AR matching repair method and system for damaged component of ancient building

Abstract

The invention provides an ancient building damaged component AR matching repair method and system, and the method comprises the steps: obtaining the surface point cloud data of a damaged component through an augmented reality technology, generating a high-precision three-dimensional point cloud model through a laser scanning and multi-view stereoscopic vision fusion method, and obtaining the digital expression of the complex geometric shape of the component; for the point cloud model, adopting voxel filtering and a normal estimation algorithm to carry out denoising and surface smoothing processing, and extracting component edge features and curved surface geometric features to obtain a smooth three-dimensional geometric model; and carrying out space mapping on the optimized mortise and tenon joint structure model and an actual component position through an augmented reality technology, generating a virtual joint guide line and stress distribution visual image, guiding a component installation process, and obtaining a high-precision component installation result.

Description

Technical Field

[0001] The present invention relates to the field of augmented reality technology, and in particular to an AR matching and repair method and system for damaged components of ancient buildings. Background Art

[0002] The restoration of ancient buildings is a core topic in the field of cultural heritage protection, carrying the dual missions of historical and cultural heritage and technological innovation. Its importance lies in preserving the material and intangible values ​​of ancient buildings through scientific methods and maintaining the material carriers of human civilization. However, when faced with damaged components, traditional restoration methods often rely on manual experience and judgment, which is limited by insufficient precision, low efficiency, and poor adaptability to traditional processes. In particular, for components with complex geometries, manual measurement and matching make it difficult to achieve high-precision joints, and the structural stability of repaired components is difficult to quantify and verify. These shortcomings make it difficult to strike a balance between aesthetics and functionality in restoration results, limiting the sustainable development of ancient building protection.

[0003] In this field, the precise matching and joining of damaged components has become a core challenge. The geometric shape of damaged components is complex and irregular, and existing technologies find it difficult to accurately capture and reconstruct their three-dimensional models, resulting in insufficient fit of the joint surfaces during the matching process. The low fit of the joint surfaces further affects the precise determination of the joint angle, making it difficult for the repaired components to achieve ideal structural stability during installation. The lack of structural stability places higher demands on the process adaptability of traditional mortise and tenon structures, because mortise and tenon structures not only require geometric matching, but also need to meet specific force distribution characteristics to ensure the long-term stability of the repaired components. These technical factors are closely linked. Unresolved geometric matching problems directly lead to difficulties in optimizing the joint angle, and the deviation of the joint angle complicates the force analysis of the mortise and tenon structure, ultimately affecting the overall reliability of the repair effect.

[0004] Therefore, how to use augmented reality technology to achieve precise geometric matching of damaged components, optimize the joint angle, and combine the characteristics of traditional mortise and tenon structures to conduct force analysis and process guidance has become a key issue in AR matching and repair of damaged components of ancient buildings. Summary of the Invention

[0005] The present invention provides an AR matching repair method for damaged components of ancient buildings, which mainly includes:

[0006] The surface point cloud data of damaged components is obtained through augmented reality technology. Laser scanning and multi-view stereo vision fusion methods are used to generate high-precision three-dimensional point cloud models, thereby obtaining a digital representation of the complex geometric shape of the components.

[0007] For the point cloud model, voxel filtering and normal estimation algorithms are used to perform denoising and surface smoothing, extract component edge features and surface geometric properties, and obtain a smooth 3D geometric model;

[0008] The iterative closest point (ICP) algorithm is used to perform initial registration of the joint surfaces of the damaged and repaired components from the smoothed 3D geometric model. The initial fit of the joint surface point cloud is calculated to determine whether the joint surface meets the preset fit threshold T1 (T1 = 0.95). If the fit is lower than T1, the registration parameters are adjusted and the registration is repeated to obtain a high-fit initial joint surface registration result.

[0009] According to the initial joint surface registration results, the joint surface normal vector is fitted using the least squares method to calculate the relative angle deviation between the joint surfaces of the two components. The joint angle is optimized until the deviation is less than the preset angle threshold T2 (T2 = 0.5 degrees) to obtain the optimized joint angle parameters.

[0010] Based on the optimized joint angle parameters, the finite element analysis method is used to simulate the force distribution of the mortise and tenon structure at this joint angle. The stress concentration area and deformation of the mortise and tenon structure are extracted to determine whether the force distribution meets the preset force threshold T3 (T3 = 80% of the traditional mortise and tenon force standard). If not, the joint angle parameters are adjusted and the simulation is repeated to obtain a mortise and tenon structure model that meets the force distribution characteristics.

[0011] Using augmented reality technology, the optimized mortise and tenon structure model is spatially mapped to the actual component position, generating virtual joint guide lines and force distribution visualization images to guide the component installation process and obtain high-precision component installation results.

[0012] From the component installation results, structured light scanning technology was used to obtain the actual 3D model of the installed component. Geometric deviation analysis was performed between this model and the optimized mortise and tenon structure model. The structural stability parameters of the component installation were calculated to determine whether the stability met the preset stability threshold T4 (T4 = 95% design stability). The final reliability assessment results of the repaired component were obtained.

[0013] Based on the final reliability assessment results, digital twin technology is used to record the geometric matching data, joint angle parameters and force distribution characteristics of the repaired components, generate digital process guidance files adapted to traditional processes, and obtain process templates suitable for subsequent repairs.

[0014] The present invention provides an AR matching and repair system for damaged components of ancient buildings, which mainly includes:

[0015] The point cloud data acquisition module is used to obtain point cloud data on the surface of damaged components through augmented reality technology. It uses laser scanning and multi-view stereo vision fusion methods to generate a high-precision three-dimensional point cloud model and obtain a digital expression of the complex geometric shape of the component;

[0016] The point cloud processing module is used to perform denoising and surface smoothing on the point cloud model using voxel filtering and normal estimation algorithms, extracting component edge features and surface geometric properties to obtain a smooth three-dimensional geometric model;

[0017] The initial registration module is used to perform initial registration of the joint surfaces of the damaged component and the repaired component from the smoothed 3D geometric model using the iterative closest point algorithm (ICP). The module calculates the initial fit of the joint surface point cloud and determines whether the joint surface meets the preset fit threshold T1 (T1 = 0.95). If the fit is lower than T1, the registration parameters are adjusted and the registration is performed again to obtain a high-fit initial joint surface registration result.

