Railway box girder full-section data monitoring method and system based on three-dimensional scanning

Abstract

The application discloses a railway box girder full-section data monitoring method and system based on three-dimensional scanning, and particularly relates to the technical field of data monitoring, which comprises the following steps: collecting the full-section data of the railway box girder through a three-dimensional scanner, determining the full-section data collection process of the railway box girder by scanning each part of the box girder for multiple times, processing the collected full-section data of the railway box girder, splicing the point cloud data by means of the aerial triangulation calculation process based on the dense point cloud data, the accurate estimation of the key point circle center position, the key plane fitting and optimization, uploading the spliced data and the special marker point coordinates to an intelligent measurement integrated platform system by an operator, collecting the point cloud information and the contour information of the prefabricated box girder, constructing an acceptance evaluation model of the prefabricated box girder, and quantifying the possible significant deviation between the box girder model and the design standard. The application is helpful to improve the efficiency of the full-section acceptance of the railway box girder.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data monitoring, and more particularly, to a railway box girder full-section data monitoring method and system based on three-dimensional scanning. BACKGROUND

[0002] For many years, the size of the railway precast box girder has been measured by manual cooperation with conventional equipment and instruments such as steel tape, level, vernier caliper, and level. The measurement accuracy is low, the measurement efficiency is low, and the labor cost is high. With the development of intelligent measurement and scanning technology, based on the need for box girder appearance size measurement, to solve the problems of low work efficiency and human error in the traditional railway box girder appearance size acceptance method.

[0003] In order to solve the above defects, a technical scheme is provided. SUMMARY

[0004] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a railway box girder full-section data monitoring method and system based on three-dimensional scanning to solve the problems raised in the background art.

[0005] To achieve the above object, the present application provides the following technical scheme:

[0006] The railway box girder full-section data monitoring method based on three-dimensional scanning specifically includes the following steps:

[0007] S1: Collecting the data of the full section of the railway box girder by a three-dimensional scanner, determining the data collection process of the full section of the railway box girder by scanning each part of the box girder multiple times;

[0008] S2: Processing the collected full-section data of the railway box girder, splicing the point cloud data by the process of aerial triangulation based on dense point cloud data, accurate estimation of the key point circle center position, and key plane fitting and optimization;

[0009] S3: The operator uploads the spliced data and special marker point coordinates to the intelligent measurement integrated platform system, the system automatically measures the current model, generates acceptance data and exports a complete acceptance form;

[0010] S4: Collecting the point cloud information and contour information of the precast box girder, constructing an acceptance evaluation model of the precast box girder, and quantifying possible significant deviations between the box girder model and the design standard.

[0011] In a preferred embodiment, the data collection process includes beam end data collection, bottom web side data collection, indoor beam chamber data collection, and top plate data collection, wherein:

[0012] In the beam end data collection, special mark points are set at the beam bottom angle, the beam top and the support center line respectively using standard parts, which are used as coordinate positioning points after splicing, and after setting the special mark points and the positioning mark balls, non-contact three-dimensional laser scanners are used for data collection at the middle position of the beam end;

[0013] In the bottom web side data collection, special mark points are set at the span line and the beam end respectively using standard parts, which are used as coordinate positioning points after splicing, and after setting the mark points, non-contact three-dimensional laser scanners are used for web data collection at the position 6m away from the beam end on the beam side, and non-contact three-dimensional laser scanners are used for bottom plate data collection at the middle position of the beam side;

[0014] In the beam indoor data collection, two positions close to the beam end and capable of observing the positioning mark balls placed at the beam end are selected as data collection points, and the non-contact three-dimensional laser scanner is adjusted to ensure that overexposure does not occur, and then the non-contact three-dimensional laser scanner is used for data collection;

[0015] In the top plate data collection, two positions close to the beam end and capable of observing the positioning mark balls placed at the beam end are selected as data collection points, and after the two data collection points are determined, the non-contact three-dimensional laser scanner is used for data collection.

[0016] In a preferred embodiment, the aerial triangulation calculation process based on dense point cloud data includes:

[0017] The absolute position coordinate system is redefined, the current existing aerial triangulation calculation method is organically combined, the calculation process is redesigned, and the error caused by the aerial triangulation calculation is minimized or eliminated, including fitting the maximum plane of the beam end and solving the plane equation, solving the angle between the plane normal vector and the absolute coordinate system normal vector, constructing a three-dimensional rotation matrix, point cloud data transposition and determining the point cloud model center point.

[0018] In a preferred embodiment, the key point center position accurate estimation and key plane fitting and optimization include:

[0019] A fitting circle model is used to accurately estimate the center position, the least squares method is used to calculate the best center position, the plane model parameters are optimized by minimizing the distance from the point to the plane, and the best plane fitting effect is obtained.

[0020] In a preferred embodiment, the operator uploads the splicing data and special mark point coordinates to the intelligent measurement integrated platform system, including:

[0021] The intelligent measurement integrated platform system adopts advanced model optimization technology to reduce model size and improve inference speed, uses a distributed learning method to divide the model into multiple parts, reduces transmission burden, utilizes parallel computing technology to exert the performance of multi-core processors and GPUs, improves system computing speed, and distributes tasks to multiple computing nodes by constructing a distributed system;

[0022] An incremental data transmission method is adopted to transmit only the changed data, block processing of the LAS file is implemented, only the part of data required at present is loaded and processed, the file processing efficiency is improved, a distributed file system or database is used and an index technology is introduced to speed up the retrieval and query operation of large data such as the LAS file.

[0023] In a preferred embodiment, the point cloud information of the prefabricated box girder comprises:

[0024] The point cloud information of the prefabricated box girder is represented by a point cloud matching error abnormality coefficient;

[0025] The acquisition logic of the point cloud matching error abnormality coefficient is: obtaining different point cloud data sets in the same coordinate system, marking the different point cloud data sets in the same coordinate system as: , wherein n = 1, 2, 3, …, N, N is a positive integer, n is the point cloud data set number of different parts of the prefabricated box girder scanned by the three-dimensional scanner, , k = 1, 2, 3, …, K, K is a positive integer, k is the number of point cloud data in the different point cloud data sets;

[0026] Based on the matching of two different point cloud data using the ICP algorithm, the error between the point cloud data in the different point cloud data set pairs is calculated by the Euclidean distance, the point cloud error threshold is set, the point cloud data between the point cloud data in the different point cloud data set pairs that is greater than the point cloud error threshold is obtained, the number of point cloud data greater than the point cloud error threshold is obtained, and the number of point cloud data greater than the point cloud error threshold is marked as: , wherein m is the number of different point cloud data set pairs, m = 1, 2, 3, …, M, M is a positive integer, M = N-1;

[0027] The total number of point cloud data deleted in the point cloud data splicing process is obtained, and the total number of point cloud data deleted in the point cloud data splicing process is marked as: , wherein ;

[0028] The point cloud matching error abnormality coefficient is calculated, and the calculation formula is: ; wherein is the point cloud matching error abnormality coefficient, is the number of point cloud data in the different point cloud data sets.

