods.

Description

Technical Field

[0001] The present invention belongs to the technical field of prefabricated box girders, and in particular relates to a quality control method and system for small box girders. Background Art

[0002] Traditional bridge construction typically uses on-site cast-in-place concrete, which has limitations in terms of construction time, cost, environmental impact, and quality control. With technological advancements, modern bridge construction is increasingly adopting precast box girders and virtual installation techniques to improve construction efficiency, reduce costs, minimize environmental impact, and ensure higher quality and safety.

[0003] Currently, 71.72% of my country's highway bridges utilize precast prestressed concrete small box girder structures. Research on prefabrication control technology for precast small box girders is of great significance to the quality of modern bridges. In recent years, composite small box girder bridges have played an increasingly important role in bridge construction due to their exceptional safety, convenience, and durability. This structure is not only adaptable to a variety of complex terrains and environmental conditions, but also exhibits significant advantages during construction and maintenance, making it a preferred choice for modern bridge construction. Precast box girders are fabricated at a prefabrication site and then transported to the site for installation, significantly shortening the construction cycle. On-site box girder prefabrication sites require rigorous quality control to ensure that the precast box girders meet design specifications and standards during fabrication. The flatness, smoothness, and lack of deformation of the prefabrication site's formwork surface, as well as the control technology for formwork installation accuracy, determine whether the precast box girder's appearance meets the design dimensional requirements. Currently, prefabrication sites primarily rely on measuring tools such as box rulers to control the closing of formwork at key locations, lacking overall control. Therefore, it is crucial to develop a quality control method and system for small box girders to achieve digital quality control of box girder prefabrication. Summary of the Invention

[0004] The purpose of the present invention is to provide a quality control method and system for small box girders, aiming to solve the technical problems of low precision and poor efficiency in the prior art.

[0005] To achieve the above object, the technical solution adopted by the present invention is to provide a quality control method for small box beams, comprising:

[0006] According to the bridge design drawings, a BIM model of the small box girder design is constructed; then the BIM design model of the small box girder is converted into a discrete point cloud;

[0007] Obtaining the prefabrication site template point cloud and preprocessing the prefabrication site template point cloud;

[0008] The point cloud model is registered with the BIM model;

[0009] Control points are selected on the registered point cloud model as the benchmark for external model control layout, and the on-site layout data is calculated and evaluated.

[0010] Preferably, the step of constructing a small box girder design BIM model according to the bridge design drawings; and then converting the small box girder BIM design model into a discrete point cloud comprises:

[0011] By creating a structure to store the model information of the small box girder design BIM model;

[0012] Read the triangular facet data of the small box girder BIM model and determine the number of faces;

[0013] Get the point cloud corresponding to the BIM geometric model.

[0014] Preferably, obtaining the point cloud corresponding to the BIM geometric model includes:

[0015] Create a function to traverse the indices of all facets and perform equidistant interpolation on each triangle facet;

[0016] The output is a TXT file containing a virtual point cloud, which is merged with the measured point cloud for subsequent registration.

[0017] Preferably, the obtaining of the prefabricated site template point cloud and preprocessing the prefabricated site template point cloud includes:

[0018] Collect field data;

[0019] Preprocessing the prefabricated site three-dimensional point cloud to obtain a three-dimensional point cloud model; wherein the preprocessing includes: one or more of: denoising, downsampling, and splicing;

[0020] First perform plane fitting and then calculate the surface flatness.

[0021] Preferably, the collecting of on-site data includes:

[0022] Conduct exploration and site selection and control point layout around the prefabrication site;

[0023] Use a scanner to scan on-site and collect the data required for the 3D point cloud of the prefabrication site;

[0024] Perform fill scan on the mirror reflection area to reduce data loss.

[0025] Preferably, the point cloud model is registered with the BIM model, comprising:

[0026] Initial alignment is achieved through coarse registration, wherein the coarse registration is performed using the SAC-IA algorithm or / and the 4PCS algorithm;

[0027] Refined alignment is achieved through fine registration, wherein fine registration is performed using an ICP algorithm and / or a key point-based registration algorithm;

[0028] The root mean square error was used to evaluate the registration results.

[0029] Preferably, achieving initial alignment through coarse registration includes:

[0030] Input two site clouds: the site cloud to be registered P and the reference site cloud Q; calculate the normal information of each point cloud: N P 、N Q ;

[0031] Calculate the fast feature histogram FPFH of each point cloud: F P 、F Q ;

[0032] Randomly sample m points P1, P2, P3, ..., P in the cloud of sites to be registered. m ; and find the m points Q1, Q2, Q3, ..., Q that are closest to the point feature fast histogram of the above m points in the base station site cloud Q m ; Then use these m groups of point pairs to calculate the transformation matrix T, and apply the matrix T to the site cloud to be registered: P ' =P*T; find the corresponding point;

[0033] Calculate P and Q ' The registration error ∈ between them, store the corresponding transformation matrix T;

[0034] Determine whether the number of iterations i_iter is less than the set number MaxIterCounts, and if not, exit the loop;

[0035] Compare all groups of registration errors in the loop process, screen out a group of minimum registration errors, output the corresponding transformation matrix T, and apply it to the original site cloud P to complete the initial registration.