[0018] The joint angle optimization module is used to fit the joint surface normal vector using the least squares method based on the initial joint surface alignment results, calculate the relative angle deviation between the joint surfaces of the two components, optimize the joint angle until the deviation is less than the preset angle threshold T2 (T2 = 0.5 degrees), and obtain the optimized joint angle parameters;

[0019] The stress analysis module is used to simulate the stress distribution of the mortise and tenon structure at the optimized joint angle parameters using the finite element analysis method, extract the stress concentration area and deformation of the mortise and tenon structure, and determine whether the stress distribution meets the preset stress threshold T3 (T3 = 80% of the traditional mortise and tenon stress standard). If not, the joint angle parameters are adjusted and the simulation is repeated to obtain a mortise and tenon structure model that meets the stress distribution characteristics;

[0020] The installation guidance module is used to spatially map the optimized mortise and tenon structure model with the actual component position through augmented reality technology, generate virtual joint guide lines and force distribution visualization images, guide the component installation process, and obtain high-precision component installation results;

[0021] The stability assessment module is used to obtain the actual three-dimensional model of the installed component from the component installation results using structured light scanning technology, perform geometric deviation analysis with the optimized mortise and tenon structure model, calculate the structural stability parameters of the component installation, determine whether the stability meets the preset stability threshold T4 (T4 = 95% design stability), and obtain the final reliability assessment result of the repaired component;

[0022] The process template generation module is used to record the geometric matching data, joint angle parameters and force distribution characteristics of the repaired components based on the final reliability assessment results using digital twin technology, generate digital process guidance files adapted to traditional processes, and obtain process templates suitable for subsequent repairs.

[0023] The present invention includes the following beneficial effects:

[0024] The present invention obtains a high-precision three-dimensional point cloud model of the damaged component, performs surface treatment and feature extraction, and achieves precise alignment of the joint surface between the repair component and the damaged component. By optimizing the joint angle and force analysis of the mortise and tenon structure, the structural stability of the repair component is ensured. Augmented reality technology is used to guide the installation of the component, and the installation accuracy is evaluated through structured light scanning. Finally, a digital process guidance file is generated. This method realizes the digitization and precision of the traditional building component repair process, improves the repair efficiency and quality, and provides an innovative solution for the protection of traditional buildings. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] For better understanding and implementation, the technical solution of the present application is described in detail below with reference to the accompanying drawings.

[0026] Figure 1 This is a flow chart of an AR matching repair method for damaged components of ancient buildings according to the present invention.

[0027] Figure 2 Schematic diagram of an AR matching repair method and system for damaged components of ancient buildings according to the present invention.

[0028] Figure 3 This is another schematic diagram of an AR matching repair method and system for damaged components of ancient buildings according to the present invention.

[0029] Figure 4 This is a structural diagram of an AR matching repair method and system for damaged components of ancient buildings according to the present invention. DETAILED DESCRIPTION

[0030] To further illustrate the technical means and effectiveness of the present invention in achieving its intended purpose, exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all possible embodiments consistent with the present application. Rather, they are merely examples of methods and systems consistent with certain aspects of the present application, as detailed in the appended claims.

[0031] The terms used in this application are for the purpose of describing specific embodiments only and are not intended to limit this application. As used in this application and the appended claims, the singular forms "a," "an," "the," and "the" are intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0032] The following describes in detail the specific implementation methods, features and effects of the present invention in conjunction with the accompanying drawings and preferred embodiments.

[0033] Example 1

[0034] like Figure 1-3 As shown, the AR matching repair method and system for damaged components of ancient buildings in this embodiment may specifically include:

[0035] Step S101: Obtain surface point cloud data of the damaged component through augmented reality technology, and use laser scanning and multi-view stereo vision fusion method to generate a high-precision three-dimensional point cloud model to obtain a digital expression of the complex geometric shape of the component.

[0036] The surface point cloud data of the damaged component is obtained by using augmented reality technology, and the initial point cloud set is captured using a preset sensor array to obtain a first point cloud data set;

[0037] If there is noise in the first point cloud data set, the point cloud data is denoised using a mean filtering algorithm to obtain a second point cloud data set;

[0038] Laser scanning technology is used to perform high-resolution scanning on the damaged components to obtain high-density point cloud data and obtain the third point cloud data set;

[0039] Capturing image sequences of components at different angles through multi-view stereo vision technology, reconstructing a three-dimensional point cloud, and obtaining a fourth point cloud data set;

[0040] If the spatial coordinates of the second, third, and fourth point cloud datasets are not aligned, point cloud registration is performed using an iterative closest point algorithm to obtain a unified fifth point cloud dataset;

[0041] Based on the fifth point cloud data set, a Poisson reconstruction algorithm is used to generate a high-precision three-dimensional point cloud model to obtain a digital geometric expression of the component;

[0042] The surface details of the three-dimensional point cloud model are enhanced by stereo microscopy technology to obtain the final complex geometric digital model.

[0043] For example, when acquiring point cloud data on the surface of a damaged component, augmented reality technology can project a virtual point cloud grid through a head-mounted display device or a mobile terminal to assist in sensor positioning.

[0044] For example, when repairing wooden beams of ancient buildings, technicians wear augmented reality glasses to capture the surface morphology of the beam in real time, generating an initial point cloud set with a resolution of about 1 mm / point, covering about 80% of the beam surface;

[0045] This method is intuitive and efficient, and enhances the accuracy of surface data acquisition for complex components;

[0046] For denoising of the first point cloud dataset, the mean filtering algorithm smoothes the noise by calculating the mean of the neighborhood point coordinates.

[0047] In one possible implementation, for the timber beam point cloud data, a neighborhood radius of 2 mm is set, and noise points that deviate from the mean by more than 1.5 standard deviations are removed to obtain a second point cloud dataset. After denoising, the point cloud smoothness is improved by approximately 30%, effectively reducing subsequent registration errors. When laser scanning technology generates high-density point cloud data, a fixed laser scanner can be used.

