Automatic measurement system for flatness of large-size prefabricated u-beam based on laser scanning

Abstract

The application relates to the technical field of length measurement, in particular to a large-size prefabricated U-beam flatness automatic measurement system based on laser scanning. The system comprises the following steps: acquiring a three-dimensional point cloud to determine a main shaft direction; presetting a window to fit the point cloud to acquire a normal vector, and calculating a curvature to determine a section through the curvature and a slice interval; determining local continuity and global continuity based on the distance in the section neighborhood, the main shaft difference and the global distance distribution of the section; determining a topological stability coefficient based on the point cloud of the section and the whole, the perimeter difference, determining a standard section based on the three, and determining flatness by weighting the distance between the section and the standard section through a standard type score. The application improves the calculation efficiency.

Description

Technical Field

[0001] This application relates to the field of length measurement technology, specifically to an automatic measurement system for the flatness of large-size prefabricated U-beams based on laser scanning. Background Technology

[0002] Precast U-beams are a type of prestressed concrete bridge structure with advantages such as light weight, low structural height, and low noise, and are frequently used in urban rail transit projects such as subways and light rail. Their cross-section is a transversely open channel-shaped structure, consisting of a bottom plate, two side webs, and a top flange. The bottom plate, located at the bottom of the U-shape, directly bears the wheel-rail load of the rail transit vehicles. The two side webs extend upwards from both ends of the bottom plate, forming the two side walls of the U-shape. The top flange extends horizontally outwards from the top of the webs, serving both as a structural load-bearing structure and a track-laying platform. Because U-beams may exhibit surface unevenness during prefabrication, such as localized bulges and depressions, these defects directly affect the accuracy of track laying and the safety of train operation. Therefore, U-beams must undergo flatness testing before leaving the factory.

[0003] Existing flatness detection methods based on 3D laser scanning first acquire 3D point cloud data of the surface using a laser scanner, then fit a reference plane and calculate the vertical distance from each point to the reference plane as a flatness index. For planar structures like the base plate, the least squares method can be used for planar fitting, while for curved surfaces like the web, B-splines can be used for surface fitting. However, during the precasting process of large-size U-beams, uneven heat of hydration in the concrete and thermal stress after demolding can cause flatness differences at different locations. The shrinkage at the ends is greater than in the middle, and local air bubbles and voids may appear. When uniformly fitting the reference plane, these local anomalies will affect the overall fitting result, causing the reference plane itself to deviate from the true shape, thus leading to inaccurate flatness calculations. Summary of the Invention

[0004] To address the technical problem of inaccurate flatness calculation, this application provides an automatic flatness measurement system for large-size precast U-beams based on laser scanning. The specific technical solution adopted is as follows: This application proposes an automatic measurement system for the flatness of large-size precast U-beams based on laser scanning. The system includes the following modules: The point cloud acquisition module is used to acquire three-dimensional point clouds by laser scanning of the U-beam and determine its principal axis direction; The section selection module is used to preset multiple windows along the principal axis direction, and obtain the plane and normal vector corresponding to the window by fitting the point cloud of adjacent windows; calculate the curvature by using the normal vectors of two adjacent windows; limit the slice interval by setting the tolerance of the angle between the preset normal vectors, and then calculate the slice interval by combining the curvature; determine the section under different coordinates of the principal axis by the slice interval; The standard section determination module is used to determine local continuity based on the distance and principal axis coordinate differences between sections in a preset neighborhood; determine global continuity based on the discrete distribution of distances between global windows; determine topological stability based on the deviation of the point cloud number and perimeter of the section from the total number of points in all sections; and construct standard sections by calculating standardization scores through local continuity, global continuity, and topological stability. The flatness acquisition module obtains the flatness by weighting the distance deviation between each cross section and the standard cross section through a standardity score.