[0029] In a preferred embodiment, the profile information of the prefabricated box girder comprises:

[0030] The profile information of the prefabricated box girder is represented by a cross-section profile deviation coefficient and a spacing deviation coefficient;

[0031] The acquisition logic of the cross-section profile deviation coefficient is: after the splicing of the prefabricated box girder point cloud data is completed, a three-dimensional point cloud model of the prefabricated box girder is obtained, point cloud data of different cross sections of the three-dimensional point cloud model of the prefabricated box girder is obtained, curve fitting is performed on the point cloud data of different cross sections, two-dimensional cross-section curves of different cross sections are generated, and the two-dimensional cross-section curves of different cross sections are marked as: , wherein i=1, 2, 3, …, I, I is a positive integer, and i is the number of different cross sections selected;

[0032] The two-dimensional cross-section standard curves of different cross sections are set, and the two-dimensional cross-section standard curves of different cross sections are marked as: The distance deviation between each measurement point of the two-dimensional cross-section curve and the two-dimensional cross-section standard curve of different cross sections is calculated by Euclidean distance, and the distance deviation between each measurement point of the two-dimensional cross-section curve and the two-dimensional cross-section standard curve of different cross sections is marked as: , wherein j=1, 2, 3, …, J, J is a positive integer, and j is the number of each measurement point between the two-dimensional cross-section curve and the two-dimensional cross-section standard curve of different cross sections;

[0033] The total deviation of different cross sections is calculated, and the calculation formula is: ; wherein is the total deviation of the i-th cross section of the prefabricated box girder.

[0034] The average value and the standard deviation of the total deviation of different cross sections of the prefabricated box girder are calculated, and the average value and the standard deviation of the total deviation of different cross sections of the prefabricated box girder are marked as: and ; wherein , ;

[0035] The cross-section profile deviation coefficient is calculated, and the calculation formula is: ; wherein is the cross-section profile deviation coefficient;

[0036] The acquisition logic of the spacing deviation coefficient is: after the splicing of the prefabricated box girder point cloud data is completed, the positions of the special marked points in the three-dimensional point cloud model of the prefabricated box girder are obtained, and the position coordinates of the special marked points in the three-dimensional point cloud model of the prefabricated box girder are marked as: , wherein , q=1, 2, 3, …, Q, Q is a positive integer, and q is the number of special marked points;

[0037] obtaining the position coordinates of the preset special marker points, and marking the position coordinates of the preset special marker points as: wherein, ;

[0038] calculating the spacing deviation coefficient, and the calculation formula is: ; wherein, is the spacing deviation coefficient.

[0039] In a preferred embodiment, the acceptance evaluation model of the prefabricated box girder is constructed, comprising:

[0040] Through comprehensive analysis of the point cloud information and the contour information of the prefabricated box girder, the point cloud matching error anomaly coefficient, the cross-section contour deviation coefficient and the spacing deviation coefficient are weighted and calculated, the acceptance evaluation model of the prefabricated box girder is constructed, and the acceptance evaluation coefficient is generated, and the calculation formula is: ; wherein, is the acceptance evaluation coefficient, , , the proportion coefficients of the point cloud matching error anomaly coefficient, the cross-section contour deviation coefficient and the spacing deviation coefficient respectively, , , all greater than 0;

[0041] The acceptance evaluation coefficient threshold is set, the acceptance evaluation coefficient is compared with the acceptance evaluation coefficient threshold, if the acceptance evaluation coefficient is greater than the acceptance evaluation coefficient threshold, a warning signal is generated, indicating that the prefabricated box girder produced at present has a large error, if the acceptance evaluation coefficient is less than the acceptance evaluation coefficient threshold, no warning signal is generated, indicating that the prefabricated box girder produced at present has high accuracy.

[0042] In a preferred embodiment, the railway box girder full-section data monitoring system based on three-dimensional scanning comprises a data acquisition module, a model splicing module, a data uploading module and an acceptance evaluation module;

[0043] The data acquisition module is used for scanning the railway box girder multiple times by the three-dimensional scanner to capture the full-section point cloud data;

[0044] The model splicing module is used for restoring the three-dimensional coordinates of the point cloud by using the space three calculation through the multi-view dense point cloud data, extracting the key geometric points in the point cloud, accurately determining the center positions of these points by using the algorithm, fitting the plane part in the scanning area, and optimizing the splicing error between the point cloud data;

[0045] The data uploading module is used for uploading the complete spliced point cloud data and the special marker point information on the box girder to the intelligent measurement integrated platform;

[0046] The acceptance evaluation module is used for obtaining point cloud information and contour information of the prefabricated box girder from the point cloud model, calculating deviation of the box girder model from the design standard, generating an acceptance evaluation coefficient, and providing a scientific basis for quality control.

[0047] Technical effects and advantages of the present application:

[0048] In the intelligent acceptance of the external dimensions of the prefabricated box girder, the laser scanner is used to scan each surface of the box girder to obtain high-precision three-dimensional data, the dense point cloud reverse modeling method, the point cloud empty three calculation algorithm and the improved RANSAC plane fitting algorithm are introduced, the accurate measurement of the cross section of the railway box girder is realized, and the point cloud information and the contour information of the prefabricated box girder are collected to quantify the possible significant deviation between the box girder model and the design standard. The present application realizes the rapid, accurate and non-contact measurement of the full cross section of the railway box girder, improves the automation and intelligent level of the measurement, and helps to improve the efficiency of the full cross section acceptance of the railway box girder. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to facilitate the understanding of those skilled in the art, the present application will be further described below with reference to the accompanying drawings;

[0050] Figure 1 The flowchart of the railway box girder full cross section data monitoring method based on three-dimensional scanning of the present application is shown.

[0051] Figure 2 The structural diagram of the railway box girder full cross section data monitoring system based on three-dimensional scanning of the present application is shown. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0053] Embodiment 1

[0054] Figure 1 The flowchart of the railway box girder full cross section data monitoring method based on three-dimensional scanning of the present application is shown. Specifically, the following steps are included:

[0055] S1: Collect the data of the full cross section of the railway box girder by the three-dimensional scanner, and determine the data collection process of the full cross section of the railway box girder by scanning each part of the box girder multiple times;

[0056] S2: Process the collected full-section data of the railway box girder, and splice the point cloud data through the aerial triangulation calculation process based on dense point cloud data, accurate estimation of the center position of the key point, and key plane fitting and optimization.

[0057] S3: The operator uploads the spliced data and special marker point coordinates to the intelligent measurement integrated platform system, which automatically measures the current model, generates acceptance data, and exports a complete acceptance table.