[0048] It should be noted that when scanning the same wooden beam, the equipment collected data point by point at a spacing of 0.5 mm to generate a third point cloud dataset with a point cloud density of 1,000 points per square centimeter, which is 5 times more accurate than the first point cloud dataset. This high-resolution data provides a reliable foundation for subsequent model reconstruction. Multi-view stereo vision technology reconstructs three-dimensional point clouds through multi-angle image sequences.

[0049] Specifically, an industrial camera was used to shoot from four angles: front, back, left, and right. 20 frames of images were collected at each angle, with an inter-frame overlap rate of 70%. The fourth point cloud dataset was reconstructed through triangulation, with approximately 2 million points. This method compensates for the occlusion problem of a single perspective and enhances the integrity of the point cloud. If the coordinates of the second, third, and fourth point cloud datasets are not aligned, the iterative nearest point algorithm can achieve alignment.

[0050] In one embodiment, the third point cloud dataset is used as a reference to calculate the nearest point distances of other point clouds, and the rotation and translation matrices are iteratively optimized. The registration error is controlled within 0.1 mm to generate a unified fifth point cloud dataset. This registration process ensures the spatial consistency of multi-source point clouds and improves model accuracy. The Poisson reconstruction algorithm generates a three-dimensional model based on the fifth point cloud dataset.

[0051] Preferably, taking the point cloud of a wooden beam as an example, the algorithm estimates the point cloud normal vector, constructs an implicit surface, and generates a triangular mesh model with a surface error of less than 0.2 mm. This model fully expresses the geometric details of the wooden beam, facilitating subsequent analysis and repair design. Stereoscopic microscopy technology further enhances the surface details of the model.

[0052] For example, in the micro-crack area on the surface of the wooden beam, a stereo microscope was used to scan at a 50x magnification to extract texture and crack depth information, which was then integrated into a three-dimensional model with a crack width resolution of 0.01 mm. The final complex geometric digital model not only accurately restored the morphology of the wooden beam, but also highlighted the microscopic damage characteristics, providing a scientific basis for the formulation of repair plans.

[0053] It's no secret that this technological chain, through multi-source data fusion and sophisticated processing, has significantly improved the accuracy and completeness of the digitized representation of damaged components. The resulting models can be used for structural analysis, restoration simulation, and cultural heritage preservation, significantly improving restoration efficiency and effectiveness.

[0054] In step S102 , a voxel filter and a normal estimation algorithm are used to perform denoising and surface smoothing on the point cloud model, extracting component edge features and surface geometric properties to obtain a smooth three-dimensional geometric model.

[0055] The point cloud model is denoised using a voxel filtering algorithm to generate the first point cloud data set;

[0056] Calculate the surface normal vectors of the first point cloud dataset using a normal estimation algorithm to generate a second point cloud dataset. If the second point cloud dataset contains areas of normal vector discontinuity, determine the discontinuity points using local curvature analysis to generate an edge point set.

[0057] According to the edge point set, the surface fitting technology is used to generate the initial geometric model to obtain the first geometric model

[0058] Enhance the surface details of the first geometric model by using stereo microscopy technology to generate a second geometric model;

[0059] If the second geometric model has surface defects, the defects are repaired through the mesh optimization algorithm to obtain the final geometric model;

[0060] According to the final geometric model, stereo projection technology is used to generate 3D reconstructed views and obtain a 3D view dataset.

[0061] Exemplarily, the voxel filtering algorithm performs denoising on the point cloud model by dividing the point cloud data into a regular voxel grid.

[0062] For example, for point cloud data of ancient stone pillars, voxel filtering divides the space into a cubic grid with a side length of 3 mm. The centroid of the point cloud within each voxel replaces the original point to generate the first point cloud dataset. This method simplifies the data by reducing the point cloud density, preserving the main geometric features and facilitating subsequent processing. The normal estimation algorithm calculates the surface normal vector based on the first point cloud dataset.

[0063] It should be noted that for the stone pillar surface point cloud, the algorithm uses principal component analysis to estimate normal vectors, centering on each point and selecting neighboring points with a radius of 5 mm. This generates a second point cloud dataset. Normal vectors reflect surface orientation and provide a basis for subsequent surface analysis. If the second point cloud dataset contains areas of normal vector discontinuity, local curvature analysis can identify these discontinuities.

[0064] Preferably, for the point cloud on the stone pillar surface, the algorithm calculates the neighborhood curvature of each point, sets the curvature threshold to 0.1, extracts points with sudden changes in curvature, and generates edge point sets. These edge points usually correspond to the corners or damaged areas of the stone pillar, providing key information for geometric modeling. Based on the edge point set, the surface fitting technology generates the initial geometric model.

[0065] Specifically, B-spline surface fitting was used to construct a continuous surface at the edge points of the stone pillar, generating a first geometric model. This model initially expressed the geometric shape of the stone pillar and was suitable for further detail enhancement. Surface detail enhancement of this first geometric model was performed using stereo microscopy.

[0066] In one embodiment, a stereo microscope scans the surface of the stone pillar at a 30x magnification, extracts tiny pits and texture information, and integrates it into the first geometric model to generate a second geometric model. This method improves the surface realism of the model and facilitates damage analysis. If the second geometric model has surface defects, the mesh optimization algorithm can repair the defects.

[0067] For example, for local mesh holes in a stone pillar model, the algorithm fills the holes with Laplacian smoothing, adjusts the mesh vertex positions, and generates a final geometric model. This model has improved surface continuity and is suitable for subsequent view generation. Based on this final geometric model, stereoscopic projection technology is used to generate 3D reconstructed views.

[0068] It is understandable that for the stone pillar model, the technology generates a three-dimensional view dataset including front, back, left and right perspectives through multi-angle projection, with a resolution of 1920×1080 pixels for each view. This dataset intuitively presents the overall picture of the stone pillar and provides support for restoration design and display.

[0069] In step S103, the iterative closest point (ICP) algorithm is used to perform initial registration of the joint surfaces of the damaged component and the repaired component from the smoothed 3D geometric model, and the preliminary degree of fit of the joint surface point cloud is calculated to determine whether the joint surface meets the preset fit threshold T1 (T1 = 0.95). If the fit is lower than T1, the registration parameters are adjusted and re-registration is performed to obtain an initial joint surface registration result with high fit.