[0036] Preferably, achieving refined alignment through precise registration includes:

[0037] Initialization: P0 = P, k=0;

[0038] Cloud P of the site to be registered k In the benchmark site cloud Q, the neighborhood search method is used to search for the nearest neighbor points corresponding to each point in the point cloud P' to form neighboring point pairs.

[0039] Use all the nearest point pairs to calculate the parameters corresponding to the three-dimensional transformation matrix

[0040] R k 、T k Apply it to the point cloud P to be registered to obtain a new point cloud to be registered

[0041] Compare the difference between the current error value and the previous error value to see if it is less than the given threshold ε:d k -d k+1 <ε, if it is satisfied, the iteration stops.

[0042] It should be noted that this algorithm requires that the two site clouds are in a good initial superposition state at the beginning, and requires that the point cloud to be registered is a subset of the reference site cloud.

[0043] Preferably, the step of selecting control points on the registered point cloud model as references for external model control of the lofting and performing computational evaluation on the on-site lofting data includes:

[0044] Select template assembly control points and extract coordinates;

[0045] Output external template layout adjustment suggestions.

[0046] The present invention also provides a quality control system for a small box girder, comprising:

[0047] The BIM model module is used to build a BIM model of the small box girder design according to the bridge design drawings; the BIM design model of the small box girder is then converted into a discrete point cloud;

[0048] 3D point cloud module, used to obtain and pre-process the prefabricated field template point cloud;

[0049] Registration module, used to register point cloud models with BIM models;

[0050] The evaluation module is used to select control points on the registered point cloud model as the benchmark for external model control layout, and to perform calculation and evaluation on the on-site layout data.

[0051] The beneficial effects of the quality control method and system for small box girders provided by the present invention are as follows: compared with the prior art, the quality control method and system for small box girders provided by the present invention adopt BIM model interpolation to generate point clouds, combine point cloud-point cloud registration algorithm to complete high-precision registration, and evaluate the registration effect based on the root mean square error. Closed-loop quality control from design to construction reduces manual intervention and can achieve full process digitization. Through algorithm fusion (4PCS+ICP+FPFH), the registration error is ≤4cm, which meets engineering requirements and effectively improves accuracy. Avoid traditional measurement rework, shorten template adjustment time by more than 50%, and improve efficiency. It solves the pain points of low precision and poor efficiency of traditional methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0053] Figure 1 A schematic diagram of a small box girder design BIM model constructed by a quality control method for small box girders provided in an embodiment of the present invention;

[0054] Figure 2 Schematic diagram of the effect of model interpolation and completion of the BIM model of the small box girder design in a quality control method for a small box girder provided in an embodiment of the present invention Figure 1 ;

[0055] Figure 3 Schematic diagram of the effect of model interpolation and completion of the BIM model of the small box girder design in a quality control method for a small box girder provided in an embodiment of the present invention Figure 2 ;

[0056] Figure 4 The registration effect of a quality control method for a small box beam provided by an embodiment of the present invention Figure 1 ;

[0057] Figure 5 The registration effect of a quality control method for a small box beam provided by an embodiment of the present invention Figure 2 ;

[0058] Figure 6 The registration effect of a quality control method for a small box beam provided by an embodiment of the present invention Figure 3 ;

[0059] Figure 7 The registration effect of a quality control method for a small box beam provided by an embodiment of the present invention Figure 4 ;

[0060] Figure 8 This is a structural block diagram of a quality control system for a small box girder provided in an embodiment of the present invention.

[0061] In the figure: 1. BIM model module; 2. 3D point cloud module; 3. Registration module; 4. Evaluation module. DETAILED DESCRIPTION

[0062] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0063] The terms "first," "second," and the like in the specification and claims of this application are used to distinguish similar objects, and are not used to describe a specific order or precedence. It should be understood that the terms used in this manner are interchangeable where appropriate, so that the embodiments of this application can be implemented in an order other than that illustrated or described herein, and that the objects distinguished by "first," "second," and the like are generally of the same type, and do not limit the number of objects; for example, the first object can be one or more. In addition, the term "and / or" in the specification and claims refers to at least one of the connected objects, and the character " / " generally indicates that the objects connected are in an "or" relationship.

