A method for detecting surface defects in aluminum alloy templates

Abstract

This application relates to the field of aluminum alloy formwork quality inspection technology, and particularly to a method for detecting surface defects in aluminum alloy formwork. It utilizes a 3D laser scanner to acquire an overall point cloud, followed by scanning to obtain a local high-precision point cloud. The local high-precision point cloud is then fused with the overall point cloud. The measured 3D model is registered with a standard model. A multi-feature region growing algorithm is used to segment the formwork into different structural components. For each component, an improved random sampling consistency algorithm is used to fit a dynamic reference plane. The normal deviation of each component's point cloud from the reference plane is calculated, and a deviation chromatogram is generated for visualization analysis. This application elevates traditional qualitative inspection relying on manual experience to data-driven quantitative assessment. The inspection process is standardized, and parameter determination is objective, significantly improving the scientific rigor, reliability, and efficiency of aluminum formwork quality control.

Description

Technical Field

[0001] This application relates to the field of aluminum alloy template quality inspection technology, and in particular to a method for detecting surface defects in aluminum alloy templates. Background Technology

[0002] As a key reusable material in modern industrialized construction systems, the surface quality of aluminum alloy formwork is crucial to concrete forming results, construction efficiency, and overall cost control. During production, transportation, and repeated use, aluminum alloy formwork is highly susceptible to various defects such as scratches, dents, corrosion, localized deformation, and coating peeling. Simultaneously, its geometric dimensions (such as opening location, diameter, and overall flatness) are prone to deviations. If these defects and dimensional deviations are not detected promptly, they will directly replicate on the concrete surface, affecting the structural appearance and quality. Furthermore, they will exacerbate the adhesion between the formwork and concrete, leading to difficulties in demolding, accelerated formwork wear, and even concrete surface damage. Therefore, efficient and accurate identification and control of surface defects and geometric dimensions of aluminum alloy formwork are core elements in ensuring concrete forming quality, increasing formwork turnover, and promoting the development of smart construction sites.

[0003] Currently, similar quality inspection methods for aluminum formwork mostly rely on manual visual inspection or single inspection techniques. Such methods have limitations such as low efficiency, strong subjectivity, and limited inspection dimensions: manual inspection is difficult to achieve accurate quantification of geometric dimensions; traditional contact measurement methods are inefficient and cannot provide comprehensive coverage; existing 3D scanning technologies mainly focus on macroscopic dimension comparison and are insufficient in identifying local micro-deformations and early buckling risks. Summary of the Invention

[0004] This application aims to at least partially address one of the aforementioned technical problems in the prior art. To this end, embodiments of this application provide a method for detecting surface defects in aluminum alloy templates. This application elevates traditional qualitative testing, which relies on manual experience, to a data-driven quantitative assessment. The testing process is standardized, and parameter determination is objective, significantly improving the scientific rigor, reliability, and efficiency of aluminum template quality control.

[0005] A method for detecting surface defects in aluminum alloy templates includes the following steps: Step 1: Use a 3D laser scanner to scan the aluminum alloy template in medium resolution mode to obtain the overall point cloud. Then, switch to high resolution mode to scan the key areas to obtain local high-precision point clouds. The local high-precision point clouds and the overall point cloud are fused together using feature matching and iterative nearest point algorithm to establish a complete measured 3D model P. Step 2: Register the measured 3D model P with the standard CAD model Q. First, perform coarse registration using feature points, and then perform fine registration using the iterative nearest point algorithm. Calculate the root mean square error value after registration and compare it with the threshold specified in the standard. Determine whether the overall dimensions are qualified based on the comparison results. If they are not qualified, terminate the subsequent inspection process. Step 3: Use the region growing algorithm to segment the point cloud clusters of each circular hole from the measured 3D model P, perform cylindrical surface fitting on each point cloud cluster of circular holes, calculate its center coordinates and diameter, calculate the hole diameter rationality index based on the measured data of all circular holes, evaluate the processing accuracy of individual holes and the consistency between holes, compare the hole diameter rationality with the qualified threshold, and determine whether the hole quality meets the standard. Step 4: The template is segmented into different structural components using a multi-feature region growing algorithm. An improved random sampling consensus algorithm is used to fit a dynamic reference plane for each component. The normal deviation of the point cloud of each component to the reference plane is calculated, and a deviation chromatogram is generated for visualization analysis. Based on the standard, it is determined whether the local deformation exceeds the allowable range.