[0005] In the above scheme, this application uses a cross-sectional comparison method to detect flatness. The standard cross-section is automatically generated through data-driven calculation, eliminating the need for selecting fitting types and parameters. Compared to traditional datum surface fitting methods where different parameters lead to significant differences in detection results, this significantly improves the adaptability of flatness detection. This application can also identify and reduce the weight of abnormal cross-sections. Compared to traditional datum surface fitting methods that apply equal weight to all point clouds, local defects can affect the fitting effect of the datum surface, causing it to deviate from its true shape. This scheme, through a standard scoring mechanism, ensures that the construction of the standard cross-section is not affected by local defects, and the calculated overall flatness can more accurately reflect the quality level of normal areas. This application uses an adaptive slicing mechanism to dynamically adjust the slicing interval based on the axial curvature. Because the fixed-interval slicing method uses a uniform interval across the entire beam length, it can lead to oversampling in smooth areas and undersampling in deformable areas. This application increases the interval in smooth areas to improve efficiency and decreases the interval in high-curvature areas to ensure accuracy, significantly improving calculation accuracy.

[0006] In one embodiment, the principal axis direction is the direction with the longest point cloud distribution determined by principal component analysis of all point clouds, and the plane perpendicular to the principal axis is used as the cross-sectional plane.

[0007] In one embodiment, the curvature is calculated as follows: The normal vector is multiplied by its adjacent normal vector, and the angle between the two vectors is obtained by taking the inverse cosine of the dot product. The ratio of the angle to the window width is used as the curvature.

[0008] In one embodiment, the expression for the slice interval is: , This represents the tolerance of the angle between the defined normal vectors. Let represent the curvature of the i-th window. Represents a very small positive number. This represents the slice interval of the i-th window. This indicates the preset maximum slice interval. This indicates the preset minimum slice interval.

[0009] In one embodiment, the method for determining the cross-section at different coordinates of the principal axis by slicing interval is as follows: Along the main axis, the right side of the first window is taken as the first slice section. The curvature of the two windows on the left and right sides of the slice section is calculated to obtain the slice interval. For the first slice section, the position of the slice interval is taken as the second slice section, and so on to obtain all slice sections.

[0010] In one embodiment, the local continuity is negatively correlated with the shape change rate; the shape change rate is positively correlated with the distance between the cross section and all cross sections in its neighborhood set, and negatively correlated with the difference in principal axis coordinates between the cross section and all cross sections in its neighborhood set.

[0011] In one embodiment, the global continuity is negatively correlated with the coefficient of variation of the distance between all windows.

[0012] In one embodiment, the topological stability is negatively correlated with the deviations of the cross section and all cross sections in terms of point cloud quantity and perimeter.

[0013] In one embodiment, the standardization score is the mean of a first standardization score and a second standardization score; the first standardization score is the sum of local continuity and global continuity, and the second standardization score is topological stability.

[0014] In one embodiment, all cross-sections are sorted from largest to smallest according to their standardization scores. A predetermined number of cross-sections are grouped into a high-standard set. For each cross-section in the high-standard set, its central axis is selected, and samples are taken at equal intervals along the central axis according to the arc length. At each sampling point, a ray is emitted along the normal direction of the central axis, intersecting with the inner and outer contours of the cross-section, resulting in two distance values: the inner distance and the outer distance. The coordinates of the central axis and the normal angle of all high-standard cross-sections are weighted and averaged using the standard scores to obtain the standard central axis and the standard normal angle. The inner distance and the outer distance at each sampling point are weighted and averaged using the standard scores to obtain the standard thickness of each sampling point. The Cartesian coordinates of the inner and outer contours of each sampling point are calculated using the standard central axis, the standard normal angle, and the standard thickness.

[0015] The beneficial effects of this application are as follows: This application uses a cross-sectional comparison method to detect flatness. The standard cross-section is automatically generated through data-driven calculations, eliminating the need for selecting fitting types and parameters. Compared to traditional datum surface fitting methods, where different parameter selections can lead to significant differences in detection results, this significantly improves the adaptability of flatness detection. This application can also identify and reduce the weight of abnormal cross-sections. Unlike traditional datum surface fitting methods that apply equal weight to all point clouds, local defects can affect the fitting effect of the datum surface, causing it to deviate from its true shape. This scheme, through a standardized scoring mechanism, ensures that the construction of the standard cross-section is not affected by local defects, and the calculated overall flatness more accurately reflects the quality level of normal areas. This application uses an adaptive slicing mechanism to dynamically adjust the slicing interval based on the axial curvature. Fixed-interval slicing methods use a uniform interval across the entire beam length, leading to oversampling in smooth areas and undersampling in deformable areas. This application increases the interval in smooth areas to improve efficiency and decreases the interval in high-curvature areas to ensure accuracy, significantly improving calculation accuracy. Attached Figure Description

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

[0017] Figure 1 The flowchart illustrates an automatic measurement system for the flatness of large-size prefabricated U-beams based on laser scanning, as provided in one embodiment of this application. Detailed Implementation

[0018] An embodiment of an automatic flatness measurement system for large-size prefabricated U-beams based on laser scanning: The following description, in conjunction with the accompanying drawings, details the specific scheme of the automatic measurement system for the flatness of large-size precast U-beams based on laser scanning provided in this application.