[0064] Below, in conjunction with the accompanying drawings, a quality control method for a small box girder provided in an embodiment of the present application is described in detail through specific embodiments and application scenarios.

[0065] Please also refer to Figures 1 to 8 The present invention now provides a method for controlling the quality of small box girders. The method comprises scanning and constructing a small box girder outer formwork, obtaining an internal point cloud of the small box girder outer formwork, analyzing the flatness of the bottom and side formworks, and comparing the laser point cloud with the design BIM model to find the precise locations of the control points for assembling the outer formwork. Adjustment suggestions are then made for the closure of the tire frame, thereby achieving quality control of the prefabrication of the small box girder. Specifically, the method comprises the following steps:

[0066] Step S1: Construct a BIM model for the design of a small box girder according to the bridge design drawings; convert the BIM design model of the small box girder into a discrete point cloud. A parametric small box girder model is created using BIM software, which includes geometric features (bottom, side walls, top), prestressed ducts, steel mesh and other details. The method for converting the BIM design model of the small box girder into a discrete point cloud is to interpolate all triangular facets of the BIM model one by one according to the barycentric coordinate rule, with the point spacing consistent with the point spacing of the prefabrication site template point cloud, to obtain a point cloud corresponding to the BIM geometric model.

[0067] The specific implementation process of this step is:

[0068] Step S1.1: Create a structure to store model information of the small box girder design BIM model, where the model information includes but is not limited to vertices, normals, and texture coordinates;

[0069] Step S1.2: Read the triangular facet data of the small box girder BIM model and determine the number of facets. Specifically, read and parse the obj file of the model to obtain the facet data of the small box girder and determine the number of triangular facets;

[0070] Step S1.3, obtain the point cloud corresponding to the BIM geometric model. Specifically, create a function to traverse the index of all facets and perform equidistant interpolation on each triangular facet; set a parameter to control the density of the interpolation, determine the optimal parameter value through experiments, and finally output it as a TXT file containing a virtual point cloud, which is merged with the measured point cloud for subsequent alignment. Among them, the best effect is achieved when the interpolation density parameter is 14. Specifically, all triangular facets of the BIM model are interpolated one by one according to the barycentric coordinate law, and the point spacing is consistent with the point spacing of the prefabrication site template point cloud to obtain the point cloud corresponding to the BIM geometric model.

[0071] Step S2: Obtain the prefabrication site template point cloud and preprocess the prefabrication site template point cloud.

[0072] In this step, 3D laser scanning can replicate the real scene. Through on-site point cloud scanning, splicing, denoising, data resampling, data analysis, and results inspection, the scanned point cloud is processed to obtain a 3D point cloud of the prefabrication site.

[0073] The specific implementation process of this step is:

[0074] Step S2.1: Collecting on-site data. Specifically, it includes:

[0075] Step S2.1.1, survey and select points and arrange control points around the prefabrication site;

[0076] Step S2.1.2: Scan the site using a scanner to collect the data required for the 3D point cloud of the prefabrication site;

[0077] Step S2.1.3: Perform fill scans on the specular reflection area (such as the bottom plate coating) to reduce data loss;

[0078] Step S2.2: Preprocess the prefabrication site's 3D point cloud to obtain a 3D point cloud model. This preprocessing includes one or more of denoising, downsampling, and splicing. More specifically, denoising, downsampling, and splicing are performed sequentially to filter out ambient noise, reduce point cloud density, improve computational efficiency, and align multi-station scan data based on control points to form a complete template point cloud.

[0079] Step S2.3: Analyze the template flatness. Specifically, first perform plane fitting, then calculate the surface flatness. The plane fitting algorithm uses the least squares method. Flatness calculation: Count the distances from all points to the plane and calculate the average deviation.

[0080] In this step, the surface of the prefabrication site template base is required to be smooth and flat, the allowable deviation of flatness over a length of 2m should not exceed 2mm, and the deflection of the base or bottom mold should be no more than 2mm.

[0081] Step S3: aligning the point cloud model with the BIM model;

[0082] The specific implementation process of this step is:

[0083] Step S3.1, achieving initial alignment through coarse registration; wherein the coarse registration is performed using the SAC-IA algorithm or / and the 4PCS algorithm;

[0084] SAC-IA algorithm: Through sampling consistency initial registration, it automatically finds pairs of points with the same name and calculates the initial transformation matrix. Specifically, the box girder point cloud is first used to search for the nearest point pair corresponding to each point in the mold point cloud. The found point pairs are then used to calculate the rigid body transformation matrix. This transformation matrix is ​​applied to the box girder point cloud, and the mean square error of the distance between the transformed box girder point cloud and the mold point cloud is calculated. The algorithm iterates the process from searching for the nearest point to the final error assessment until the two errors are less than the threshold, at which point the iteration ends.