[0006] In an optional or preferred embodiment, the step of fine registration using the iterative nearest point algorithm in step 2 is as follows: Based on the coarse registration, the overall distance error between the measured 3D model P and the standard CAD model Q is optimized using a nonlinear least squares problem. The formula for calculating the overall distance error E(R,T) is: ; Where: N p R is the number of points in the measured 3D model P; R and T are the optimal rotation matrix and translation vector to be solved; and T is the number of points in the standard CAD model Q after the transformation (R·p). i +T) is the point with the closest Euclidean distance.

[0007] In an optional or preferred embodiment, the formula for calculating the root mean square error value after registration in step 2 is: ; in Points in the registered point cloud; when RMSE>2.25mm, the overall size is deemed unqualified.

[0008] In an optional or preferred embodiment, the formula for calculating the aperture reasonableness σ in step 3 is: ; in: As a weighting factor, and k represents the total number of identified holes; D j The diameter calculated by fitting the cylindrical surface of the j-th hole; The standard diameter of the circular hole; These are the maximum and minimum values ​​among all measured apertures; This refers to the predefined allowable tolerance for aperture diameter.

[0009] In an optional or preferred embodiment, the acceptable threshold for the aperture reasonableness σ is... Determined by the following formula: ; When the aperture rationality σ of the aluminum alloy template is lower than the threshold If the quality of the template hole diameter is found to be substandard, it is determined that the template hole diameter is not up to standard.

[0010] In an optional or preferred embodiment, step 4 includes: for a seed point and its neighboring vertices, the criterion for determining whether to merge them into the same region is a multi-feature constraint function F(v s , v c ): ; in: ; Where v s v is the seed point for the current region. c w is the candidate point to be examined. n ,w c ,w g The weighting factor and w n +w c +w g =1;Δ normal For the difference term of the normal vector, n(v) s ) and n(v c ) are the unit normal vectors of the seed point and the candidate point, respectively; Δ curvature H(v) is the curvature difference term, where H(v) is the average curvature at vertex v; Δ geodesic D represents the geodetic distance difference term. g (v s , v c ) is vertex v s With v c The geodesic distance between them is the shortest path length along the surface of the 3D model. L avg It is the square root of the total surface area of ​​the entire aluminum formwork model, for each (v s , v c Yes, calculate F(v) s , v c When F(v) s , v c )<θ grow When candidate point v c Merge it into the current region and use it as the new seed point; Starting from the seed point, adjacent vertices that satisfy the growth criteria are continuously added to the current region. When the current region can no longer grow, a new seed point is selected to start a new region growth, until all vertices have been processed, resulting in the point cloud subsets P of the regions to be analyzed. local ={p i}; Calculate P local The local curvature characteristics of each point in the matrix form a candidate interior point set P. candidate A plane can be defined by its unit normal vector n=(a,b,c) and its distance d from the origin. The equation of the plane is: ; In P candidate Calculate the plane model π by randomly sampling three points, and calculate P. local The distance δ from all points to the plane i : ; Will d i <ϵ plane The point is considered the consensus point of the model, where ϵ plane It is a preset distance threshold; Repeat the above process, and finally select the planar model with the most consensus points as the fitted dynamic reference plane. π base .

[0011] In an optional or preferred embodiment, step 4 involves calculating the point cloud of each component to the dynamic reference plane. π bas The normal deviation includes: for local point clouds P local Each point in p i Calculate its distance to the dynamic reference plane π base Signed normal distance ,like If the value is greater than 0.15, then a certain area of ​​the aluminum formwork is considered unqualified.

[0012] In optional or preferred embodiments, step 5 is further included, which includes the following steps: calculating the curvature characteristics and gradient distribution of each component region; constructing an equivalent curvature model to comprehensively evaluate the curvature amplitude and its rate of change; setting dynamic dual thresholds based on material properties and geometric features to identify early risk areas; performing spatial clustering and risk assessment on the identified risk points to generate a complete inspection report.

[0013] In an optional or preferred embodiment, the equivalent curvature in step 5 The calculation formula is: ; Where α is the gradient energy factor, calculated according to Saint-Venant's principle and plate and shell theory as follows: ; ν is the Poisson's ratio for aluminum; The characteristic length is denoted as .

[0014] In an optional or preferred embodiment, in step 5: for all deformation points and points of interest, a clustering algorithm based on Euclidean distance is used to aggregate spatially adjacent points into different potential deformation regions. R k Calculate its area A k Calculate the regional risk index for each region by taking the average absolute value of curvature of the vertices within the region.

[0015] ; Where A k For the area, The absolute value of the average equivalent curvature of the vertices within the region; risk index. A higher value indicates a greater risk of early buckling deformation in the region.