[0058] S4: Collect the point cloud information and contour information of the precast box girder, and construct an acceptance evaluation model of the precast box girder to quantify possible significant deviations between the box girder model and the design standard.

[0059] In step 1, in the intelligent acceptance of the precast box girder, a laser scanner is used to scan each surface of the box girder to obtain high-precision three-dimensional data. To ensure data accuracy, multiple stations are used to scan each part of the box girder multiple times to improve data integrity and accuracy.

[0060] The automatic program can control the laser scanner to collect data and automatically process and analyze the collected data. The automatic program includes the following functions:

[0061] Data collection: The automatic program can set the scanning parameters and path to control the laser scanner to collect data. During the collection process, the automatic program can also monitor the integrity and accuracy of the data in real time to ensure the reliability of the data.

[0062] Data processing: The automatic program can preprocess and analyze the collected data. For example, remove noise, smooth data, calculate the dimensions of the box girder, etc. Through the processing of the automatic program, the time and error of manual processing can be greatly reduced.

[0063] Intelligent judgment: The automatic program can use machine learning algorithms to intelligently judge the processed data. Through training and learning on a large amount of data, the automatic program can learn the qualified standards of the box girder dimensions and automatically judge whether the box girder dimensions meet the requirements.

[0064] Repeated measurement function: Since there may be some changes in the production process of the precast box girder, the data collection system needs to be reusable for multiple measurements and comparisons. The repeated measurement function can record the parameters and results of each measurement and generate a measurement report. By comparing the measurement results of different times, the trend of changes in the dimensions of the box girder can be found, and appropriate measures can be taken for quality control.

[0065] For the rapid acquisition of full-section data of railway box girder, the scope of acquisition is comprehensive coverage of every detail of the box girder, that is, the shape, inner box, bottom surface and top surface can be measured without dead angle. Through the disassembly and analysis of each step of the manual measurement process, and according to the requirements of different measurement parameters, the data acquisition process is reasonably divided into four parts, which are beam end data acquisition, bottom web side data acquisition, beam indoor data acquisition and top plate data acquisition, wherein:

[0066] In the beam end data acquisition, special marker points are set at the beam bottom inclined angle, beam top and support center line using standard parts, which are used for coordinate positioning points after splicing. After setting the special marker points and splicing positioning marker balls, non-contact three-dimensional laser scanner is used for data acquisition at the middle position of the beam end.

[0067] In the bottom web side data acquisition, special marker points are set at the span line and beam end using standard parts, which are used for coordinate positioning points after splicing. After setting the marker points, non-contact three-dimensional laser scanner is used for web data acquisition at the position 6m away from the beam end on the beam side, and non-contact three-dimensional laser scanner is used for bottom plate data acquisition at the middle position of the beam side.

[0068] In the beam indoor data acquisition, two positions close to the beam end and able to observe the positioning marker balls placed at the beam end are selected as data acquisition points. After adjusting the non-contact three-dimensional laser scanner to ensure that there is no overexposure, non-contact three-dimensional laser scanner is used for data acquisition.

[0069] In the top plate data acquisition, two positions close to the beam end and able to observe the positioning marker balls placed at the beam end are selected as data acquisition points. After determining the two data acquisition points, non-contact three-dimensional laser scanner is used for data acquisition.

[0070] Through non-contact three-dimensional laser scanner scanning of the full section of railway box girder, the full section data of box girder is covered without dead angle through multiple scanning. After completing the above data acquisition, the collected data is sent back to the operator's computer through the Internet, and then the collected data is spliced and some special marker points are extracted for subsequent processing. The spliced data is a relatively accurate three-dimensional reconstruction model.

[0071] In the full-section data acquisition of railway box girder, the support center line and span line are marked on the side of the box girder, and the marker targets are fixed around the box girder to ensure the accuracy of the scanning data and the high precision of the later splicing.

[0072] It should be noted that the support center line and the span line are key reference lines in the geometric dimensions of the box girder, providing clear reference for subsequent data calibration and splicing, and serving as the basis of the coordinate system, ensuring that the point cloud data after scanning can be accurately mapped into a unified spatial coordinate;

[0073] The marker target is usually used as a splicing reference point for data collection at multiple stations, ensuring that the data from different scanning stations can be seamlessly spliced into a complete three-dimensional model through the position information of the target. The position of the target in space is clear and easy to identify, and is the core input of the splicing algorithm. In addition, the data collected by different stations may have deviations in the coordinate system, and through the target, these data can be unified into a global coordinate system.

[0074] To adapt to the full scanning of the outer plane of the box girder and reduce errors, one scanning station is selected at each end of the beam for collecting box girder end plane point cloud. Two stations are set up on both sides of the beam 6m away from the beam end for collecting box girder end web and wing plate side point cloud. Two stations are set up at the bottom of the box girder for collecting bottom plate point cloud. Two stations are set up inside the box chamber for collecting box chamber point cloud. Two stations are set up on the top plate for collecting top plate point cloud. The splicing between stations uses target balls for splicing. Three target balls are placed on the diagonal lines of the box girder, so that each

[0075] Each station can observe at least three stationary target balls.

[0076] Box girder space position random, angular uneven point cloud collection and positioning technology. Using the characteristics of instrument recognizing object color, black and white targets are used to mark position information into point cloud. In order to reduce the amount of preliminary work, the principle of setting as few targets as possible is followed, and the principle of unnecessary target is not set. Four targets are set at each corner of the beam bottom to solve the problem of right angle point without point caused by the round corner of the box girder bottom surface. The support center line in the transverse and longitudinal directions is marked with a target to solve the problem of not being able to scan the support hole below.

[0077] After copying the data measured by the scanner to the computer, import it into the SCENE platform software. By identifying the spheres placed in the scanning field, the images collected are spliced based on these points, forming a complete box beam model. To further improve the precision of the cluster splicing, targets with splicing errors greater than the preset value are processed. The incorrect base points are processed and re-spliced by manual deletion to reduce errors. After splicing, in the 3D view, use the manipulator to move, rotate and scale the picture in a simple way, crop the excess image and delete the invisible points to reduce the point cloud volume and improve the operation fluency. Then create a project point cloud and export it. Using the recognition function of the software, the black and white targets marked on the box beam can be marked, and 22 marker point coordinates can be obtained. Finally, import the exported project point cloud and marker point coordinates into the box beam three-dimensional measurement system platform, and process the data through the platform to finally output the box beam appearance inspection original record report.

[0078] In step 2, the collected data is further processed and analyzed to determine whether the dimensions of the precast box beam meet the requirements, including the following functions:

[0079] Data cleaning: remove noise and outliers in the collected data to ensure the accuracy of data analysis;

[0080] Data transformation: transform and process the collected data to better reflect the dimensions of the precast box beam, such as smoothing, filtering, etc.

[0081] Feature extraction: extract feature information related to the dimensions of the precast box beam from the processed data, such as length, width, height, curvature, etc.