[0019] Please see Figure 1 The diagram illustrates an embodiment of an automatic flatness measurement system for large-size precast U-beams based on laser scanning, which includes the following modules: The point cloud acquisition module uses a laser scanner to collect 3D point cloud data of large-sized precast U-beams. Statistical filtering is applied to the point cloud data to remove outliers and noise, with the threshold set to the average distance plus three standard deviations (based on the 3σ principle). Principal component analysis (PCA) is used to determine the principal axis direction of the U-beam, i.e., the direction of the longest point cloud distribution. The point cloud coordinate system is rotated and aligned so that the x-axis is along the length of the beam, and this x-axis is designated as the principal axis. The yz plane is the cross-sectional plane perpendicular to the long axis of the beam.

[0020] At this point, point cloud data has been collected and preprocessed.

[0021] The section selection module sets a window of a preset size for point cloud analysis. To ensure that each window contains a sufficient number of samples for statistical analysis, in this embodiment, the window side length is 10 times the scanning resolution. The statistical window position is then calculated starting from the beginning along the x-axis.

[0022] For each window, the point cloud within the range of its adjacent window is extracted. The local plane equation is then fitted using the least squares method to obtain the normal vector for each window, which characterizes the overall tendency of the point cloud within the window. In this embodiment, the normal vector of the window is obtained by fitting the point cloud of each window and its next adjacent window.

[0023] For each window, the rate of change of the angle between its normal vector and the normal vector of its adjacent window is calculated and denoted as the axial curvature. The calculation method is as follows: Perform a dot product operation on the normal vector and its adjacent normal vector. Take the inverse cosine of the dot product result to obtain the angle between the two vectors (in radians). Divide this angle by the window width (in meters) to obtain the curvature, with units of radians / meter. Its physical meaning is the degree of bending of the beam per unit length at the window location. The result after the calculation ranges from -1 to 1. The adjacent normal vector is the normal vector of the window adjacent to the target window; in this embodiment, the adjacent window is the window immediately preceding it.

[0024] For an ideal straight beam, the normal vectors of all windows are parallel, the dot product of adjacent normal vectors is close to 1, the inverse cosine is close to zero, and the curvature is also close to zero. When the beam has local depressions, bulges, or overall bending, the direction of the normal vectors at different locations changes, the angle between the normal vectors at adjacent locations increases, and the curvature value increases accordingly. Locations with high curvature indicate rapid shape changes. If a relatively large slice interval is still used, the shape difference between adjacent sections will be too large, making it impossible to accurately assess continuity. Therefore, regions with high curvature need to have a smaller slice interval and use denser sampling to capture shape changes.

[0025] However, since there is a certain error in flatness, in order to ensure that the flatness detected later is within a certain error range, it is necessary to set the tolerance of the angle between the normal vectors of adjacent sections to further limit the slicing interval.

[0026] Curvature represents the rate of change of the angle between the normal vectors of the beam per unit length of the window. Let the interval between adjacent sections be... Then the angle between the normal vectors is To ensure that the shape difference between adjacent cross-sections is not too large, the included angle must not exceed the tolerance. ,Right now When a location with high curvature (rapid shape change) is detected, the slice interval is reduced and the sampling density is increased to ensure that the shape difference between adjacent sections is within an acceptable range.

[0027] Therefore, the slice interval at each window position is calculated based on the tolerance and curvature of the angle of the normal vector of adjacent sections.

[0028] Preferably, the expression for the slice interval is: , This represents the tolerance of the angle between the defined normal vectors. Let represent the curvature of the i-th window. Represents a very small positive number. This represents the slice interval of the i-th window. This indicates the preset maximum slice interval. This indicates the preset minimum slice interval.