[0085] 4PCS algorithm: Based on the search for four-point congruent sets, it quickly matches the geometric features of the point cloud and the model. The 4PCS (4-Points Congruent Sets) algorithm is used for coarse registration. Coarse registration achieves basic matching between the point cloud and the model. Specifically, several groups of point pairs with the same name are found in the two-site clouds, and a rigid body transformation matrix is ​​calculated using these point pairs. This matrix is ​​applied to the site cloud to be registered, so that the two site clouds are pieced together as much as possible, thereby facilitating subsequent fine registration. At the same time, once the initial registration is completed, the overlapping parts of the two site clouds can be easily calculated, thus providing good fast convergence conditions for fine registration.

[0086] The specific implementation process of this step can be:

[0087] Step S3.1.1, input two site clouds: the site cloud to be registered P and the reference site cloud Q; calculate the normal information of each point cloud: N P 、N Q ;

[0088] Step S3.1.2, calculate the fast feature histogram FPFH of each point cloud: F P 、FQ ;

[0089] Step S3.1.3: Randomly sample m points P1, P2, P3, ..., P from the cloud of sites to be registered. m ; and find the m points Q1, Q2, Q3, ..., Q that are closest to the point feature fast histogram of the above m points in the base station site cloud Q m ; Then use these m groups of point pairs to calculate the transformation matrix T, and apply the matrix T to the site cloud to be registered: P ' =P*T; find the corresponding point.

[0090] Step S3.1.4: Calculate P and Q ' The registration error ∈ between them, store the corresponding transformation matrix T;

[0091] Step S3.1.5: Determine whether the number of iterations i_iter is less than the set number MaxIterCounts. If so, return to step S2.2 and continue to step S2.4. Otherwise, exit the loop.

[0092] Step S3.1.6: Compare all groups of registration errors in the loop process, select a group with the minimum registration error, output its corresponding transformation matrix T, apply it to the original site cloud P to be registered, and complete the initial registration.

[0093] In this step, the specific details can be divided into the following modules:

[0094] Neighborhood search selection.

[0095] For the neighborhood search selection algorithm,

[0096] First, the concepts of feature descriptors and neighborhoods are clarified. For a given point P q , and its corresponding point set in the domain is To express.

[0097] These neighborhood point sets can be determined by the following expressions:

[0098]

[0099] Where, d m is the neighborhood search radius, i.e., the maximum distance between the neighborhood points and the query point, ||·|| x is an example of a Minkowski norm (different distance norms can also be used).

[0100] In addition to using radius to constrain the neighborhood point set, it can also be described by a neighborhood limit number K, that is, the number of points from point P q The nearest K points are P qTherefore, the point feature descriptor can be represented by a vector function F, which can describe the neighborhood point of point P. q The surrounding neighborhood point set P k The local geometric information reflected:

[0101] F(P q , P k )={x1,x2,x3…x n}

[0102] Where x i , i∈{1…n} represent the value of the i-th dimension of the feature vector.

[0103] Then to compare whether the features of two points p1 and p2 are consistent, we only need to compare whether the values ​​of their feature vectors F1 and F2 in each dimension are the same or close.

[0104] Let Γ be the similarity expression of points p1 and p2, and d be a scale metric function, then:

[0105] Γ=d(F1,F2)

[0106] Therefore, when Γ approaches a minimum value, that is, Γ→0, then the feature descriptors of the two points are considered to be similar. On the contrary, if Γ is a very large value, then the two points are considered to be very different from each other, that is, they should be on different geometric surfaces. The neighborhood point set of the point is found through a single point search, so the geometric features of the local surface of the point can be well expressed by the point feature descriptor. Therefore, point cloud registration can rely on this feature variable when looking for corresponding relationships, and find pairs of points with the same name that have consistent cloud features between the two stations, thereby solving the registration problem. What follows is the calculation of the normal and point feature fast histogram of the point cloud. The FLANN neighborhood search library is used to perform neighborhood search. When searching for a point set with similar features to the m points sampled from the station cloud P to be registered in the base station cloud Q, a neighborhood search is also required. The neighborhood search here mainly searches for the point closest to the fast histogram of the point features of the sample point as the neighborhood of the sample point. When calculating P ' When the registration error ∈ between , Q, the FLANN library is used to search for the transformed site cloud P to be registered ' The nearest neighbor point of each point in the benchmark site cloud Q.

[0107] Find pairs of points with the same name.

[0108] For the search of point pairs with the same name, after the normal and histogram of the point cloud are calculated, it is necessary to start selecting sample points of the site cloud P to be registered.