[0016] Based on the above technical solutions, the embodiments of this application have at least the following beneficial effects: This application combines medium-resolution rapid global scanning with high-resolution fine scanning of key areas, which greatly improves efficiency while ensuring detection accuracy; it adopts a multi-stage registration method that combines feature matching and iterative nearest point algorithm to achieve high-precision alignment between the measured model and the standard model, laying the foundation for subsequent quantitative analysis; it automatically segments structural components through a region growing algorithm with multi-feature constraints, and uses an improved random sampling consistency algorithm to fit a dynamic reference plane for each component, overcoming the limitation of traditional methods in accurately quantifying local deformation, and realizing personalized and high-precision local deformation assessment. This application upgrades the traditional qualitative detection that relies on manual experience to a data-driven quantitative assessment, standardizes the detection process, and objectifies parameter determination, significantly improving the scientificity, reliability, and efficiency of aluminum template quality control. Attached Figure Description

[0017] The present application will be further described below with reference to the accompanying drawings and embodiments; Figure 1 This is a flowchart of the implementation process of this application. Detailed Implementation

[0018] To make the above-mentioned objectives, features, and advantages of this application more apparent and understandable, the specific embodiments of this application are described in detail below with reference to the accompanying drawings. Many specific details are set forth in the following description to provide a thorough understanding of this application. However, this application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar modifications without departing from the spirit of this application. Therefore, this application is not limited to the specific embodiments disclosed below.

[0019] In the description of this application, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc., indicating the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application.

[0020] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0021] In this application, unless otherwise expressly specified and limited, the terms "installation," "connection," "joining," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components, unless otherwise expressly limited. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0022] In this application, unless otherwise expressly specified and limited, "above" or "below" the second feature can mean that the first feature is in direct contact with the second feature, or that the first feature is in indirect contact with the second feature through an intermediate medium. Furthermore, "above," "on top of," and "over" the second feature can mean that the first feature is directly above or diagonally above the second feature, or simply that the first feature is at a higher horizontal level than the second feature. "Below," "below," and "under" the second feature can mean that the first feature is directly below or diagonally below the second feature, or simply that the first feature is at a lower horizontal level than the second feature.

[0023] It should be noted that when an element is referred to as being "fixed to" or "set on" another element, it can be directly on the other element or there may be an intervening element. When an element is considered to be "connected to" another element, it can be directly connected to the other element or there may be an intervening element. The terms "vertical," "horizontal," "upper," "lower," "left," "right," and similar expressions used herein are for illustrative purposes only and do not represent the only possible implementation.

[0024] As a key reusable material in modern industrialized construction systems, the surface quality of aluminum alloy formwork is crucial to concrete forming results, construction efficiency, and overall cost control. During production, transportation, and repeated use, aluminum alloy formwork is highly susceptible to various defects such as scratches, dents, corrosion, localized deformation, and coating peeling. Simultaneously, its geometric dimensions (such as opening location, diameter, and overall flatness) are prone to deviations. If these defects and dimensional deviations are not detected promptly, they will directly replicate on the concrete surface, affecting the structural appearance and quality. Furthermore, they will exacerbate the adhesion between the formwork and concrete, leading to difficulties in demolding, accelerated formwork wear, and even concrete surface damage. Therefore, efficient and accurate identification and control of surface defects and geometric dimensions of aluminum alloy formwork are core elements in ensuring concrete forming quality, increasing formwork turnover, and promoting the development of smart construction sites.

[0025] Currently, similar quality inspection methods for aluminum formwork mostly rely on manual visual inspection or single inspection techniques. Such methods have limitations such as low efficiency, strong subjectivity, and limited inspection dimensions: manual inspection is difficult to achieve accurate quantification of geometric dimensions; traditional contact measurement methods are inefficient and cannot provide comprehensive coverage; existing 3D scanning technologies mainly focus on macroscopic dimension comparison and are insufficient in identifying local micro-deformations and early buckling risks.

[0026] Reference Figure 1 This application provides a method for detecting surface defects in aluminum alloy templates, comprising the following steps: Step 1: Using a 3D laser scanner, the aluminum alloy template is scanned as a whole in medium resolution mode to obtain an overall point cloud. Then, key areas are scanned in high resolution mode to obtain local high-precision point clouds. The local high-precision point clouds are then fused with the overall point cloud using feature matching and an iterative nearest-point algorithm to establish a complete measured 3D model P. By combining rapid medium-resolution global scanning with high-resolution fine scanning of key areas, efficiency is significantly improved while ensuring detection accuracy.

[0027] Step 2: Register the measured 3D model P with the standard CAD model Q. First, perform coarse registration using feature points, then perform fine registration using the iterative nearest point algorithm. Calculate the root mean square error value after registration and compare it with the threshold specified in the standard. Determine whether the overall dimensions are qualified based on the comparison results. If not, terminate the subsequent inspection process. By adopting a multi-stage registration method combining feature matching and the iterative nearest point algorithm, high-precision alignment between the measured model and the standard model is achieved, laying the foundation for subsequent quantitative analysis.