[0082] Model construction: use the extracted feature information to construct a model to predict whether the dimensions of the precast box beam meet the requirements. Various machine learning algorithms can be used for model construction, such as support vector machines, neural networks, decision trees, etc.

[0083] The splicing of point cloud data mainly involves preprocessing the original point cloud data obtained from the three-dimensional scanning device to improve the quality of the data and the accuracy of the splicing. The preprocessing includes the following steps:

[0084] Data denoising: due to various factors such as device error, environmental lighting, etc. during the scanning process, noise points may exist in the obtained point cloud data. Through data denoising, these noise points can be removed to improve data quality.

[0085] Data smoothing: smooth the point cloud data to reduce sharp parts and abrupt points in the data, making the data smoother and more continuous.

[0086] Data reduction: Since the amount of point cloud data can be very large, directly performing registration may consume a lot of computing resources and time. Therefore, it is necessary to reduce the point cloud data to reduce the amount of data and improve the efficiency of registration.

[0087] Determine the registration reference stage: Select a stable and accurate point cloud as the reference for registration. You can choose the point cloud where the common target is located as the reference, or choose other point clouds with good stability as the reference. The principle of selecting the reference is to ensure the accuracy and stability of the registration.

[0088] Selecting appropriate registration methods and feature extraction and matching algorithms can improve the accuracy and efficiency of registration. Common registration methods include ICP algorithm, NDT algorithm, transformation estimation method, etc. Among them, ICP algorithm is one of the most commonly used point cloud registration algorithms, which can quickly converge to the optimal solution and is suitable for large-scale point cloud processing.

[0089] In the registration process, rigid body transformation estimation needs to be performed on two point clouds, that is, a transformation matrix is found to make the two point clouds correspond to each other. Rigid body transformation estimation can be solved by least squares method and other algorithms. Through the estimation of the rigid body transformation matrix, the two point clouds can be spliced.

[0090] After completing the registration, the quality of the registration result needs to be evaluated and corrected. Quality evaluation mainly checks the accuracy of the registration result, including checking whether there are obvious deviations and discontinuous places. If there are problems, they need to be corrected. The correction method can be based on manual operation or use automated methods for repair.

[0091] The computing scheme of point cloud data includes data preprocessing, key point extraction, feature description, point cloud registration, three-dimensional reconstruction and data visualization of point cloud data, wherein:

[0092] Data preprocessing: mainly clean and filter the obtained point cloud data to eliminate noise and irrelevant data. This step usually includes data filtering and data reduction. Data filtering can use smoothing filtering, median filtering and other methods to eliminate noise and outliers. Data reduction can use Euclidean distance, normal distance and other methods to reduce data volume and improve computing efficiency;

[0093] Key point extraction: extract key points from point cloud data, which can use methods similar to SIFT, SURF and other methods in two-dimensional image processing. These methods can extract local features such as curvature, normal direction and surface material from point cloud. In addition, there are some algorithms specifically for point cloud data, such as ISS3D, Harris3D, which can extract key points from point cloud.

[0094] Feature Description: For the extracted key points, it is necessary to describe their features. Common feature description algorithms include PFH, FPFH, etc. These algorithms can calculate the distribution of points around the key points, thereby obtaining the feature description of the key points.

[0095] Point Cloud Registration: Point cloud registration is the process of aligning multiple point cloud data. Common point cloud registration algorithms include ICP (Iterative Closest Point) and its improved algorithms, NDT (Normal Distribution Transform) and its improved algorithms, etc. These algorithms can find the best transformation matrix between two point clouds, so that the corresponding points between them reach the minimum error.

[0096] Three-dimensional reconstruction: Three-dimensional reconstruction is the process of generating a three-dimensional model from point cloud data. Common three-dimensional reconstruction algorithms include point cloud triangulation and voxel grid methods. These algorithms can convert point cloud data into a three-dimensional model for subsequent analysis and processing.

[0097] Data visualization: Data visualization is the process of visualizing processed point cloud data. Common visualization methods include point cloud direct drawing and voxel grid visualization. These methods can convert processed point cloud data into images or models for observation and analysis.

[0098] According to the point cloud data obtained by the non-contact three-dimensional laser scanner, the coordinate values of the point cloud data are relative position coordinates relative to the scanner. When performing aerial triangulation, it is necessary to redefine the absolute coordinate system. According to the current existing aerial triangulation method, the calculation process is re-designed by organic combination, and the error caused by aerial triangulation is minimized or eliminated to the greatest extent. The calculation process can be roughly divided into four parts, which are fitting the maximum plane at the beam end and solving the plane equation, solving the angle between the plane normal vector and the absolute coordinate system normal vector, constructing a three-dimensional rotation matrix, transposing the point cloud data, and determining the center point of the point cloud model, including:

[0099] Select the beam bottom end target point, support line target point and its surrounding 200 nearest points, and use the RANSAC plane segmentation method to obtain the maximum plane equation based on the beam bottom end target point, support line target point and its surrounding 200 nearest points;

[0100] Obtain the maximum plane and the absolute coordinate system normal vector, construct a three-dimensional rotation matrix, and solve the plane angle;

[0101] Determine the initial center point of the model, then construct the reference coordinate system based on the coordinate system of the initial center point, and finally transpose the whole point cloud data to the reference coordinate system.

[0102] Through the above steps, the point cloud model is accurately converted to the absolute position coordinate system, and the coordinate zero point is the midpoint of the line connecting the two beam bottom bevel mark points of the beam end. Then, according to the measurement requirements, the point cluster at the required position is selected, and the measurement average value is accurately calculated.

[0103] Specifically, the point cloud processing stage: using point cloud filtering, point cloud registration and point cloud segmentation technology to preliminarily process the point cloud data. Among them, point cloud filtering can remove noise and invalid points to improve data quality; point cloud registration can align multiple point cloud data to obtain more comprehensive and accurate information; point cloud segmentation can divide point cloud into different objects or parts, which is convenient for subsequent processing and analysis;

[0104] Feature extraction stage: extract key points from point cloud data and describe their features. The extracted features include edges, corner points, normal vectors, etc. These features will be used for geometric information reconstruction and calculation.

[0105] Circle center position estimation stage: to realize high-precision reconstruction and calculation, key point circle center position estimation is needed. A fitting circle model is used to accurately estimate the circle center position. The fitting circle model uses the least squares method to calculate the best circle center position.

[0106] Plane fitting and optimization stage: for the case of needing to fit key planes, use least squares method, RANSAC and other optimization algorithms to fit plane model, and optimize the parameters of the plane model by minimizing the distance from the point to the plane, to obtain the best plane fitting effect.

[0107] Data processing and algorithm optimization stage: in order to improve the processing efficiency and accuracy, innovative data processing methods and algorithm optimization strategies are used. GPU acceleration is used for incremental processing of large-scale data; deep learning and other methods are used to extract features and perform model fitting.