[0070] The iterative closest point algorithm is used to perform initial registration on the joint surface of the damaged component and the repaired component in the smooth geometric model, and the degree of fit of the point cloud data is calculated to obtain the initial registration result.

[0071] If the degree of fit of the initial registration result is lower than the preset threshold, the registration parameters are adjusted using the least squares method, and the registration is performed again to obtain the optimized registration result;

[0072] Based on the optimized registration results, local geometric features are extracted from the point cloud data of the joint surface to generate a joint surface feature set. The principal component analysis algorithm is used to reduce the dimension of the joint surface feature set to obtain low-dimensional feature vectors and determine the feature vector set.

[0073] If the distribution variance of the feature vector set is lower than a preset threshold, the joint surface is locally geometrically adjusted by using a grid subdivision technique to obtain an adjusted joint surface;

[0074] According to the adjusted joint surface, a three-dimensional mesh model of the joint surface is generated to obtain a final joint surface model;

[0075] The final joint surface model is rendered using stereoscopic projection technology to generate a three-dimensional view dataset.

[0076] In one possible implementation, when the iterative closest point algorithm is used for point cloud registration, the closest point pair between two sets of point clouds is found and the transformation matrix is ​​gradually optimized to achieve geometric alignment of the joint surfaces of the damaged component and the repaired component.

[0077] For example, in a cultural relic restoration scenario, for a broken pottery, the algorithm first roughly aligns the point cloud of the damaged part with the point cloud of the repaired part. Assuming that the maximum number of iterations in the initial iteration is set to 50 and the point-to-point distance threshold is 0.5 mm, the average distance error of the two sets of point clouds is calculated through multiple iterations to obtain the initial alignment result.

[0078] It should be noted that the core of the iterative closest point algorithm is to gradually reduce the geometric deviation between point clouds through transformation matrix optimization.

[0079] Specifically, if the initial registration accuracy is lower than a preset threshold, for example, the average distance error is greater than 0.3 mm, the least squares method is used to adjust the registration parameters.

[0080] In one embodiment, the least square method optimizes the rotation and translation parameters by constructing an error function.

[0081] For example, for the pottery joint surface, the algorithm recalculates the transformation matrix and sets the new number of iterations to 30. The adjusted error is reduced to 0.1 mm, generating an optimized alignment result.

[0082] It can be understood that the optimized registration improves the geometric consistency of the joint surface and lays the foundation for subsequent feature extraction.

[0083] Preferably, when extracting local geometric features of the joint surface from the optimized registration result, the algorithm focuses on the curvature and normal vector changes of the point cloud.

[0084] For example, in pottery restoration, the joint surface may contain curved edges and high curvature areas. The algorithm analyzes the neighborhood characteristics of the point cloud, extracts points with curvature values ​​greater than 0.2, and generates a joint surface feature set. This feature set contains about 5,000 key points that describe the geometric details of the joint surface.

[0085] It should be noted that the accuracy of feature extraction directly affects the reliability of subsequent analysis.

[0086] In one embodiment, a principal component analysis algorithm performs dimensionality reduction processing on the joint surface feature set to reduce computational complexity.

[0087] For example, the original dimension of the feature set is reduced from 10 dimensions to 3 dimensions, 95% of the variance information is retained, and a low-dimensional feature vector set is generated. Assuming that the distribution variance of the feature vector set is 0.05, which is lower than the preset threshold of 0.1, it indicates that the feature distribution is too concentrated, which may lead to the loss of geometric details. At this time, the joint surface geometry is adjusted through mesh subdivision technology.

[0088] For example, for the local concave and convex areas of the pottery joint surface, mesh subdivision reduces the edge length of the original triangle mesh from 2 mm to 0.5 mm, generating a smoother geometric structure.

[0089] For example, when the adjusted joint surface is used to generate a three-dimensional mesh model, the algorithm converts the point cloud into a continuous triangular mesh based on the Delaunay triangulation technology. It is assumed that the final mesh contains 100,000 triangular faces, covering all the geometric features of the pottery joint surface.

[0090] Preferably, the stereoscopic projection technology performs view rendering on the final joint surface model to generate a multi-angle three-dimensional view data set.

[0091] For example, the rendering process sets five viewing angles with a resolution of 1920x1080 to generate a set of views containing joint surface details, which can be used for subsequent evaluation or display of restoration effects.

[0092] It can be understood that each step of the above method revolves around the improvement of the geometric accuracy of the joint surface, from alignment to feature extraction, dimensionality reduction, geometric adjustment to final rendering, forming a complete logical chain. The implementation methods of each step support each other through the cultural relics restoration scenario, jointly ensuring the high-quality generation of the joint surface model.

[0093] In step S104, based on the initial joint surface alignment result, the joint surface normal vector is fitted using the least squares method, the relative angle deviation between the joint surfaces of the two components is calculated, and the joint angle is optimized until the deviation is less than the preset angle threshold T2 (T2 = 0.5 degrees), thereby obtaining the optimized joint angle parameters.

[0094] The least squares method is used to fit the normal vectors of the damaged and repaired components from the joint surface point cloud data to obtain an initial normal vector set. Based on the initial normal vector set, the relative angular deviation between the joint surface of the damaged and repaired components is calculated to obtain the angular deviation value.

[0095] If the angle deviation value is greater than the preset threshold, the registration parameters are iteratively adjusted through the gradient descent method, the registration position of the joint surface point cloud is updated, and the adjusted angle deviation value is obtained;

[0096] According to the adjusted angle deviation value, local curvature features are extracted from the joint surface point cloud data to generate a joint surface curvature feature set;

[0097] The principal component analysis algorithm is used to reduce the dimension of the joint surface curvature feature set, obtain low-dimensional curvature feature vectors, and determine the feature vector set;

[0098] If the distribution variance of the feature vector set is lower than the preset threshold, the mesh smoothing technology is used to perform geometric optimization on the joint surface point cloud to obtain the optimized joint surface point cloud model;

[0099] According to the optimized joint surface point cloud model, the three-dimensional mesh structure of the joint surface is generated to obtain the final joint surface mesh model.