[0028] Step 3: Use the region growing algorithm to segment the point cloud clusters of each circular hole from the measured 3D model P. Perform cylindrical surface fitting on each point cloud cluster of circular holes, calculate its center coordinates and diameter, calculate the hole diameter rationality index based on the measured data of all circular holes, evaluate the processing accuracy of individual holes and the consistency between holes, compare the hole diameter rationality with the qualified threshold, and determine whether the hole quality meets the standard.

[0029] Step 4: The template is segmented into different structural components using a multi-feature region growing algorithm. For each component, an improved random sampling consensus algorithm is used to fit a dynamic reference plane. The normal deviation of the point cloud of each component to the reference plane is calculated, and a deviation chromatogram is generated for visualization analysis. Based on the standard, it is determined whether the local deformation exceeds the allowable range. By automatically segmenting the structural components using a multi-feature constrained region growing algorithm and fitting a dynamic reference plane for each component using an improved random sampling consensus algorithm, the limitations of traditional methods in accurately quantifying local deformation are overcome, achieving personalized and high-precision local deformation assessment.

[0030] Step 5: Calculate the curvature characteristics and gradient distribution of each component region; construct an equivalent curvature model to comprehensively evaluate the curvature amplitude and its rate of change; set dynamic dual thresholds based on material properties and geometric characteristics to identify early risk areas; perform spatial clustering and risk assessment on the identified risk points to generate a complete inspection report. By equating the curvature gradient contribution to curvature increments, an equivalent curvature formula with clear physical meaning is constructed, and dynamic dual thresholds are automatically calculated based on material properties, enabling effective identification of early buckling risks that have not yet formed significant displacement but have already shown abrupt curvature changes.

[0031] This invention upgrades the traditional qualitative testing that relies on human experience to a data-driven quantitative assessment. The testing process is standardized and the parameters are objectively determined, which significantly improves the scientific nature, reliability and efficiency of aluminum formwork quality control and provides strong technical support for building industrialization.

[0032] In some embodiments, the step of fine registration using the iterative nearest point algorithm in step 2 is as follows: Based on the coarse registration, the overall distance error between the measured 3D model P and the standard CAD model Q is optimized using a nonlinear least squares problem. The formula for calculating the overall distance error E(R,T) is: ; Where: N p R is the number of points in the measured 3D model P; R and T are the optimal rotation matrix and translation vector to be solved. It is the point (R·p) in the standard CAD model Q and the transformed point. i +T) is the point with the closest Euclidean distance.

[0033] In some embodiments, the formula for calculating the root mean square error value after registration in step 2 is: ; in Points in the registered point cloud; when RMSE>2.25mm, the overall size is deemed unqualified.

[0034] The formula for calculating the aperture rationality σ in step 3 is: ; in: As a weighting factor, and k represents the total number of identified holes; D j The diameter calculated by fitting the cylindrical surface of the j-th hole; The standard diameter of the circular hole; These are the maximum and minimum values ​​among all measured apertures; This refers to the predefined allowable tolerance for aperture diameter.

[0035] In some embodiments, the acceptable threshold for aperture suitability σ is determined by the following formula: ; When the aperture rationality σ of the aluminum alloy template is lower than the threshold If the quality of the template hole diameter is found to be substandard, it is determined that the template hole diameter is not up to standard.

[0036] In some embodiments, step 4 includes: for a seed point and its neighboring vertices, the criterion for determining whether to merge them into the same region is a multi-feature constraint function F(v s , v c ):

[0037] in: ; Where v s v is the seed point for the current region. c w is the candidate point to be examined. n ,w c ,w g The weighting factor and w n +w c +w g=1;Δ normal For the difference term of the normal vector, n(v) s ) and n(v c ) are the unit normal vectors of the seed point and the candidate point, respectively; Δ curvature H(v) is the curvature difference term, where H(v) is the average curvature at vertex v; Δ geodesic D represents the geodetic distance difference term. g (v s , v c ) is vertex v s With v c The geodesic distance between them is the shortest path length along the surface of the 3D model. L avg It is the square root of the total surface area of ​​the entire aluminum formwork model, for each (v s , v c Yes, calculate F(v) s , v c When F(v) s , v c )<θ grow When candidate point v c Merge it into the current region and use it as the new seed point; Starting from the seed point, adjacent vertices that satisfy the growth criteria are continuously added to the current region. When the current region can no longer grow, a new seed point is selected to start a new region growth, until all vertices have been processed, resulting in the point cloud subsets P of the regions to be analyzed. local ={p i}; Calculate P local The local curvature characteristics of each point in the matrix form a candidate interior point set P. candidate A plane can be defined by its unit normal vector n=(a,b,c) and its distance d from the origin. The equation of the plane is: ; In P candidate Calculate the plane model π by randomly sampling three points, and calculate P. local The distance δ from all points to the plane i : Will d i <ϵ plane The point is considered the consensus point of the model, where ϵ plane It is a preset distance threshold; Repeat the above process, and finally select the planar model with the most consensus points as the fitted dynamic reference plane. π base .