[0109] According to the principle of measurement control, control points should be evenly distributed, and the control points with the same name in the registration work should also follow this principle. Therefore, geometric constraints are required when randomly sampling the sample points of the benchmark site cloud, and a distance threshold ε between the sample points needs to be set. dist This threshold is determined by the point cloud density, and it is sufficient to ensure that the sampled points are not distributed in a small area.

[0110] If a point cannot be sampled within the set threshold, the algorithm should be designed to automatically reduce the set threshold and continue sampling until a suitable sample point is collected.

[0111] Since the purpose of sampling points and subsequent search for point pairs is to calculate a transformation matrix, and the rigid body transformation matrix between point sets includes three rotation angles and three translation amounts, the necessary number of point pairs to solve the transformation matrix should be 3 sets of three-dimensional point pairs. Therefore, only three points need to be sampled in the algorithm, and the number of sampling points can also be appropriately increased.

[0112] The three sample points P1, P2, and P3 in the site cloud P to be registered search for corresponding point sets with similar features in the reference site cloud Q. Each sample point searches for k neighborhood values, forming the following point pair set:

[0113] C={c i ||c i = <P I , Q i1 , Q i2 ,...,Q ik >, 0 <i<3}

[0114] In the above formula, each sampling point corresponds to k most similar feature points. Here, a point is randomly sampled from its corresponding k neighborhood points as a sample point pair, thereby constructing the correspondence between the two site clouds.

[0115] Transformation matrix calculation.

[0116] For transformation matrix calculation, once the correspondence between the two site clouds is established, the corresponding transformation matrix T is calculated.

[0117] The three-dimensional transformation matrix is ​​mainly composed of three modules: rotation, translation, and scaling. If the original coordinates of the three-dimensional point are agreed to be [xy z], the new coordinates after transformation are [x ' y ' z ' ]. Then we have:

[0118] ①The formula for rotation transformation is:

[0119]

[0120] ②The translation transformation formula is:

[0121]

[0122] ([T x T y T z ] is the translation amount of the three-dimensional coordinate value)

[0123] ③The scaling transformation formula is:

[0124]

[0125] ([S x S y S z ] is the scaling factor on the three-dimensional axis)

[0126] Step S3.2: Implement refined alignment through fine registration; wherein, fine registration is performed through an ICP algorithm and / or a key point-based registration algorithm.

[0127] ICP algorithm: Utilizes the geometric invariance (vector / line segment ratio, Euclidean distance between points) in rigid body transformations. Based on the fact that the ratio of line segments occupied by intersection points remains unchanged after rigid transformation and the Euclidean distance between points remains unchanged, the algorithm tries to find four approximately coplanar points (approximately congruent four-point set) in the target point cloud to correspond to them. The transformation matrix is ​​then calculated using the least squares method. Multiple bases are iteratively selected based on the RANSAC algorithm framework, and the optimal transformation is obtained by comparison based on the evaluation criteria of the largest common point set (LCP). Specifically, an initial transformation is selected for the two sets of point clouds (which can be random or estimated). In each iteration, the nearest neighbor points (points in the target point cloud) of each point in the source point cloud are found, and the error between each pair of points is calculated. The optimal transformation (translation and rotation) is then calculated by minimizing these errors (commonly using the least squares method). Finally, the transformation is updated and the iteration is repeated until the error converges.

[0128] The specific implementation process of this step can be: the existing site cloud to be registered P = {p i |p i ∈R 3 ; i = 1, 2, 3, ..., n} and the reference site cloud Q = {q i |q i ∈R 3 ; i = 1, 2, 3, ..., m}, where n ≤ m, the algorithm first traverses each point P in the point cloud P i , search for the nearest point Y in the point cloud Q i As the corresponding point. If the symbol is defined To search for the nearest neighbor operation, we have: After the point pairs are established, the transformation matrix can be calculated. The least squares principle is used to calculate the transformation matrix, so that the transformation matrix with the smallest error measure is the required matrix.

[0129]

[0130] Solution error measure Minimum rigid body transformation parameters Right now:

[0131]

[0132] Where d is the mean square error of the Euclidean distance between the two station clouds after the corresponding transformation, that is The obtained rigid body transformation is applied to the site cloud P to be registered, which is recorded as: Then iterate repeatedly until the iterative convergence condition is met.

[0133] That is, the specific implementation process of step S3.2 can be:

[0134] S3.2.1, Initialization: P0 = P, k=0;

[0135] S3.2.2. Cloud P of the site to be registered k In the benchmark site cloud Q, the neighborhood search method is used to search for the nearest neighbor points corresponding to each point in the point cloud P' to form neighboring point pairs.

[0136] S3.2.3. Use all nearest point pairs to calculate the parameters corresponding to the three-dimensional transformation matrix

[0137] S3.2.4, R k 、T k Apply it to the point cloud P to be registered to obtain a new point cloud to be registered

[0138] S3.2.5. Compare the difference between the current error value and the previous error value to see if it is less than the given threshold ε: d k -d k+1 <ε, if it is satisfied, stop the iteration, otherwise repeatedly iterate steps S3.2.2 to S3.2.4.