[0100] In a possible implementation, when the least squares method is used to fit the normal vector, the local planar characteristics of the joint surface can be calculated using the coordinates of each point in the point cloud data.

[0101] For example, the point cloud data of a damaged component contains thousands of points. The local plane is fitted using the least squares method to obtain the normal vector, which reflects the orientation of the joint surface.

[0102] Preferably, the point cloud at the edge of the joint surface is selected and fitted to obtain an initial normal vector set to avoid noise interference in the central area. Assuming that the normal vector of a damaged component is [0.1, 0.2, 0.98] and that of the repaired component is [0.12, 0.18, 0.97], the normal vector set can be used for subsequent angular deviation analysis. This method ensures that the normal vector accurately reflects the geometric characteristics, facilitating subsequent alignment adjustments.

[0103] Specifically, when calculating the relative angle deviation, the angle formula between the normal vectors can be used to obtain the joint surface deviation between the damaged component and the repaired component.

[0104] For example, the normal vector angle is about 2.5 degrees. If the preset threshold is 1 degree, the deviation exceeds the standard.

[0105] It should be noted that the angle deviation reflects the degree of fit between the joint surfaces of the two components. Excessive deviation may lead to unstable joints. By recording the deviation value, it can provide a basis for subsequent optimization.

[0106] In one embodiment, when adjusting the registration parameters using the gradient descent method, the rotation and translation parameters of the point cloud may be iteratively updated based on the angular deviation.

[0107] For example, after the initial registration, the point cloud of the damaged component needs to be rotated 0.5 degrees around the Z axis and translated 0.2 mm. The gradient descent method gradually reduces the deviation to 0.8 degrees through multiple iterations to meet the threshold requirements. This method gradually optimizes the registration position through small-step adjustments and improves the fitting accuracy of the joint surface.

[0108] For example, when extracting local curvature features, the curvature value of each point in the point cloud can be calculated to generate a curvature feature set. Assume that the curvature value of a certain joint surface area is between 0.01 and 0.05, indicating that the surface is relatively smooth.

[0109] Preferably, high curvature areas are selected as feature points to generate a feature set that reflects the geometric changes of the joint surface. This feature extraction helps to identify key geometric information of the joint surface.

[0110] It can be understood that when principal component analysis is used for dimensionality reduction, the high-dimensional curvature feature set is mapped to a low-dimensional space.

[0111] For example, the original feature set contains 10-dimensional data. After dimensionality reduction, 3-dimensional main eigenvectors are retained with a distribution variance of 0.02. If the variance is lower than the threshold of 0.05, it indicates that the eigenvector effectively captures the geometric characteristics. This dimensionality reduction process simplifies the data, retains key information, and facilitates subsequent optimization.

[0112] In one possible implementation, when mesh smoothing technology optimizes the joint surface point cloud, irregular mesh points in the point cloud can be adjusted using a Laplace smoothing algorithm.

[0113] For example, the grid point spacing in a certain area is uneven, but after smoothing, the point spacing tends to 0.1 mm, and the surface is more continuous. This optimization improves the geometric consistency of the joint surface model.

[0114] Specifically, when generating the final joint surface mesh model, a triangular mesh structure can be constructed based on the optimized point cloud.

[0115] For example, a mesh model consisting of approximately 5,000 triangular facets was generated for the interface between a damaged component and a repaired one, creating a smooth surface with no noticeable gaps. This mesh model provides a high-quality geometric foundation for subsequent processing or analysis.

[0116] Step S105, for the optimized joint angle parameters, the finite element analysis method is used to simulate the force distribution of the mortise and tenon structure at the joint angle, the stress concentration area and deformation of the mortise and tenon structure are extracted, and it is determined whether the force distribution meets the preset force threshold T3 (T3 = 80% of the traditional mortise and tenon force standard). If not, the joint angle parameters are adjusted and the simulation is repeated to obtain a mortise and tenon structure model that meets the force distribution characteristics.

[0117] Finite element analysis is used to generate an initial mesh model of the mortise and tenon structure from the optimized joint angle parameters to obtain initial force distribution data.

[0118] Extract stress concentration area and deformation characteristics from initial stress distribution data to generate stress distribution feature set;

[0119] The stress distribution feature set is reduced in dimension by the principal component analysis algorithm to obtain low-dimensional stress feature vectors and determine the feature vector set;

[0120] If the distribution variance of the feature vector set is lower than the preset threshold, the initial mesh model is geometrically adjusted using mesh optimization technology to obtain an optimized mortise and tenon structure mesh model;

[0121] Extract key stress points from the optimized mortise and tenon structure grid model, generate a stress distribution feature map, and determine whether the stress distribution meets the preset stress threshold T3;

[0122] If the force distribution does not meet the preset threshold T3, the joint angle parameters are iteratively adjusted using the gradient descent method, the mesh model is updated, and the finite element analysis is re-performed to obtain new force distribution data;

[0123] The stress concentration area and deformation are extracted from the new force distribution data to generate the final mortise and tenon structure model that meets the force distribution characteristics.

[0124] For example, a finite element analysis method is used to generate an initial mesh model of the mortise and tenon structure, aiming to simulate the stress conditions at the joint.

[0125] For example, in the design of wooden mortise and tenon structures, the initial mesh model of the tenon and mortise joint can be divided into tetrahedral meshes using 3D modeling software. The mesh size is controlled at around 1 mm to balance calculation accuracy and efficiency. The acquisition of initial force distribution data relies on applying simulated loads, such as 500N of vertical pressure, to analyze the stress distribution and deformation of the mortise and tenon joint.

[0126] It should be noted that finite element analysis can reveal the stress concentration phenomenon at the edge of the tenon, which provides a data basis for subsequent optimization. The stress concentration area and deformation characteristics are extracted from the initial force distribution data to form a stress distribution feature set.

[0127] Specifically, stress concentration areas usually appear at the sharp corners where the tenon and mortise contact. The maximum stress value may reach 20 MPa and the deformation may be 0.2 mm. By recording the positions and values ​​of these feature points, a feature set containing stress peak values ​​and deformation distribution is formed.

[0128] For example, this feature set can be presented as a heat map through data visualization tools to intuitively display the weak areas of the mortise and tenon structure. The stress distribution feature set can be reduced in dimensionality through the principal component analysis algorithm to obtain a low-dimensional stress feature vector.