[0038] In some embodiments, step 4 calculates the point cloud of each component to the dynamic reference plane. π basThe normal deviation includes: for local point clouds P local Each point in p i Calculate its distance to the dynamic reference plane π base Signed normal distance ,like If the value is greater than 0.15, then a certain area of ​​the aluminum formwork is considered unqualified.

[0039] In some embodiments, the equivalent curvature in step 5 The calculation formula is: ;

[0040] Where α is the gradient energy factor, calculated according to Saint-Venant's principle and plate and shell theory as follows: ; ν is the Poisson's ratio for aluminum; The characteristic length is denoted as .

[0041] In some embodiments, in step 5: for all deformation points and points of interest, a clustering algorithm based on Euclidean distance is used to aggregate spatially adjacent points into different potential deformation regions. R k Calculate its area A k Calculate the regional risk index for each region by taking the average absolute value of curvature of the vertices within the region.

[0042] ; Where A k For the area, The absolute value of the average equivalent curvature of the vertices within the region; risk index. A higher value indicates a greater risk of early buckling deformation in the region.

[0043] The specific steps for this application are detailed below: Step 1: 3D point cloud data acquisition and model reconstruction based on a global-local hierarchical strategy.

[0044] Step 1.1: Using a handheld 3D laser scanner, perform a rapid, all-around scan of the sample aluminum template in medium resolution mode (point cloud spacing set to 1-2mm).

[0045] Step 1.2: Based on the initial model generated in Step 1.1, the system automatically identifies or the operator manually selects vulnerable or critical quality areas such as the area around bolt holes, weld joints, and corners. Then, the scanner is switched to high-resolution mode (point cloud spacing set to ≤0.1mm) to perform close-range, multi-angle supplementary scanning of these specific areas.

[0046] Step 1.3: Combine the local high-precision point cloud obtained in Step 1.2 with the overall point cloud obtained in Step 1.1, and perform multi-view point cloud data fusion in a unified global coordinate system. A hybrid registration method based on features and iterative nearest point (ICP) algorithms is adopted to ensure seamless stitching of point cloud data at different resolutions.

[0047] Step 1.4: Perform outlier filtering and noise smoothing preprocessing on the fused complete point cloud dataset, and then use the greedy projection triangulation algorithm to reconstruct the surface, finally generating a high-fidelity 3D mesh model for subsequent quantitative analysis.

[0048] Step 2: Quantitative analysis of overall size deviation based on multi-stage point cloud registration Step 2.1: Measure the point cloud of the aluminum template (set as follows) Standard CAD model (converted to point cloud format) In this study, a rigid transformation matrix was initially calculated using the nearest neighbor search and random sampling consensus (RANSAC) algorithm. (in R 0 is the initial rotation matrix. T (0 is the initial translation vector), to complete the initial coarse registration of the two point clouds, and to initially align the measured point cloud to the standard coordinate system.

[0049] Step 2.2: Based on the coarse registration, the measured point cloud is... P Compared with the standard model Q Overall distance error between E ( R , T This is defined as a nonlinear least squares optimization problem, which aims to minimize the following: ; in: N p It is a measured point cloud P The number of points in the middle. R and T It is the optimal rotation matrix and translation vector to be solved. q nearest(i) It is a standard point cloud Q The transformed point ( R · p i + TThe point with the closest Euclidean distance.

[0050] Obtain the optimal transformation matrix with high precision. Applying this transformation, the final registered measured point cloud is obtained. .

[0051] Step 2.3: Calculate the overall root mean square error of registration (RMSE), which represents the distance error from each point in the measured point cloud to the nearest point on the standard model. This serves as an evaluation index for registration quality and overall shape consistency. ; Referring to industry-recognized standards such as "General Industrial Aluminum and Aluminum Alloy Plates and Strips" (GB / T 3880.3), "Dimensional Deviations of Aluminum and Aluminum Alloy Extruded Profiles" (GBT14846-2014), and "General Industrial Aluminum and Aluminum Alloy Extruded Profiles" (GB / T6892-2015), the threshold is determined to be 1.5mm. Taking a safety factor k of 1.5, when RMSE>2.25, the overall dimensions are considered unqualified.