[0139] It should be noted that this algorithm requires that the two site clouds are in a good initial superposition state at the beginning, and requires that the point cloud to be registered is a subset of the reference site cloud.

[0140] The key point-based registration algorithm completes the registration process:

[0141] Based on the geometric characteristics of the small box girder and mold, a keypoint-based registration algorithm is used to perform manual registration. This method achieves automatic registration by automatically detecting key feature points in the point cloud. Common feature extraction methods include SIFT (Scale-Invariant Feature Transform), SURF (Speeded Robust Features), and FPFH (Fast Point Feature Histograms). These methods extract local features from the point cloud and match them for registration. The Euclidean mean square error (MSE) between points and surfaces is used as an error measure. Based on this least squares principle, the transformation matrix is ​​calculated using the quaternion method. The transformation result judgment module then determines whether the iteration has reached convergence conditions. If the conditions are met, the final refined registration result is output. Keypoint matching can extract FPFH feature descriptors, enhancing the robustness of local geometric feature matching.

[0142] The specific implementation process of this step can be as follows: (High-precision alignment is achieved through automatic feature matching, transformation solution and convergence judgment. Its innovation lies in combining key point features (such as FPFH) and robust filtering strategies to reduce dependence on manual intervention, and is suitable for efficient alignment of complex structures such as small box beam molds)

[0143] Input the site cloud P to be registered and the reference site cloud Q;

[0144] Initialization: transformation matrix T0 = I, data simplification and filtering of the registration site cloud P and the reference site cloud Q: P ' =datafilter(P),Q ' =tafilter(Q), set the iterative convergence threshold: ε Angle , ε Trans ;

[0145] Start iterating and looping the following process ad until the iteration converges:

[0146] a. The point cloud P' to be registered undergoes three-dimensional transformation: P' k =T k-1 (P' K );

[0147] b. Find the nearest neighbor point pair: M k =match(P' K , Q');

[0148] c. Outlier pair filtering: W k =outlier(M k );

[0149] d. Least squares estimation:

[0150] e. Transformation result judgment:

[0151] Output the final registration matrix: Where m is the number of iterations when the iteration converges.

[0152] For the point cloud downsampling part, this algorithm uses random sampling filtering.

[0153] The random sampling set expression is as follows:

[0154] P′=datafilter(P)={p∈P:η(p)<θ}

[0155] Where η∈[0,1), η is any random value in the range. θ∈[0,1], θ is the ratio setting value of point cloud sampling, ranging from 0 to 1.

[0156] The normal calculation set expression is as follows:

[0157]

[0158] Where normal(p) is the normal of point p estimated based on the neighborhood around point p.

[0159] By comparing the feature descriptors of key points, the best matching point pairs are found. The point pair matching module mainly uses the to-be-registered site cloud P to find the corresponding nearest point pairs in the reference site cloud Q, which are ultimately used for subsequent transformation matrix calculations.

[0160] There are three modes for searching neighboring point pairs: ① Euclidean distance matching of 3D point coordinates, ② matching with added feature description information, and ③ hybrid matching. Neighborhood point matching is used for fine-grained matching. Its core is to use KD-TREE linear heaps to partition the point cloud space. The neighborhood search uses a range search mechanism. Given a maximum distance constraint threshold for a point pair, the nearest point within the threshold is searched as a neighboring point. The point pair matching set expression is as follows:

[0161]

[0162] Where d(P n , Q x ) is point P n and Q x The Euclidean distance between That is, filter out the corresponding P in the index of the benchmark site cloud Q n The index x of the nearest neighboring point.

[0163] Then use the least squares method to calculate and optimize the rigid transformation (rotation and translation) as shown in the following formula:

[0164]

[0165] Where n i Q is the midpoint of the Q cloud at the benchmark site i The three-dimensional normal vector, R and t are the rotation parameters and translation parameters of the transformation matrix T respectively. Finally, the transformation is adjusted to ensure that the two point clouds are aligned.

[0166] Step S3.3: Evaluate the registration result. Specifically, the root mean square error is used to evaluate the registration result.

[0167] In this method, the root mean square error (RMSE) between points is used as the error evaluation metric. RMSE quantifies the spatial distance error between point clouds. Specifically, RMSE calculates the distance from each point in the source point cloud to the nearest point in the target point cloud, squares and averages these distances, and then takes the square root of the square root.