[0029] The tolerance of the included angle of the normal vectors is limited by the flatness. In this embodiment, the flatness error is within 5mm, therefore, on a 3-meter-wide U-beam cross-section, the maximum allowable angular deviation is approximately... Radius (approximately 0.1 degrees). In reality, local defects can lead to larger local angle changes, requiring more stringent precision control. The actual value is approximately one-fifth of the theoretical value, i.e. Milliradians (approximately 0.02 degrees). To represent extremely small positive numbers, to prevent the denominator from being 0, the maximum slice interval is 5cm, and the minimum slice interval is 1cm.

[0030] From the starting point of the beam ( Starting with the first window, the right side represents the first slice position. The axial curvature is calculated from the two windows to its left and right to obtain the first slice interval. Then, starting from the next slice position, the axial curvature is calculated from the two windows to its left and right, and the slice interval is calculated, generating a series of slice position coordinates cyclically. Each slice position corresponds to a section to be extracted. Because the interval is dynamically changing, the final number of slices obtained is... It is also adaptive, with sparse slices in smooth areas to reduce computational complexity, and dense slices in high-curvature areas to better capture local defects.

[0031] For each slice location, point clouds within a certain tolerance range before and after it are extracted. The tolerance range is set to twice the scanning resolution; in this embodiment, the scanning resolution is 2.5 mm, and the tolerance range is 5 mm. A tolerance range that is too small will miss points near the boundary due to the discreteness of the point cloud, while a tolerance range that is too large will mix in points from adjacent sections, resulting in a blurred outline. The extracted 3D point cloud is projected onto the yz plane (perpendicular to the beam length direction) to obtain a 2D cross-sectional outline point cloud. The projection operation involves discarding the x-coordinate and retaining the y and z coordinates. This 2D outline point cloud represents the cross-sectional shape of the beam at that location and is used for subsequent cross-sectional comparison.

[0032] At this point, the cross-section of the U-beam has been obtained.

[0033] The standard cross-section determination module calculates the distance between two cross-sections. In this embodiment, the distance between the cross-sections is the Hausdorff distance. Specifically, for each point on a cross-section, it finds the nearest distance to another cross-section, and then sets the cross-section... The average of the shortest distances at all points is used. This average value is the one-way Hausdorff distance between the two cross-sections, measured in millimeters. Physically, it represents the average degree of deviation of the cross-sectional shape.

[0034] For each pair of cross-sections, calculate the distance. If there are M cross-sections, then combine all the distances into a single distance. The distance matrix is ​​asymmetric because the distance from section A to section B is different from the distance from section B to section A. The distance matrix contains shape similarity information between all sections, which is subsequently used to evaluate the standardization of each section.

[0035] Because directly averaging all cross sections as the standard section can be affected by local defects—such as air bubbles, voids, or chipped corners on the beam—these abnormal sections, when averaged together with normal sections, can skew the standard section. Therefore, this application first assesses the standardization of each cross section, i.e., its representativeness, and then uses a weighted average of sections with high standardization. This avoids the interference of defects.

[0036] Considering the continuity of the cross section, the standard cross section is constructed to determine the standard cross section. The standard cross section needs to have local continuity (smooth transition with the shape of the preceding and following neighbors, without abrupt changes, excluding cracks and chipped corners) and global continuity (similar to most cross sections of the whole, not in special locations, excluding ends and joints).

[0037] Therefore, for each cross section, its neighborhood set is first obtained. The neighborhood set is the cross section consisting of k cross sections before and after each cross section. In this embodiment, k is 10.

[0038] The shape change rate of two cross sections is determined based on the difference in principal axis coordinates and the distance between each cross section and every cross section in its neighborhood set. Local continuity is then calculated using the shape change rate. The smaller the shape change rate, the smoother the shape transition, and in this case, the greater the difference in principal axis coordinates between the distances between the cross sections.

[0039] Therefore, the local continuity is negatively correlated with the shape change rate; the shape change rate is positively correlated with the distance between the cross section and all cross sections in its neighborhood set, and negatively correlated with the difference in principal axis coordinates between the cross section and all cross sections in its neighborhood set.

[0040] It should be noted that positive correlation means that when one variable increases, the other variable also increases, and the two variables change in the same direction. When one variable changes from large to small or from small to large, the other variable also changes from large to small or from small to large. The specific relationship is determined by the actual application, and this application does not impose any special restrictions.