[0052] Step 3: Identification and evaluation of opening location and diameter Step 3.1: Based on the registered 3D point cloud model, the point cloud clusters of each circular hole on the surface of the aluminum template are automatically identified and segmented using a region growing algorithm.

[0053] Step 3.2: For each identified cluster of circular hole point clouds, calculate the spatial coordinates of the central axis of the fitted cylindrical surface, which will be used as the position of the center point of the circular hole. x j , y j , z j The positional deviation Δ is calculated by comparing it with the corresponding theoretical coordinates in the standard model. L .

[0054] Step 3.3: Extract the diameter of the fitted cylindrical surface. D j The standard diameter of the circular hole stored in the cloud database D Compare the results and calculate the aperture deviation Δ. D A comprehensive analysis of the reasonableness of the aperture size of the sample aluminum template. s The calculation formula is as follows: ; in: c 1 and c 2 is the weighting factor, and c 1+ c 2 = 1; k This represents the total number of circular holes identified on the sample aluminum template. Dmax and D min These are the maximum and minimum values ​​among all measured apertures; D D′′ represents the predefined allowable deviation of the hole diameter. The first measure is the degree to which each hole diameter conforms to the standard diameter. The second measure is the consistency of diameters among all holes. The weighting factor can be set in advance according to site requirements. To ensure the assembly function of the aluminum template and the standard connecting bolts, the D′′ value should not exceed half of the difference between the maximum bolt diameter and the minimum diameter of the template hole, and is usually set in the range of 0.3mm to 0.8mm.

[0055] The reasonableness of the aperture s qualified threshold s critical It is determined by the following formula: ; When the aperture of the sample aluminum template is reasonable s Below the threshold s critical If the quality of the template hole diameter is found to be substandard, it is determined that the template hole diameter is not up to standard.

[0056] Step 4: Quantitative detection of local deformation based on dynamic reference plane fitting and normal deviation analysis. Step 4.1: Create a triangular mesh M Represented as a diagram G ( V , E ), where the vertices V These are grid vertices and edges. E These are mesh edges. Traverse all vertices and assign an initial "part candidate label" to each according to the rules described above. L initial ( v i ).

[0057] Step 4.2: For a seed point and its neighboring vertices, the criterion for determining whether to merge them into the same region is a multi-feature constraint function. F ( v s , v c ): ; in: v s This is the seed point for the current region. v c These are the candidate points to be examined. w n , w c ,w g It is a preset weighting factor, and w n + w c + w g =1. These determine the relative importance of the three features: normal vector, curvature, and distance.

[0058] Δ normal For the difference term of the normal vector, n ( v s )and n ( v c ) are the unit normal vectors of the seed point and the candidate point, respectively, and • represents the dot product operation of the vectors, the result of which is equal to the cosine of the angle between the two vectors.

[0059] Δ curvature For the curvature difference term, H ( v ) is a vertex v The average curvature at ϵ. ϵ is a very small positive number to prevent the denominator from being zero.

[0060] Δ geodesic This is the geodetic distance difference term. D g ( v s , v c ) is a vertex v s and v c The geodesic distance between them is the shortest path length along the surface of the three-dimensional model. L avg It is the square root of the total surface area of ​​the entire aluminum template model, used to normalize the distance so that it is independent of the model size.

[0061] For each ( v s , v c Yes, calculate F ( v s , v c ).if F ( v s , v c )< i grow Then the candidate points v cMerge it into the current region and use it as the new seed point.

[0062] Step 4.3: Starting from the seed point, continuously add adjacent vertices that satisfy the growth criteria to the current region. When the current region can no longer grow, select a new seed point to start a new region growth, until all vertices have been processed. Based on the design characteristics of the aluminum template, automatically identify and segment easily deformable regions such as panels, edge ribs, and middle ribs to obtain the point cloud subsets of each region to be analyzed. P local ={ p i} Step 4.4: Calculation P local Based on the local curvature characteristics of each point, points with significantly excessive curvature (suspected deformation areas) are temporarily excluded, forming a candidate set of interior points. P candidate A plane can be represented by its unit normal vector. n =( a , b , c and distance to the origin d The equation of the plane is defined as follows: ; exist P candidate Calculate a planar model by randomly sampling three points. π .calculate P local The distance from all points to the plane d i : ; Will d i <ϵ plane The point is considered the consensus point of the model, where ϵ plane It is a preset distance threshold.

[0063] Step 4.5: Repeat the above process, and finally select the planar model with the most consensus points as the fitted "dynamic reference plane". π base .