[0108] Example 2

[0109] Example 1 uses a three-dimensional scanner to scan the prefabricated box girder to obtain point cloud data of the prefabricated box girder, splices the point cloud data of the prefabricated box girder to generate a model of the prefabricated box girder, and example 2 automatically measures the model of the prefabricated box girder and collects data generated during the process of generating the model of the prefabricated box girder to determine the acceptance result of the prefabricated box girder.

[0110] In step 3, after the data splicing is completed, the operator uploads the spliced data and special marker point coordinates to the intelligent measurement integrated platform system. The system will automatically measure the current model, generate acceptance data and export a complete acceptance table. At the same time, the system supports historical data query management and operator account management, providing convenient and effective data access interfaces and tools. It meets the requirements of scientific organization and storage of data, efficient acquisition and maintenance of data. The intelligent measurement integrated platform system includes:

[0111] Model optimization and compression: Advanced model optimization techniques such as pruning, quantization and distillation are used to reduce model size and improve inference speed. Distributed learning methods are used to divide the model into multiple parts, distributed training and deployment to reduce transmission burden. Parallel computing techniques are used to fully utilize multi-core processors and GPUs to improve system computing speed. Distributed systems are constructed to distribute tasks to multiple computing nodes to improve overall system performance.

[0112] Incremental data transmission and processing: Incremental data transmission method is used to transmit only the changed data to reduce transmission volume. Block processing of LAS files is implemented to load and process only the data needed at the moment, improving file processing efficiency. Efficient data storage engines such as distributed file systems or databases are used to improve data reading and storage speed. Index technology is introduced to accelerate the retrieval and query operations of large data such as LAS files.

[0113] User interface optimization: Design a user-friendly interface to provide intuitive data display and operation interface for users to understand and control measurement data in real time. Complex data information can be visualized through charts, animations, reports, etc. Intuitive and simple user interaction design is adopted to reduce user operation steps. Visualization technology is used to present complex measurement results in a graphical way, making it easier for users to understand. Automatic processes are introduced to reduce user manual operations and improve system automation.

[0114] Through incremental data transmission and processing innovation, the system only transmits the changed data, reducing the transmission volume and improving the data transmission efficiency. Automatic process and intelligent recommendation innovation introduces automatic process and intelligent recommendation mechanism to reduce user manual operation and improve system intelligence. Model optimization and compression technology innovation uses advanced model optimization techniques to reduce model size and improve inference speed, solving the problem of large model transmission and improving system performance and reducing the demand for computing resources.

[0115] In step 4, in the process of processing the prefabricated box girder point cloud data, the ICP algorithm is used to match different point clouds, in order to ensure the matching of different point cloud data, the abnormal point cloud data greater than the set distance threshold is deleted, the point cloud information of the prefabricated box girder is determined, after the point cloud data splicing is completed, according to the model of the prefabricated box girder, the contour information of the prefabricated box girder is determined, the point cloud information of the prefabricated box girder is expressed by the point cloud matching error abnormal coefficient, and the contour information of the prefabricated box girder is expressed by the cross section contour deviation coefficient and the interval deviation coefficient.

[0116] The acquisition logic of the point cloud matching error abnormal coefficient is that different point cloud data sets in the same coordinate system are obtained, and the different point cloud data sets in the same coordinate system are marked as: , wherein n=1, 2, 3, …, N, N is a positive integer, n is the point cloud data set number of the prefabricated box girder scanned by the three-dimensional scanner at different positions, , k=1, 2, 3, …, K, K is a positive integer, k is the number of point cloud data in different point cloud data sets;

[0117] Based on the matching of two different point cloud data using the ICP algorithm, the error between the point cloud data in the different point cloud data set pair is calculated by the Euclidean distance, the point cloud error threshold is set, the point cloud data between the point cloud data in the different point cloud data set pair which is greater than the point cloud error threshold is obtained, the number of point cloud data greater than the point cloud error threshold is obtained, and the number of point cloud data greater than the point cloud error threshold is marked as: , wherein m is the number of different point cloud data set pairs, m=1, 2, 3, …, M, M is a positive integer, and M=N-1;

[0118] The total number of point cloud data deleted in the point cloud data splicing process is obtained, and the total number of point cloud data deleted in the point cloud data splicing process is marked as: , wherein ;

[0119] It should be noted that a reference point cloud (usually the first collected point cloud) is first selected, and then matched with the second point cloud. After matching, the second point cloud is matched with the third point cloud, and so on. Once all the point clouds are aligned through iterative optimization and registration, they are fused together to form a complete three-dimensional model. In the matching process, if the error of some points is large (for example, exceeds the preset threshold), these points may be considered as matching error points and may be deleted or excluded. The point cloud error threshold is set by professional staff in the field, and will not be described here.

[0120] The point cloud matching error abnormal coefficient is calculated, and the calculation formula is: ; wherein is the point cloud matching error abnormal coefficient, The number of point cloud data in different point cloud data sets.

[0121] As can be seen from the formula, the greater the point cloud matching error anomaly coefficient, the greater the proportion of abnormal points or point cloud data with large errors in the process of point cloud data splicing, indicating that the quality of splicing may be poor, and more point cloud data is deleted, which may result in large errors in the production process of the prefabricated box girder.

[0122] The acceptance of the prefabricated box girder requires accurate measurement data to ensure that it meets the design and quality standards. When using a three-dimensional scanner to obtain point cloud data, point cloud data from multiple scanning positions needs to be spliced together to form a complete three-dimensional model. If there is a large point cloud matching error in the splicing process, the spliced box girder model may be misaligned, distorted or have other geometric inconsistencies. The point cloud matching error anomaly coefficient can help evaluate the quality of the splicing process and the measurement accuracy. If the coefficient is too large, it indicates that the splicing error is large, which may result in the generated model not accurately reflecting the actual size and shape of the prefabricated box girder, thereby affecting the acceptance result.

[0123] There may be error points or abnormal points in the point cloud data due to differences in scanning angle, scanning device or scanning environment. Abnormal points refer to points that do not conform to the actual structure and are usually identified and deleted during the splicing process. If the point cloud matching error anomaly coefficient is high, it indicates that the proportion of abnormal points is large, which may indicate that there are quality control problems in the production process, such as irregular surface or defects of the box girder. By monitoring this coefficient, these potential problems can be effectively discovered, and the production process can be further improved or corrected.

[0124] The three-dimensional model of the prefabricated box girder is the basis for subsequent structural analysis, load capacity evaluation and practical application. If a large number of errors or abnormal points occur during the splicing process, the accuracy and reliability of the model will be affected, which will in turn affect the structural safety analysis. The point cloud matching error anomaly coefficient can quantify the splicing error and help determine whether further optimization of the splicing process or data cleaning is needed to ensure the accuracy of the three-dimensional model and ultimately ensure that the quality of the box girder meets the standards.