[0168] The specific implementation process of this step is:

[0169] For each point pi in the source point cloud, find the point qi in the target point cloud that is closest to it

[0170] Calculate the distance between each pair of points di = |||pi-qi|||;

[0171] Calculate the average of the squared distances of all pairs of points and then take the square root:

[0172] Where NN is the number of points in the point cloud, and di is the distance between point pi and its nearest neighbor point qi.

[0173] Step S4: Select control points on the registered point cloud model as the benchmark for external model control layout, and perform calculation and evaluation on the on-site layout data.

[0174] The specific implementation process of this step is:

[0175] Step S4.1: Select template assembly control points and extract coordinates.

[0176] In the area where the registered point cloud and the BIM model overlap, select five pairs of control points (such as the four corners of the base plate and the center of the side wall). Output the 3D coordinates (X, Y, Z) and deviation values ​​of the control points as the basis for the layout.

[0177] Step S4.2: Output external template layout adjustment suggestions.

[0178] In this step, deviation analysis is performed first, followed by the generation of adjustment suggestions and dynamic feedback. Specifically, the measured control point coordinates are compared with the theoretical coordinates of the BIM model to calculate the translation and rotation angles. Adjustment instructions (such as the amount of cradle movement and bolt tightness) are automatically generated through the "Small Box Girder Precast Quality Control Procedure." After adjustment, rescanning and verification are performed until the error meets the specifications.

[0179] This invention provides a quality control method for small box girders. This method scans the prefabricated small box girder exterior formwork to obtain an internal point cloud. This point cloud is then compared with the designed BIM model. Adjustment recommendations are made during the cradle closure process to control the quality of the prefabricated small box girders. The completed prefabricated small box girders and bridge cap beams are scanned, and the laser point cloud is compared with the designed box girder model to determine dimensional deviations. Based on the control benchmarks for the small box girder installation, optimized boundary conditions are determined, and mathematical algorithms are developed for optimization. Installation data is obtained to provide data support for physical installation.

[0180] The present invention also provides a quality control system for small box girders, comprising: a BIM model module 1, a three-dimensional point cloud module 2, a registration module 3 and an evaluation module 4, wherein the BIM model module 1 is used to construct a small box girder design BIM model according to a bridge design drawing; and then perform model interpolation and completion on the small box girder design BIM model; the three-dimensional point cloud module 2 is used to obtain a prefabrication site template point cloud and pre-process the prefabrication site template point cloud; the registration module 3 is used to register the point cloud model with the BIM model; and the evaluation module 4 is used to select control points on the registered point cloud model as a benchmark for external mold control layout, and to perform calculation and evaluation on the on-site layout data.

[0181] Compared with the existing technology, the quality control method and system for small box girders provided by the present invention use BIM model interpolation to generate point clouds, combine point cloud-point cloud registration algorithms to achieve high-precision registration, and evaluate the registration effect based on the root mean square error. Closed-loop quality control from design to construction reduces manual intervention and can achieve full process digitization. Through algorithm fusion, the registration error is ≤4cm, meeting engineering requirements and effectively improving accuracy. Traditional measurement rework is avoided, template adjustment time is shortened by more than 50%, and efficiency is improved. The pain points of low precision and poor efficiency of traditional methods are solved.

[0182] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A quality control method for small box beams, characterized in that: The following steps are involved: According to the bridge design drawings, a BIM model of the small box girder design is constructed; then the BIM design model of the small box girder is converted into a discrete point cloud; Obtaining the prefabrication site template point cloud and preprocessing the prefabrication site template point cloud; The point cloud model is registered with the BIM model; Select control points on the registered point cloud model as the benchmark for external model control layout, and perform calculation and evaluation on the on-site layout data; The point cloud model is registered with the BIM model, including: Initial alignment is achieved through coarse registration, wherein the coarse registration is performed using the SAC-IA algorithm or / and the 4PCS algorithm; Refined alignment is achieved through fine registration, wherein fine registration is performed using an ICP algorithm and / or a key point-based registration algorithm; The root mean square error was used to evaluate the registration results; The initial alignment is achieved through coarse registration, comprising: Input two site clouds: the site cloud to be registered P and the reference site cloud Q; calculate the normal information of each point cloud: N P 、N Q ; Calculate the fast feature histogram FPFH of each point cloud: F P 、F Q ; Randomly sample m points P1, P2, P3, ..., P in the cloud of sites to be registered. m ; and find the m points Q1, Q2, Q3, ..., Q that are closest to the point feature fast histogram of the above m points in the benchmark site cloud Q m ; Then use these m groups of point pairs to calculate the transformation matrix T, and apply the matrix T to the site cloud to be registered: P ′ =P*T; find the corresponding point; Calculate P and Q ′ The registration error ∈ between them, store the corresponding transformation matrix T; Determine whether the number of iterations i_iter is less than the set number MaxIterCounts, and if not, exit the loop; Compare all groups of registration errors in the loop process, select a group with the minimum registration error, output its corresponding transformation matrix T, and apply it to the original site cloud P to complete the initial registration; According to the bridge design drawings, a small box girder design BIM model is constructed; and then the small box girder BIM design model is converted into a discrete point cloud, including: By creating a structure to store the model information of the small box girder design BIM model; Read the triangular facet data of the small box girder BIM model and determine the number of faces; Get the point cloud corresponding to the BIM geometric model.