[0064] Step 4.6: Precise calculation of normal deviation for local point clouds. P local Each point in p i Calculate its distance to the dynamic reference plane π baseThe signed normal distance is determined to be 0.15 mm according to industry-recognized standards such as "General Industrial Aluminum and Aluminum Alloy Plates and Strips" (GB / T3880.3), "Dimensional Deviations of Aluminum and Aluminum Alloy Extruded Profiles" (GBT14846-2014), and "General Industrial Aluminum and Aluminum Alloy Extruded Profiles" (GB / T6892-2015). If it is greater than 0.15 mm, then a certain area of ​​the aluminum template is considered unqualified.

[0065] Step 4.7: Perform the above operation on each area of ​​the template to identify all unqualified areas.

[0066] Step 5: Early Deformation Warning Based on Multi-Curvature Features and Dual-Threshold Judgment Step 5.1: Create a 3D triangular mesh model for the local area. M For each grid vertex v Based on its neighborhood grid, its average curvature is calculated. H and Gaussian curvature K .

[0067] Step 5.2: Calculate the local gradient magnitude of the curvature at each vertex ||▽ H ||and||▽ K ||, to capture abrupt changes in curvature.

[0068] Step 5.3: Set dual threshold judgment rules Primary threshold i 1: Set a low sensitivity threshold. When the vertex's ||▽ H ||and||▽ K || Exceeding i When the value is 1, mark the vertex as a "point of concern" to indicate that there is a potential risk.

[0069] Advanced threshold i 2: Set a higher confirmation threshold ( i 2> i 1) When the aforementioned curvature eigenvalue of the vertex exceeds i At time 2, the vertex is directly identified as the "deformation point".

[0070] First, define a comprehensive equivalent curvature. H eq : ; in, α Let be the gradient energy factor, a fixed dimensionless constant. Based on Saint-Venant's principle and plate and shell theory, its value can be theoretically derived as follows: ; in vThis is the Poisson's ratio of aluminum (approximately 0.33). Therefore, α is calculated to be approximately 0.1. L char The characteristic length is denoted as .

[0071] ; Step 5.4: For all "deformation points" and "points of interest", use a clustering algorithm based on Euclidean distance to aggregate spatially adjacent points into different "potential deformation regions". R k Calculate its area. A k The average absolute value of curvature at the vertices within the region. Combining area and curvature severity, calculate the risk index for each region: ; Risk Index r k The higher the value, the greater the risk of early buckling deformation in that region.

[0072] The embodiments of this application have been described in detail above with reference to the accompanying drawings. However, this application is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of this application.

Claims

1. A method for detecting surface defects in aluminum alloy templates, characterized in that, Includes the following steps: Step 1: Use a 3D laser scanner to scan the aluminum alloy template in medium resolution mode to obtain the overall point cloud. Then, switch to high resolution mode to scan the key areas to obtain local high-precision point clouds. The local high-precision point clouds and the overall point cloud are fused together using feature matching and iterative nearest point algorithm to establish a complete measured 3D model P. Step 2: Register the measured 3D model P with the standard CAD model Q. First, perform coarse registration using feature points, and then perform fine registration using the iterative nearest point algorithm. Calculate the root mean square error value after registration and compare it with the threshold specified in the standard. Determine whether the overall dimensions are qualified based on the comparison results. If they are not qualified, terminate the subsequent inspection process. Step 3: Use the region growing algorithm to segment the point cloud clusters of each circular hole from the measured 3D model P, perform cylindrical surface fitting on each point cloud cluster of circular holes, calculate its center coordinates and diameter, calculate the hole diameter rationality index based on the measured data of all circular holes, evaluate the processing accuracy of individual holes and the consistency between holes, compare the hole diameter rationality with the qualified threshold, and determine whether the hole quality meets the standard. Step 4: The template is segmented into different structural components using a multi-feature region growing algorithm. An improved random sampling consensus algorithm is used to fit a dynamic reference plane for each component. The normal deviation of the point cloud of each component to the reference plane is calculated, and a deviation chromatogram is generated for visualization analysis. Based on the standard, it is determined whether the local deformation exceeds the allowable range.

2. The method for detecting surface defects in aluminum alloy templates according to claim 1, characterized in that: The step 2 of fine registration using the iterative nearest point algorithm is as follows: Based on the coarse registration, the overall distance error between the measured 3D model P and the standard CAD model Q is optimized using a nonlinear least squares problem. The formula for calculating the overall distance error E(R,T) is: ; Where: N p R is the number of points in the measured 3D model P; R and T are the optimal rotation matrix and translation vector to be solved. It is the point (R·p) in the standard CAD model Q and the transformed point. i +T) is the point with the closest Euclidean distance.