[0125] The acquisition logic of the cross-section profile deviation coefficient is as follows: after the point cloud data of the prefabricated box girder is spliced, a three-dimensional point cloud model of the prefabricated box girder is obtained, the point cloud data of different sections of the prefabricated box girder three-dimensional point cloud model is obtained, the point cloud data of different sections is curve-fitted, and two-dimensional cross-section curves of different sections are generated. Different cross-section curves are marked as: where i = 1, 2, 3, …, I, I is a positive integer, and i is the number of different sections selected;

[0126] It should be noted that in order to extract the cross section of the box girder, a set of point cloud data is cut at the predetermined cross section position, and the point cloud data is extracted on the specified plane by "cutting" the three-dimensional point cloud data, which means finding the intersection of the plane and the point cloud data by mathematical method (such as normal vector calculation), obtaining the point cloud data on the plane, including selecting a plane intersecting with the point cloud, obtaining the points on the plane, determining the direction of the cross section according to the calculation of the normal vector, extracting the point cloud parallel to the direction, and the curve fitting method of the point cloud data of different cross sections includes least square fitting and spline interpolation method, and different types of curve models are selected according to the actual situation to fit the data, such as circular, elliptical or polynomial curve, etc., which will not be repeated here.

[0127] The two-dimensional cross section standard curve of different cross sections is set, and the two-dimensional cross section standard curve of different cross sections is marked as: The distance deviation between each measurement point of the two-dimensional cross section curve of different cross sections and the two-dimensional cross section standard curve is calculated by Euclidean distance, and the distance deviation between each measurement point of the two-dimensional cross section curve of different cross sections and the two-dimensional cross section standard curve is marked as: Wherein, j=1, 2, 3, …, J, J is a positive integer, and j is the number of each measurement point between the two-dimensional cross section curve of different cross sections and the two-dimensional cross section standard curve.

[0128] It should be noted that the two-dimensional cross section curve of different cross sections is the actual cross section obtained by fitting the point cloud data, and the cross section curve is fitted by scanning the point cloud of a certain cross section, which represents the actual shape of the box girder at the cross section, and the two-dimensional cross section standard curve of different cross sections is the ideal cross section shape defined in advance, which is usually obtained according to the design requirements or standard specifications, and the measurement point is a specific point with spatial coordinates extracted from the point cloud data. By comparing the measurement points of the two-dimensional cross section standard curve and the two-dimensional cross section curve, the deviation of the cross section can be quantified.

[0129] The total deviation of different cross sections is calculated, and the calculation formula is: Wherein, is the total deviation of the i-th cross section of the prefabricated box girder.

[0130] The average value and standard deviation of the total deviation of different cross sections of the prefabricated box girder are calculated, and the average value and standard deviation of the total deviation of different cross sections of the prefabricated box girder are marked as: And Wherein, , ;

[0131] The cross section profile deviation coefficient is calculated, and the calculation formula is: Wherein, is the cross section profile deviation coefficient.

[0132] It can be seen from the formula that the greater the cross-section profile deviation coefficient, the greater the deviation amplitude of different cross-sections of the prefabricated box girder, indicating that there may be a large error in the production process of the prefabricated box girder.

[0133] Specifically, if the cross-section profile deviation coefficient is large, it may mean that there is a large geometric error in the production process, especially in the production process of the prefabricated box girder. This error may be caused by factors such as mold deviation, material unevenness, temperature change, etc., resulting in changes in the shape of the box girder;

[0134] The cross-section profile deviation coefficient may also be affected by the quality of the point cloud data. If there is a large error in the splicing process of the point cloud data, or the scanning accuracy is not high, it may lead to a large error in the cross-section fitting, and further lead to a large deviation coefficient;

[0135] It may be necessary to strengthen the monitoring and adjustment in the production process to ensure that the geometric shape of the box girder better meets the design requirements. For this purpose, it may be necessary to adjust the mold, improve the production process, or use more accurate technology in the measurement and data processing stage.

[0136] The acquisition logic of the spacing deviation coefficient is as follows: after the splicing of the prefabricated box girder point cloud data is completed, the positions of the special marker points in the prefabricated box girder three-dimensional point cloud model are obtained, and the position coordinates of the special marker points in the prefabricated box girder three-dimensional point cloud model are marked as: , wherein, q=1, 2, 3, …, Q, Q is a positive integer, and q is the number of special marker points;

[0137] The position coordinates of the preset special marker points are obtained, and the position coordinates of the preset special marker points are marked as: , wherein, ;

[0138] It should be noted that the special marker points are markers with known positions placed on the surface or key positions of the box girder, such as black and white targets, reflective balls or other types of positioning markers. When performing three-dimensional scanning, the special marker points can help accurately determine the spatial position and attitude of the point cloud data. When the data of multiple scanning stations needs to be spliced, the marker points can help align different point cloud data to ensure that the overall model after splicing is accurate.

[0139] The spacing deviation coefficient is calculated, and the calculation formula is: ; wherein, is the spacing deviation coefficient.

[0140] It can be seen from the formula that the greater the spacing deviation coefficient, the greater the spacing deviation between the special marker points in the prefabricated box girder model, indicating that there may be a large error in the production process of the prefabricated box girder.

[0141] It should be noted that the spacing between the actual position of the mark point and the design position has a significant difference, indicating that the positioning deviation of the mark point setting may exist, or the error caused by the equipment precision or environmental interference in the measurement process, reflecting that the actual size of the prefabricated box girder deviates from the design requirement, if the spacing deviation coefficient continuously increases in multiple measurements, it indicates that there may be cumulative deviation in the production process of the prefabricated box girder, and the production process or equipment may need to be checked and adjusted.

[0142] Through comprehensive analysis of the point cloud information and the contour information of the prefabricated box girder, the point cloud matching error abnormality coefficient, the cross-section contour deviation coefficient and the spacing deviation coefficient are weighted and calculated to construct an acceptance evaluation model of the prefabricated box girder, and an acceptance evaluation coefficient is generated, and the calculation formula is: ; wherein, is the acceptance evaluation coefficient, , , the proportion coefficients of the point cloud matching error abnormality coefficient, the cross-section contour deviation coefficient and the spacing deviation coefficient respectively, , , all greater than 0.

[0143] As can be seen from the formula, the greater the proportion coefficients of the point cloud matching error abnormality coefficient, the cross-section contour deviation coefficient and the spacing deviation coefficient, the greater the acceptance evaluation coefficient, which indicates that there may be a significant deviation between the box girder model determined by the three-dimensional scanner and the design standard, affecting the evaluation of the quality of the box girder.