[0041] It should be noted that negative correlation means that when one variable increases, the other variable decreases accordingly, and the two variables change in opposite directions. When one variable changes from large to small or from small to large, the other variable also changes from small to large or from large to small. The specific relationship is determined by practical application, and this application does not impose any special restrictions.

[0042] Preferably, in this embodiment, the expression for local continuity is: , This represents the i-th cross section. This represents the j-th cross section in the neighborhood set of the i-th cross section. Representing the cross section With cross section The distance; Represents the principal axis coordinates of the i-th section. This represents the principal axis coordinate of the j-th cross-section within the neighborhood set of the i-th cross-section. Let i represent the neighborhood set of the i-th section. It represents half the number of sections in the neighborhood set. This represents an exponential function with the natural constant as its base. This indicates the local continuity of the i-th section.

[0043] Defects can cause abrupt changes in the cross-sectional shape relative to its neighbors, manifested as an abnormally increased rate of shape change. Therefore, neighborhood consistency testing can effectively eliminate local defects. Geometric continuity is quantified by calculating the rate of shape change of the cross-section and its neighbors. The smaller the rate of change, the smoother the shape transition. The rate of change is mapped to the 0-1 interval using an exponential function. The exponential function can more significantly penalize severe abrupt changes while having a higher tolerance for normal fluctuations.

[0044] Due to the constraints of the mold and the special nature of the demolding process, the shape of the ends of U-beams often differs systematically from that of the middle (such as end shrinkage and chamfers). Although these positions are manufactured normally, they are not suitable as standards. The coefficient of variation can identify the special nature of such geometric positions. By statistically analyzing the dispersion of the overall distance distribution, non-representative positions such as ends and joints can be eliminated, ensuring that the standard section comes from the typical part of the beam.

[0045] Therefore, global continuity is calculated based on the coefficient of variation of the distance between each cross section and all other cross sections, in order to analyze the dispersion of the overall distance distribution.

[0046] The global continuity is negatively correlated with the coefficient of variation.

[0047] Preferably, in this embodiment, the expression for global continuity is: , Let represent the coefficient of variation of the distance between the i-th cross section and all other cross sections. This represents the global continuity of the i-th section.

[0048] The coefficient of variation (CV) measures the dispersion of the distances from a cross section to all other cross sections. A small CV indicates that the distances from the cross section to all other locations are relatively uniform, and it is located at the geometric center, representing the overall shape. A large CV indicates that the distances from the cross section to different locations vary significantly, and it is located at the geometric edge or in a special position.

[0049] The sum of global continuity and local continuity is used as the first standard score.

[0050] In addition, a standard score is constructed through topological stability, meaning that a standard cross-section must possess topological stability (complete outline, normal point count and perimeter, excluding scan occlusion and missing data). Therefore, topological stability is determined by the difference in point cloud count and perimeter between each cross-section and all cross-sections.

[0051] The topological stability is negatively correlated with the differences in point cloud quantity and perimeter of the cross section and all cross sections.

[0052] Preferably, in this embodiment, the expression for topological stability is: , This represents the number of point clouds in the i-th section. This represents the median number of point clouds across all cross sections. This represents the perimeter of the i-th cross-section. This represents the median of the perimeter of all cross-sections. This represents the absolute deviation in the number of point clouds across all cross sections. This represents the absolute deviation of the perimeter of all cross-sectional profiles. This represents an exponential function with the natural constant as its base. This represents the topological stability of the i-th cross section. The absolute deviation is used to eliminate units for easier calculation; when the denominator is 0, the absolute deviation is recorded as 1.

[0053] Issues such as scan occlusion, excessively small laser incident angle, and abnormal surface reflectivity can lead to missing or sparse point clouds, while edge chipping and mold defects can result in incomplete contours, manifested as excessively large perimeters. Both of these problems affect the accuracy of comparing cross-sections with standard cross-sections. Statistical testing of topological features can automatically identify and reduce the weight of these low-quality cross-sections, preventing them from participating in standard cross-section construction and flatness evaluation. Topological stability is tested for the integrity of cross-sectional data using two geometrically invariant variables: point cloud count and perimeter. A deviation in point count from the median indicates abnormal scan sampling, while a deviation in perimeter indicates incomplete contour boundaries. The two deviations are normalized to MAD and added as a penalty term. An exponential function maps the penalty to a 0-1 score. The median and MAD are used instead of the mean and standard deviation to combat outlier interference.