3. The method for detecting surface defects in aluminum alloy templates according to claim 2, characterized in that: The formula for calculating the root mean square error value after registration in step 2 is as follows: ; in Points in the registered point cloud; when RMSE>2.25mm, the overall size is deemed unqualified.

4. The method for detecting surface defects of aluminum alloy templates according to claim 1, characterized in that: The formula for calculating the aperture rationality σ in step 3 is: ; in: As a weighting factor, and k represents the total number of identified holes; D j The diameter calculated by fitting the cylindrical surface of the j-th hole; The standard diameter of the circular hole; and These are the maximum and minimum values ​​among all measured apertures; This refers to the predefined allowable tolerance for aperture diameter.

5. The method for detecting surface defects of aluminum alloy templates according to claim 4, characterized in that: The acceptable threshold for the aperture reasonableness σ Determined by the following formula: ; When the aperture rationality σ of the aluminum alloy template is lower than the threshold If the quality of the template hole diameter is found to be substandard, it is determined that the template hole diameter is not up to standard.

6. The method for detecting surface defects in aluminum alloy templates according to claim 1, characterized in that: Step 4 includes: for a seed point and its neighboring vertices, the criterion for determining whether to merge them into the same region is a multi-feature constraint function F(v). s , v c ): ; in: ; Where v s v is the seed point for the current region. c w is the candidate point to be examined. n ,w c ,w g The weighting factor and w n +w c +w g =1;Δ normal For the difference term of the normal vector, n(v) s ) and n(v c ) are the unit normal vectors of the seed point and the candidate point, respectively; Δ curvature H(v) is the curvature difference term, where H(v) is the average curvature at vertex v; Δ geodesic D represents the geodetic distance difference term. g (v s , v c ) is vertex v s With v c The geodesic distance between them is the shortest path length along the surface of the 3D model. L avg It is the square root of the total surface area of ​​the entire aluminum formwork model, for each (v s , v c Yes, calculate F(v) s , v c When F(v) s , v c )<θ grow When candidate point v c Merge it into the current region and use it as the new seed point; Starting from the seed point, adjacent vertices that satisfy the growth criteria are continuously added to the current region. When the current region can no longer grow, a new seed point is selected to start a new region growth, until all vertices have been processed, resulting in the point cloud subsets P of the regions to be analyzed. local ={p i }; Calculate P local The local curvature characteristics of each point in the matrix form a candidate interior point set P. candidate A plane can be defined by its unit normal vector n=(a,b,c) and its distance d from the origin. The equation of the plane is: ; In P candidate Calculate the plane model π by randomly sampling three points, and calculate P. local The distance δ from all points to the plane i : ; Will δ i <ϵ plane The point is considered the consensus point of the model, where ϵ plane It is a preset distance threshold; Repeat the above process, and finally select the planar model with the most consensus points as the fitted dynamic reference plane. π base .

7. The method for detecting surface defects in aluminum alloy templates according to claim 6, characterized in that: In step 4, the point cloud of each component is calculated to the dynamic reference plane. π bas The normal deviation includes: for local point clouds P local Each point in p i Calculate its distance to the dynamic reference plane π base Signed normal distance ,like If the value is greater than 0.15, then a certain area of ​​the aluminum formwork is considered unqualified.

8. The method for detecting surface defects of aluminum alloy templates according to claim 1, characterized in that: The method also includes step 5, which includes the following steps: calculating the curvature characteristics and gradient distribution of each component region; constructing an equivalent curvature model to comprehensively evaluate the curvature amplitude and its rate of change; setting dynamic dual thresholds based on material properties and geometric characteristics to identify early risk areas; performing spatial clustering and risk assessment on the identified risk points to generate a complete inspection report.

9. The method for detecting surface defects in aluminum alloy templates according to claim 8, characterized in that: The equivalent curvature in step 5 The calculation formula is: ; Where α is the gradient energy factor, calculated according to Saint-Venant's principle and plate and shell theory as follows: ; ν is the Poisson's ratio for aluminum; The characteristic length is denoted as .

10. The method for detecting surface defects in aluminum alloy templates according to claim 1, characterized in that: In step 5: For all deformation points and points of interest, a clustering algorithm based on Euclidean distance is used to aggregate spatially adjacent points into different potential deformation regions. R k Calculate its area A k Calculate the regional risk index for each region by taking the average absolute value of curvature of the vertices within the region. : ； Where A k For the area, The absolute value of the average equivalent curvature of the vertices within the region; Risk Index A higher value indicates a greater risk of early buckling deformation in the region.

Images