[0129] In one possible implementation, the original feature set contains data from thousands of stress points. After principal component analysis, 95% of the variance is retained and the dimension is reduced to a 10-dimensional feature vector. This dimensionality reduction process can reduce computational complexity while retaining key stress distribution information.

[0130] Preferably, if the distribution variance of the feature vector set is lower than a preset threshold, such as 0.01, it means that the feature extraction is stable and suitable for further optimization. If the distribution variance of the feature vector set meets the requirements, the initial grid model is adjusted through grid optimization technology.

[0131] For example, for stress concentration areas, the local mesh density is increased to 0.5 mm, and a smoothing algorithm is used to eliminate sharp geometric features to generate an optimized mortise and tenon structure mesh model. This optimization can improve the uniformity of stress distribution.

[0132] It should be noted that the mesh-optimized model exhibits a lower stress peak in the stress analysis, such as a drop from 20 MPa to 15 MPa. Key stress points are extracted from the optimized mesh model to generate a stress distribution characteristic diagram to determine whether the preset stress threshold T3 is met, such as if the maximum stress is less than 18 MPa.

[0133] It can be understood that the force distribution characteristic diagram displays the stress distribution through color coding, which is convenient for intuitively judging the stability of the model. If the requirements are not met, the joint angle parameters are iteratively adjusted through the gradient descent method.

[0134] For example, the tenon angle is adjusted from 90 degrees to 89.5 degrees, the mesh model is regenerated and finite element analysis is performed to obtain new force distribution data. The new data may show that the stress peak is reduced to 16 MPa, which meets the threshold requirement.

[0135] In one embodiment, the stress concentration area and deformation amount are extracted from the new force distribution data to generate the final mortise and tenon structure model.

[0136] For example, the final model achieved uniform stress along the tenon edges, with maximum deformation controlled within 0.15mm. This model ensures the long-term stability of the mortise and tenon structure.

[0137] Preferably, the final model can also be prototyped using 3D printing technology to conduct physical testing to verify the reliability of the analysis results. This method significantly improves the mechanical properties of the mortise and tenon structure through multi-step optimization.

[0138] Step S106: spatially map the optimized mortise and tenon structure model to the actual component position through augmented reality technology to generate virtual joint guide lines and force distribution visualization images to guide the component installation process and obtain high-precision component installation results.

[0139] The optimized mortise and tenon structure model is scanned in three dimensions using augmented reality technology to obtain the position coordinate data of the model and the actual components. A spatial coordinate mapping matrix M is generated, where M contains the correspondence between the model coordinates (x_m, y_m, z_m) and the component coordinates (x_c, y_c, z_c), and the spatial coordinate mapping result is determined.

[0140] Extract the relative position deviation between the model and the component from the spatial coordinate mapping matrix M, optimize the deviation using the least squares method, and generate the corrected coordinate mapping matrix M', where M' contains the adjusted coordinate correspondence, to obtain the corrected spatial coordinate mapping;

[0141] Based on the corrected coordinate mapping matrix M', a virtual joint guide line L is generated by an augmented reality rendering engine, where L represents the three-dimensional path curve of the mortise and tenon joint, and the spatial position of the virtual joint guide line is determined;

[0142] Extract the force distribution data from the optimized mortise and tenon structure model, combine it with the corrected coordinate mapping matrix M', and generate a force distribution visualization image V through an augmented reality rendering engine. V uses a color gradient to represent the stress intensity, thus obtaining a force distribution visualization result.

[0143] If the deviation between the virtual joint guide line L and the actual component position exceeds the preset threshold T1, the coordinate mapping matrix M' is adjusted through the iterative optimization algorithm, the virtual joint guide line L is updated, and the force distribution visualization image V is regenerated to determine whether the deviation requirement is met;

[0144] Based on the virtual joint guide line L and the force distribution visualization image V, the augmented reality display device outputs the installation guidance information, generates the component installation path P, where P includes the installation sequence and angle parameters, and obtains the component installation guidance result;

[0145] The component position data after installation is extracted from the component installation path P, combined with the spatial coordinate mapping matrix M', and the geometric consistency detection algorithm is used to determine whether the installation accuracy meets the preset threshold T2 to obtain a high-precision installation result.

[0146] For example, the application of augmented reality technology in mortise and tenon structure optimization involves technical topics such as 3D spatial scanning, coordinate mapping, virtual guide line generation, force distribution visualization, and installation guidance. The following analysis and examples address each topic, focusing on the mortise and tenon structure optimization scenario, maintaining a consistent scope, rigorous logic, and progressive approach. Augmented reality 3D spatial scanning is used to obtain positional data between the mortise and tenon model and the actual component.

[0147] For example, a laser radar scanner is used to scan the mortise and tenon components with a resolution of up to 0.1 mm to generate high-precision point cloud data. The scanning process needs to be carried out under stable light to avoid reflection interference. The point cloud data contains model coordinates and component coordinates, forming the basis of initial spatial data. The beneficial effect of this is to ensure the accuracy of subsequent coordinate mapping. The generation of the spatial coordinate mapping matrix M is based on the scan data. In one possible implementation method, the model coordinates (x_m, y_m, z_m) obtained by scanning and the component coordinates (x_c, y_c, z_c) are aligned through an algorithm to generate a matrix M.

[0148] For example, the model coordinate point (10, 20, 30) corresponds to the component coordinate point (10.2, 20.1, 30.3). The matrix M records these pairing relationships. The matrix M reflects the spatial relationship between the model and the component, providing data support for deviation analysis. The relative position deviation optimization uses the least squares method to generate the correction matrix M'.

[0149] Specifically, if the deviation between the model and the component is 0.3 mm, the least squares method adjusts the coordinate pair through iterative calculation to generate M' with a deviation less than 0.1 mm.

[0150] Preferably, multiple iterations can improve mapping accuracy. The advantage of this method is that it reduces position errors during installation. The virtual joint guide line L is generated by an augmented reality rendering engine and is expressed as a three-dimensional path curve.

[0151] For example, the rendering engine draws the guide line L of the mortise and tenon joint according to M', and the curve is highlighted in green to indicate the installation path.