[0054] Topological stability is denoted as the second standardization score, and the average of the first and second standardization scores is taken as the standardization score of the cross section.

[0055] Sort all cross-sections by standardization score from largest to smallest, and select cross-sections with scores exceeding the 70th percentile to form a high standardization set (equivalent to retaining the top 30% of cross-sections by score). For each high standardization cross-section, use the centerline of the distance variation calculator to intercept the centerline of the main part of the U-beam, obtaining a continuous single-parameter curve. Then, sample at equal intervals along the centerline by arc length. At each sampling point, emit a ray along the normal direction of the centerline, intersecting with the inner and outer contours of the cross-section, obtaining two distance values, namely the inner distance and the outer distance. For all high standardization cross-sections, the centerline coordinates and normal angles are weighted and averaged using the standard scores to obtain the standard centerline and standard normal angle. The inner distance and outer distance at each sampling point are weighted and averaged using the standard scores to obtain the standard thickness of each sampling point. The Cartesian coordinates of the inner and outer contours of each sampling point are calculated using the standard centerline, standard normal angle, and standard thickness.

[0056] At this point, the standard cross-section was obtained.

[0057] The flatness acquisition module calculates the distance between each cross-section and a standard cross-section as the flatness deviation of that cross-section. This distance reflects the degree to which the cross-section shape deviates from the standard cross-section. The distance is calculated using the average Hausdorff distance, which requires first calculating the geometric centroid of the cross-section point cloud and then aligning the centroids of the cross-sections being compared. A larger deviation indicates a greater difference between the cross-section shape and the standard shape, resulting in poorer flatness. A deviation close to zero indicates that the cross-section is almost identical to the standard cross-section, indicating good flatness. The flatness calculated based on the Hausdorff distance in this application essentially covers both the overall profile and local flatness deviation of the cross-section, making it a comprehensive form and position evaluation index.

[0058] Calculate the rate of change of deviation between each cross section and its adjacent preceding cross section; the rate of change is the ratio of the deviation difference to the spatial interval, representing the change in flatness deviation per unit length. A cross section deviation itself may be small, but if it changes rapidly relative to the preceding cross section, it indicates a local abrupt change. Calculate the median and the absolute deviation of the median (MAD) for all rates of change. Defect screening is performed using the rates of change. The method for calculating the absolute deviation of the median is as follows: for a number of values, obtain the median; subtract each value from the median; sum the differences and the ratio of this sum to the number of values; this ratio is taken as the absolute deviation of the median.

[0059] In this embodiment, if the rate of change at a certain cross-sectional location exceeds the median plus 5 times the MAD, it is determined to be a local abrupt change point. The threshold strength of 5 times MAD is comparable to 3σ (MAD is approximately 0.67σ under a normal distribution, and 5 times MAD is approximately 3.35σ), ensuring the consistency of the judgment criteria. σ represents the standard deviation in the 3σ principle of normal distribution.

[0060] This method can identify local defects such as chipped edges, protrusions, and steps. Even if these defects are not obvious in the overall deviation, they can still be detected because of their large rate of change.

[0061] Since the deviation of defective sections can be large, these large deviations will significantly increase the overall average value. Therefore, if the flatness deviation of all sections is directly calculated by arithmetic mean, it will be biased by abnormal sections, and the evaluation results will not truly reflect the normal quality level of the beam.

[0062] The overall flatness is obtained by weighting the flatness deviations of each cross section using the ratio of the standard score of each cross section to the sum of the standard scores of all cross sections. Compared with the traditional arithmetic mean or median method, this weighted method retains the information of all data (instead of simply discarding outliers) and effectively suppresses the influence of outliers, resulting in a more stable and reliable evaluation result.

[0063] This completes the flatness measurement.