[0144] The threshold value of the acceptance evaluation coefficient is set, and the acceptance evaluation coefficient is compared with the threshold value of the acceptance evaluation coefficient, if the acceptance evaluation coefficient is greater than the threshold value of the acceptance evaluation coefficient, a warning signal is generated, indicating that the prefabricated box girder produced at present has a large error, if the acceptance evaluation coefficient is less than the threshold value of the acceptance evaluation coefficient, no warning signal is generated, indicating that the prefabricated box girder produced at present has high accuracy.

[0145] In the intelligent acceptance of the outer dimensions of the prefabricated box girder, the laser scanner is used to scan each surface of the box girder to obtain high-precision three-dimensional data, the dense point cloud reverse modeling method, the point cloud air three calculation algorithm and the improved RANSAC plane fitting algorithm are introduced, the accurate measurement of the cross section of the railway box girder is realized, and the point cloud information and the contour information of the prefabricated box girder are collected to quantify the possible significant deviation between the box girder model and the design standard, the present application realizes the rapid, accurate and non-contact measurement of the whole cross section of the railway box girder, improves the automation and intelligent level of the measurement, and helps to improve the efficiency of the whole cross section acceptance of the railway box girder.

[0146] Embodiment 3

[0147] Figure 2The application provides a structure diagram of a railway box girder full-section data monitoring system based on three-dimensional scanning, and specifically comprises a data acquisition module, a model splicing module, a data uploading module and an acceptance evaluation module.

[0148] The data acquisition module is used for multiple scanning of the railway box girder by a three-dimensional scanner to capture point cloud data of the full section.

[0149] The model splicing module is used for restoring three-dimensional coordinates of the point cloud by space three calculation, extracting key geometric points in the point cloud, accurately determining the center positions of the points by an algorithm, fitting the planar part in the scanning area, and optimizing the splicing error between the point cloud data.

[0150] The data uploading module is used for uploading the complete spliced point cloud data and special marker point information on the box girder to an intelligent measurement integrated platform.

[0151] The acceptance evaluation module is used for obtaining point cloud information and contour information of the prefabricated box girder from the point cloud model, calculating the deviation of the box girder model from the design standard, generating an acceptance evaluation coefficient, and providing a scientific basis for quality control.

[0152] The above formulas are dimensionless numerical calculations, the formulas are obtained by software simulation of a large amount of data to obtain a formula of the nearest real situation, and the preset parameters in the formula are set by a person skilled in the art according to the actual situation.

[0153] The above embodiments can be realized by software, hardware, firmware or any combination thereof, in whole or in part. When realized by software, the above embodiments can be realized in the form of a computer program product in whole or in part. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the processes or functions described in the embodiments of the application are generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable devices. The computer instructions can be stored in a computer readable storage medium or transmitted from one computer readable storage medium to another, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through a wired or wireless (such as infrared, wireless, microwave, etc.) manner. The computer readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center and the like containing one or more available medium collections. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD) or a semiconductor medium. The semiconductor medium can be a solid state disk.

[0154] It should be understood that the size of the sequence number of the above processes does not mean the order of execution in various embodiments of the present application, and the execution order of the processes should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0155] Those skilled in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be realized by electronic hardware or a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solutions. Those skilled in the art can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0156] Those skilled in the art can clearly understand that, for the convenience and brevity of the description, the specific working processes of the above-described system, device and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be repeated here.

[0157] In several embodiments provided in the present application, it should be understood that the disclosed system, device and method can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or other forms.

[0158] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application or the part of the technical solutions that essentially contribute to the prior art or the part of the technical solutions can be embodied in the form of software products. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0159] The above merely provides the specific implementation of the present application, but the protection scope of the present application is not limited to this. Any person skilled in the art can easily think of the changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A method for monitoring full-section data of railway box girders based on a 3D scanner, characterized in that, Specifically, the following steps are included: S1: Data of the entire cross-section of the railway box girder is collected using a 3D scanner. By scanning each part of the box girder multiple times, the data collection process for the entire cross-section of the railway box girder is determined. S2: Process the collected full-section data of railway box girders. Through the aerial triangulation calculation process based on dense point cloud data, accurate estimation of the center position of key points, and fitting and optimization of key planes, the point cloud data is stitched together. S3: The operator uploads the spliced ​​data and the coordinates of special marker points to the intelligent measurement integration platform system. The system will automatically measure the current model, automatically generate acceptance data, and export a complete acceptance form. S4: Collect point cloud and contour information of precast box girders, construct an acceptance evaluation model for precast box girders, and quantify the deviation between the box girder model and the design standards; The point cloud information of the precast box girder is represented by the point cloud matching error anomaly coefficient, and the calculation formula is as follows: ;in, The point cloud matching error anomaly coefficient, The number of point cloud data in different point cloud datasets, This represents the total number of point cloud data points deleted during the point cloud data stitching process. The profile information of the precast box girder is represented by the cross-sectional profile deviation coefficient and the spacing deviation coefficient. The formula for calculating the cross-sectional profile deviation coefficient is as follows: ;in, This is the cross-sectional profile deviation coefficient. This represents the average of the total deviations of different sections of the precast box girder. This represents the standard deviation of the total deviation of different sections of the precast box girder; The logic for obtaining the spacing deviation coefficient is as follows: After the precast box girder point cloud data is stitched together, the positions of special marker points in the three-dimensional point cloud model of the precast box girder are obtained, and the position coordinates of the special marker points in the three-dimensional point cloud model of the precast box girder are marked as follows: ,in, q = 1, 2, 3, ..., Q, where Q is a positive integer and q is the number of a special marker point; Obtain the position coordinates of the preset special marker points, and mark the position coordinates of the preset special marker points as follows: ,in, ; The spacing deviation coefficient is calculated using the following formula: ;in, This is the spacing deviation coefficient; Construct an acceptance evaluation model for precast box girders, including: By comprehensively analyzing the point cloud and contour information of precast box girders, the point cloud matching error anomaly coefficient, cross-sectional contour deviation coefficient, and spacing deviation coefficient are weighted and calculated to construct an acceptance evaluation model for precast box girders, generating acceptance evaluation coefficients. The calculation formula is as follows: ;in, This is the acceptance evaluation coefficient. , , These are the proportional coefficients for the point cloud matching error anomaly coefficient, the cross-sectional profile deviation coefficient, and the spacing deviation coefficient, respectively. , , All are greater than 0; Set an acceptance evaluation coefficient threshold, compare the acceptance evaluation coefficient with the acceptance evaluation coefficient threshold. If the acceptance evaluation coefficient is greater than the acceptance evaluation coefficient threshold, an early warning signal is generated, indicating that there is a large error in the precast box girder currently being produced. If the acceptance evaluation coefficient is less than the acceptance evaluation coefficient threshold, no early warning signal is generated, indicating that the accuracy of the precast box girder currently being produced is high.