[0152] It should be noted that the deviation between the guide line and the actual component position must be less than T1 (e.g., 0.2 mm). Otherwise, an iterative optimization algorithm will adjust M' and update L. This approach intuitively guides workers and improves installation efficiency. The force distribution visualization image V uses a color gradient to represent stress intensity.

[0153] In one embodiment, red indicates high-stress areas (e.g., 100 MPa) and green indicates low-stress areas (e.g., 20 MPa). The rendering engine, combined with M', overlays the force data onto the mortise and tenon model, forming an intuitive image V. It is understood that workers can use image V to quickly identify stress concentration areas, adjust installation strategies, and reduce structural risks. Installation guidance information is output as the component installation path P via the augmented reality device.

[0154] For example, path P indicates that the tenon should be installed first, with a 45-degree rotation angle, from left to right. The device displays P on a head-mounted display, and workers follow the instructions. This approach reduces human error and improves installation consistency. Geometric consistency testing determines whether the installation accuracy meets threshold T2.

[0155] In a possible implementation, the component position data after installation is compared with M'. If the deviation is less than T2 (eg, 0.15 mm), it is determined to be qualified.

[0156] For example, the detection algorithm found that the deviation of a certain component was 0.1 mm, which met the requirements. This detection ensures the stability and accuracy of the mortise and tenon structure. Through the coordinated application of the above technical themes, the mortise and tenon structure installation process from scanning to detection has formed a closed-loop process, significantly improving accuracy and efficiency. The implementation methods of each theme support each other and jointly ensure the achievement of high-precision installation goals.

[0157] In step S107, the actual three-dimensional model of the installed component is obtained from the component installation results using structured light scanning technology, and a geometric deviation analysis is performed with the optimized mortise and tenon structure model to calculate the structural stability parameters of the component installation. It is determined whether the stability meets the preset stability threshold T4 (T4 = 95% design stability) to obtain the final reliability assessment result of the repaired component.

[0158] The structured light scanning technology is used to perform a three-dimensional scan on the installed component to generate an actual three-dimensional model M1, where M1 contains the component surface point cloud data (x1, y1, z1), and obtain the actual three-dimensional model;

[0159] The actual 3D model M1 is geometrically aligned with the optimized mortise and tenon model M2 through the point cloud registration algorithm, where M2 contains the reference point cloud data (x2, y2, z2), and the alignment matrix T is calculated to obtain the aligned actual 3D model M1';

[0160] According to the aligned actual 3D model M1' and the optimized mortise and tenon model M2, the geometric deviation D between the point clouds is calculated using the Euclidean distance, where D represents the average distance difference between the point clouds, and the geometric deviation data is obtained;

[0161] Extract the deviation distribution feature F from the geometric deviation data D, where F includes the deviation concentration area and the maximum deviation point. Use the principal component analysis algorithm to determine the main direction of the deviation distribution and obtain the deviation distribution feature;

[0162] According to the deviation distribution characteristics F and the optimized mortise and tenon model M2, the structural stability parameter S is calculated, where S includes the stress concentration factor of the component's load point, and the stability parameter is obtained;

[0163] If the stability parameter S is lower than the preset threshold value T4, the stress state of the component is simulated by the finite element analysis algorithm to generate the repair adjustment matrix R, where R contains the component geometric adjustment vector, and the repair adjustment data is obtained;

[0164] Based on the repair adjustment data R and the stability parameter S, the reliability assessment result E is generated, where E includes the stability score of the component after repair, and the final assessment result is obtained.

[0165] For example, structured light scanning technology generates high-precision three-dimensional point cloud data by projecting a specific grating pattern and capturing the reflection from the surface of the component.

[0166] For example, after the mortise and tenon components are installed, a structured light scanner is used to project stripe light to capture the point cloud on the component surface to form the actual three-dimensional model M1. The scanner records the point cloud coordinates (x1, y1, z1) with an accuracy of 0.1 mm, generating an M1 model containing millions of points to ensure the reliability of subsequent geometric analysis. The point cloud registration algorithm is used to align the actual three-dimensional model M1 with the optimized mortise and tenon model M2.

[0167] It can be understood that the registration calculates the rigid transformation matrix T of the two sets of point clouds through an iterative closest point algorithm.

[0168] For example, point cloud M1 contains the surface data of a component after installation, while M2 is the reference model used during the design phase. The registration process identifies common feature points between the two models and generates a matrix T, which transforms M1 into the aligned M1'. The alignment error is kept within 0.05 mm to ensure the accuracy of subsequent deviation calculations. Euclidean distance is used to calculate the geometric deviation D between the point clouds, reflecting the difference between the actual component and the designed model.

[0169] Specifically, the point cloud coordinates of M1' and M2 are compared point by point, and the average distance difference D is calculated.

[0170] For example, the deviation D of a mortise and tenon joint is 0.2 mm, indicating that there is a slight offset in the installation. The D value intuitively quantifies the geometric consistency and provides a data basis for deviation distribution analysis. The deviation distribution feature F is extracted through principal component analysis to extract the deviation concentration area and main direction.

[0171] Preferably, the D data is analyzed to identify that the deviation is concentrated at the tenon joint, with the maximum deviation point being 0.3 mm and the main direction being along the length axis of the component. This feature F helps to locate the installation problem area.

[0172] For example, the local deviation caused by the tenon processing error guides the subsequent stability analysis, and the structural stability parameter S is used to calculate the stress concentration factor based on the deviation distribution characteristics F and the M2 model.

[0173] In one embodiment, due to the concentration of deviation F at the mortise and tenon joint, the S value shows a stress concentration coefficient of 1.5, which is lower than the ideal value of 2.0, indicating insufficient stability. The S value reflects the potential risk of the component under actual stress and provides a basis for repair. The finite element analysis algorithm simulates the stress state of the component and generates a repair adjustment matrix R.

[0174] For example, if S is lower than the threshold value T4, the simulation shows that the stress in the tenon area exceeds the limit, and the R matrix is ​​generated, indicating that the geometry is adjusted by 0.1 mm along the Z axis. R ensures that the geometry and force of the component are balanced after repair, extending the service life. The reliability evaluation result E is based on R and S to generate a stability score.