[0064] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. An automatic flatness measurement system for large-size prefabricated U-beams based on laser scanning, characterized in that, The system includes the following modules: The point cloud acquisition module is used to acquire three-dimensional point clouds by laser scanning of the U-beam and determine its principal axis direction; The section selection module is used to preset multiple windows along the principal axis, and obtain the plane and normal vector corresponding to the window by fitting the point cloud of adjacent windows; the curvature is calculated by the normal vector of two adjacent windows. The slice interval is limited by the preset normal vector angle tolerance, and then the slice interval is calculated in combination with curvature; the cross section under different coordinates of the principal axis is determined by the slice interval; The standard section determination module is used to determine local continuity based on the distance and principal axis coordinate differences between sections in a preset neighborhood; determine global continuity based on the discrete distribution of distances between global windows; determine topological stability based on the deviation of the point cloud number and perimeter of the section from the total number of points in all sections; and construct standard sections by calculating standardization scores through local continuity, global continuity, and topological stability. The flatness acquisition module obtains the flatness by weighting the distance deviation between each cross section and the standard cross section through a standardity score.

2. The automatic flatness measurement system for large-size prefabricated U-beams based on laser scanning as described in claim 1, characterized in that, The principal axis direction is the direction with the longest point cloud distribution determined by principal component analysis of all point clouds, and the plane perpendicular to the principal axis is used as the cross-sectional plane.

3. The automatic flatness measurement system for large-size prefabricated U-beams based on laser scanning as described in claim 1, characterized in that, The curvature is calculated as follows: The normal vector is multiplied by its adjacent normal vector, and the angle between the two vectors is obtained by taking the inverse cosine of the dot product. The ratio of the angle to the window width is used as the curvature.

4. The automatic flatness measurement system for large-size prefabricated U-beams based on laser scanning as described in claim 1, characterized in that, The expression for the slice interval is: , This represents the tolerance of the angle between the defined normal vectors. Let represent the curvature of the i-th window. Represents a very small positive number. This represents the slice interval of the i-th window. This indicates the preset maximum slice interval. This indicates the preset minimum slice interval.

5. The automatic flatness measurement system for large-size prefabricated U-beams based on laser scanning as described in claim 1, characterized in that, The method for determining the cross section at different coordinates of the principal axis by slicing interval is as follows: Along the main axis, the right side of the first window is taken as the first slice section. After calculating the curvature of the two windows on the left and right sides of the slice section, the slice interval is obtained. For the first slice section, the position of the slice interval is taken as the second slice section, and so on to obtain all slice sections.

6. The automatic flatness measurement system for large-size prefabricated U-beams based on laser scanning as described in claim 1, characterized in that, The local continuity is negatively correlated with the shape change rate; the shape change rate is positively correlated with the distance between the cross section and all cross sections in its neighborhood set, and negatively correlated with the difference in principal axis coordinates between the cross section and all cross sections in its neighborhood set.

7. The automatic flatness measurement system for large-size prefabricated U-beams based on laser scanning as described in claim 1, characterized in that, The global continuity is negatively correlated with the coefficient of variation of the distance between all windows.

8. The automatic flatness measurement system for large-size prefabricated U-beams based on laser scanning as described in claim 1, characterized in that, The topological stability is negatively correlated with the deviations of the cross section and all cross sections in terms of point cloud quantity and perimeter.

9. The automatic flatness measurement system for large-size prefabricated U-beams based on laser scanning as described in claim 1, characterized in that, The standardization score is the average of the first standardization score and the second standardization score; the first standardization score is the sum of local continuity and global continuity, and the second standardization score is topological stability.

10. The automatic flatness measurement system for large-size prefabricated U-beams based on laser scanning as described in claim 1, characterized in that, The method for obtaining the standard cross-section is as follows: Sort all cross-sections from largest to smallest according to their standard scores. Form a high-standard set with the first preset number of cross-sections. For each cross-section in the high-standard set, select its central axis and sample it at equal intervals along the central axis according to the arc length. At each sampling point, emit a ray along the normal direction of the central axis, intersecting the inner and outer contours of the cross-section to obtain two distance values, namely the inner distance and the outer distance. For all high-standard cross-sections, the coordinates of the central axis and the normal angle are weighted and averaged according to the standard scores to obtain the standard central axis and the standard normal angle. The inner distance and the outer distance at each sampling point are weighted and averaged according to the standard scores to obtain the standard thickness of each sampling point. Calculate the Cartesian coordinates of the inner and outer contours of each sampling point using the standard central axis, the standard normal angle, and the standard thickness.

Images