2. The method for monitoring full-section data of railway box girders based on a 3D scanner according to claim 1, characterized in that, The data acquisition process for the entire cross-section of railway box girders was determined, including: The data acquisition process is divided into beam end data acquisition, bottom and web side data acquisition, beam interior data acquisition, and top slab data acquisition, among which: During beam end data acquisition, special marker points were set using standard parts at the bottom angle of the beam, the top of the beam, and the center line of the support. These marks were used as coordinate positioning points after subsequent splicing. After setting the special marker points and the positioning marker balls for splicing, a non-contact 3D laser scanner was used to acquire data at the middle position of the beam end. During the data acquisition of the bottom web plate, special marker points were set using standard parts at the span line and beam end, respectively, for coordinate positioning points after subsequent splicing. After setting the marker points, a non-contact 3D laser scanner was used to acquire web plate data at a position 6m away from the beam end on the beam side, and a non-contact 3D laser scanner was used to acquire bottom plate data at the middle position on the beam side. During the data acquisition inside the beam, two locations were selected as data acquisition points, which are close to the beam end and where the positioning marker ball placed on the beam end can be observed. At the same time, the non-contact 3D laser scanner was adjusted to ensure that there would be no overexposure before the non-contact 3D laser scanner was used for data acquisition. During the data acquisition of the top plate, two locations were selected as data acquisition points, one on each side of the beam end, where the positioning marker ball placed on the beam end could be observed. After determining the two data acquisition points, a non-contact 3D laser scanner was used to acquire the data.

3. The method for monitoring full-section data of railway box girders based on a 3D scanner according to claim 2, characterized in that, The aerial triangulation process based on dense point cloud data includes: The absolute position coordinate system is redefined, and the calculation process is redesigned based on the existing aerial triangulation calculation methods to minimize or eliminate the errors caused by aerial triangulation calculations. This includes fitting the maximum plane at the beam end and obtaining the plane equation, obtaining the angle between the plane normal vector and the absolute coordinate system normal vector, constructing a three-dimensional rotation matrix, transposing the point cloud data, and determining the center point of the point cloud model.

4. The method for monitoring full-section data of railway box girders based on a 3D scanner according to claim 3, characterized in that, Accurate estimation of key point center positions and key plane fitting and optimization, including: A fitted circular model is used to accurately estimate the center position. The least squares method is used to calculate the optimal center position. The parameters of the planar model are optimized by minimizing the distance from the point to the plane to obtain the best planar fitting effect.

5. The method for monitoring full-section data of railway box girders based on a 3D scanner according to claim 4, characterized in that, The operator uploads the stitched data and the coordinates of the special marker points to the intelligent measurement integration platform system, including: The intelligent measurement integration platform system adopts model optimization technology and uses distributed learning methods to divide the model into multiple parts. It utilizes parallel computing technology to distribute tasks to multiple computing nodes by building a distributed system. An incremental data transfer method is adopted, which only transmits the data that has changed, realizes the block processing of LAS files, loads and processes only the part of the data that is currently needed, uses a distributed file system or database and introduces indexing technology to accelerate the retrieval and query operations of LAS files.

6. The method for monitoring full-section data of railway box girders based on a 3D scanner according to claim 5, characterized in that, Point cloud information of precast box girders includes: The logic for obtaining the point cloud matching error anomaly coefficient is as follows: obtain different point cloud datasets in the same coordinate system, and mark the different point cloud datasets in the same coordinate system as: Where n = 1, 2, 3, ..., N, N is a positive integer, and n is the point cloud dataset number of different parts of the precast box girder scanned by the 3D scanner. k = 1, 2, 3, ..., K, where K is a positive integer and k is the number of the point cloud data in different point cloud datasets; The ICP algorithm is used to match two different point cloud datasets. The error between the data points in the Euclidean distance is calculated. A point cloud error threshold is set, and point cloud data with errors greater than the threshold are identified. The number of point cloud data points with errors greater than the threshold is then marked as follows: Where m is the number of the different point cloud dataset pairs, m=1, 2, 3, ..., M, M is a positive integer, M=N-1; Obtain the total number of point cloud data deleted during the point cloud data stitching process, and mark the total number of point cloud data deleted during the point cloud data stitching process as: ,in, ; Calculate the anomaly coefficient of point cloud matching error.

7. The method for monitoring full-section data of railway box girders based on a 3D scanner according to claim 6, characterized in that, The outline information of the precast box girder includes: The logic for obtaining the cross-sectional profile deviation coefficient is as follows: After the precast box girder point cloud data is stitched together, a three-dimensional point cloud model of the precast box girder is obtained. Point cloud data of different sections of the three-dimensional point cloud model of the precast box girder are obtained. Curve fitting is performed on the point cloud data of different sections to generate two-dimensional cross-sectional curves of different sections. The two-dimensional cross-sectional curves of different sections are marked as follows: Where i = 1, 2, 3, ..., I, I is a positive integer, and i is the number of the selected section; Set up two-dimensional cross-sectional standard curves for different cross-sections, and label the two-dimensional cross-sectional standard curves for different cross-sections as follows: The distance deviation between each measurement point of the two-dimensional cross-sectional curve and the two-dimensional cross-sectional standard curve of different cross-sections is calculated using Euclidean distance, and the distance deviation between each measurement point of the two-dimensional cross-sectional curve and the two-dimensional cross-sectional standard curve of different cross-sections is marked as: , where j = 1, 2, 3, ..., J, J is a positive integer, and j is the number of each measurement point between the two-dimensional cross-sectional curve and the two-dimensional cross-sectional standard curve of different cross sections; The total deviation for different cross sections is calculated using the following formula: ;in, This represents the total deviation of the i-th section of the precast box girder; Calculate the average and standard deviation of the total deviation of different sections of the precast box girder, and label the average and standard deviation of the total deviation of different sections of the precast box girder as follows: and ;in, , ; Calculate the cross-sectional profile deviation coefficient.

8. A railway box girder full-section data monitoring system based on a 3D scanner, used to implement the railway box girder full-section data monitoring method based on a 3D scanner as described in any one of claims 1-7, characterized in that, It includes a data acquisition module, a model stitching module, a data upload module, and an acceptance evaluation module; The data acquisition module is used to scan the railway box girder multiple times using a 3D scanner to capture point cloud data of the entire cross section; The model stitching module is used to reconstruct the three-dimensional coordinates of point clouds from dense point cloud data from multiple perspectives using aerial triangulation, extract key geometric points in the point cloud, accurately determine the center position of these points using algorithms, fit the planar part in the scanned area, and optimize the stitching error between point cloud data. The data upload module is used to upload the spliced ​​complete point cloud data and the special marker point information on the box girder to the intelligent measurement integration platform; The acceptance evaluation module is used to obtain the point cloud information and contour information of the precast box girder from the point cloud model, calculate the deviation between the box girder model and the design standard, generate acceptance evaluation coefficients, and provide a scientific basis for quality control.

Images