2. A quality control method for a small box girder according to claim 1, characterized in that: Obtaining the point cloud corresponding to the BIM geometric model includes: Create a function to traverse the indices of all facets and perform equidistant interpolation on each triangle facet; The output is a TXT file containing a virtual point cloud, which is merged with the measured point cloud for subsequent registration.

3. A quality control method for a small box girder according to claim 2, characterized in that: The step of obtaining the prefabricated field template point cloud and preprocessing the prefabricated field template point cloud includes: Collect field data; Preprocessing the prefabricated site three-dimensional point cloud to obtain a three-dimensional point cloud model; wherein the preprocessing includes: one or more of: denoising, downsampling, and splicing; First perform plane fitting and then calculate the surface flatness.

4. A quality control method for a small box girder according to claim 3, characterized in that: The collecting of on-site data includes: Conduct exploration and site selection and control point layout around the prefabrication site; Use a scanner to scan on-site and collect the data required for the 3D point cloud of the prefabrication site; Perform fill scan on the mirror reflection area to reduce data loss.

5. A quality control method for a small box girder according to claim 4, characterized in that: The fine alignment achieved through precise registration includes: Initialization: P0 = P, k=0; Cloud P of the site to be registered k In the benchmark site cloud Q, the neighborhood search method is used to search for the nearest neighbor points corresponding to each point in the point cloud P' to form neighboring point pairs. Use all the nearest point pairs to calculate the parameters corresponding to the three-dimensional transformation matrix R k 、T k Apply it to the point cloud P to be registered to obtain a new point cloud to be registered Compare the difference between the current error value and the previous error value to see if it is less than the given threshold ε:d k -d k+1 <ε, if satisfied, stop iteration; It should be noted that this algorithm requires that the two site clouds are in a good initial superposition state at the beginning, and requires that the point cloud to be registered is a subset of the reference site cloud.

6. A quality control method for a small box girder according to claim 1, characterized in that: The control points are selected on the registered point cloud model as the reference for the external model control layout, and the on-site layout data are calculated and evaluated, including: Select template assembly control points and extract coordinates; Output external template layout adjustment suggestions.

7. A quality control system for small box beams, characterized in that: include: The BIM model module is used to build a BIM model of the small box girder design according to the bridge design drawings; the BIM design model of the small box girder is then converted into a discrete point cloud; 3D point cloud module, used to obtain the point cloud of the prefabrication site template and preprocess the point cloud of the prefabrication site template; Registration module, used to register point cloud models with BIM models; The evaluation module is used to select control points on the registered point cloud model as the benchmark for external model control layout, and to perform calculation and evaluation on the on-site layout data; The point cloud model is registered with the BIM model, including: Initial alignment is achieved through coarse registration, wherein the coarse registration is performed using the SAC-IA algorithm or / and the 4PCS algorithm; Refined alignment is achieved through fine registration, wherein fine registration is performed using an ICP algorithm and / or a key point-based registration algorithm; The root mean square error was used to evaluate the registration results; The initial alignment is achieved through coarse registration, comprising: Input two site clouds: the site cloud to be registered P and the reference site cloud Q; calculate the normal information of each point cloud: N P 、N Q ; Calculate the fast feature histogram FPFH of each point cloud: F P 、F Q ; Randomly sample m points P1, P2, P3, ..., P in the cloud of sites to be registered. m ; and find the m points Q1, Q2, Q3, ..., Q that are closest to the point feature fast histogram of the above m points in the base station site cloud Q m ; Then use these m groups of point pairs to calculate the transformation matrix T, and apply the matrix T to the site cloud to be registered: P ′ =P*T; find the corresponding point; Calculate P and Q ′ The registration error ∈ between them, store the corresponding transformation matrix T; Determine whether the number of iterations i_iter is less than the set number MaxIterCounts, and if not, exit the loop; Compare all groups of registration errors in the loop process, select a group with the minimum registration error, output its corresponding transformation matrix T, and apply it to the original site cloud P to complete the initial registration; According to the bridge design drawings, a small box girder design BIM model is constructed; and then the small box girder BIM design model is converted into a discrete point cloud, including: By creating a structure to store the model information of the small box girder design BIM model; Read the triangular facet data of the small box girder BIM model and determine the number of faces; Get the point cloud corresponding to the BIM geometric model.

Images