[0175] For example, after repair, the S value increased to 1.8, and the E score was 90, indicating a significant improvement in component stability. The E results provide a quantitative basis for installation quality and support project acceptance.

[0176] It should be noted that the above method revolves around the post-installation analysis of mortise and tenon components, and is linked together from scanning to {MD} scanning, alignment to stability assessment, forming a complete analysis chain.

[0177] For example, structured light scanning ensures high-precision data acquisition, registration and deviation calculation of positioning errors, feature extraction and stability analysis to identify risks, repair and evaluation to optimize component performance. This logically rigorous analysis chain ensures high-precision installation and long-term reliability of mortise and tenon components.

[0178] In step S108, based on the final reliability assessment results, digital twin technology is used to record the geometric matching data, joint angle parameters, and force distribution characteristics of the repaired component, generate a digital process guidance file adapted to the traditional process, and obtain a process template suitable for subsequent repairs.

[0179] Digital twin technology is used to obtain geometric matching data G, joint angle parameters A, and force distribution characteristics F from the repaired component, where G contains the point cloud coordinates of the component surface, A contains the joint surface angle value, and F contains the stress distribution of the force point, thus obtaining digital description data of the component;

[0180] The component digital description data is processed through a data mapping algorithm, and the geometric matching data G, the joint angle parameter A, and the force distribution characteristics F are matched with the traditional process parameter library to generate an initial process parameter set P, where P includes the processing path and angle constraints.

[0181] Based on the initial process parameter set P, the stress distribution characteristic F' of the component under the processing path is calculated using the force characteristic analysis method. If F' does not match the expected force distribution threshold T5, P is adjusted through the iterative optimization algorithm to generate the optimized process parameter set P'.

[0182] By using a digital process generation tool, the optimized process parameter set P' is converted into a digital process guidance file D, where D contains processing instructions and parameter sequences, thereby obtaining a digital process guidance file;

[0183] Using template adaptation rules, key process parameters are extracted from the digital process guidance file D to generate a process template M suitable for subsequent repair, where M includes a universal processing path and an angle template, thus obtaining a repair process template.

[0184] According to the repair process template M and the component repair record R, the process parameter optimization method is used to adjust the processing path and angle parameters in M ​​to generate a dedicated process template M' adapted to the specific component, thus obtaining a dedicated process template;

[0185] The digital archive data S of component repair is generated through the component repair record R and the dedicated process template M', where S contains the repair process parameters and template adaptation record, and the repair archive data is obtained.

[0186] For example, the application of digital twin technology in component repair realizes data collection and analysis by constructing virtual models. For example, in the repair of traditional wooden buildings, the surface point cloud data of the repaired component can be obtained through a high-precision laser scanner to form geometric matching data G, which contains three-dimensional coordinate information such as x, y, and z point sets with an accuracy of up to 0.1 mm. The joint angle parameter A records the angle between the mortise and tenon joint surfaces, such as 45 degrees or 90 degrees. The force distribution characteristic F captures the stress values ​​of key parts of the component through sensors, such as the stress peak of 100 MPa at the main beam connection. These data together constitute a digital description of the component, providing a basis for subsequent process matching.

[0187] In one possible implementation, a data mapping algorithm compares the digital description data with a traditional process parameter library that stores historical restoration cases, such as tool paths and angle constraints for mortise and tenon machining.

[0188] For example, for a certain tenon, the algorithm matches the tool path as linear feed with an angle constraint of ±2 degrees, and generates an initial process parameter set P.

[0189] It should be noted that the matching process takes the component material into consideration. For example, hardwood requires stricter angle control to ensure processing accuracy.

[0190] Specifically, the force characteristic analysis verifies the feasibility of the initial process parameter set P by simulating the stress distribution under the processing path.

[0191] For example, simulations showed that stress in a particular mortise and tenon joint concentrated at a single point after machining, peaking at 120 MPa, exceeding the expected threshold T5 of 100 MPa. In this case, the iterative optimization algorithm adjusted the tool path to a curved feed, generating the optimized process parameter set P' that evened out the stress distribution and reduced the peak stress to 90 MPa.

[0192] Preferably, the optimization process can also introduce the wood grain direction to reduce processing stress.

[0193] In one embodiment, the digital process generation tool converts the optimized process parameter set P' into a machining instruction, such as a G code for a CNC machine tool, and generates a digital process instruction file D.

[0194] For example, the instruction sequence specifies that the tool cuts at a speed of 0.5 mm / s and the angle deviation is controlled within 0.1 degrees. Process instruction file D ensures precise execution of the processing equipment and improves the consistency of the repair.

[0195] It can be understood that the template adaptation rule extracts key parameters from the process instruction file D, such as the processing path length of 50 mm and the angle of 90 degrees, to generate the general process template M.

[0196] For example, template M is suitable for batch repairs of similar mortise and tenon components, reducing duplication of design. Furthermore, based on component repair records R, if a component requires a 2mm widening of the tenon due to aging, the process parameter optimization method adjusts template M to generate a dedicated process template M', ensuring it is suitable for that specific component.

[0197] For example, by using component repair records R and dedicated process templates M', digital archive data S is generated to record processing path adjustments and stress distribution changes. For example, if the stress peak is stabilized at 85 MPa after repair, the archive data S supports subsequent review and process improvements, ensuring the traceability of the repair process. This method improves repair accuracy and efficiency through digital means and extends the service life of components.

[0198] Example 2

[0199] like Figure 4 As shown, this embodiment provides an AR matching and repair system for damaged components of ancient buildings, which mainly includes:

[0200] The point cloud data acquisition module is used to obtain point cloud data on the surface of damaged components through augmented reality technology. It uses laser scanning and multi-view stereo vision fusion methods to generate a high-precision three-dimensional point cloud model and obtain a digital expression of the complex geometric shape of the component;

[0201] The point cloud processing module is used to perform denoising and surface smoothing on the point cloud model using voxel filtering and normal estimation algorithms, extracting component edge features and surface geometric properties to obtain a smooth three-dimensional geometric model;

[0202] The initial registration module is used to perform initial registration of the joint surfaces of the damaged component and the repaired component from the smoothed 3D geometric model using the iterative closest point algorithm (ICP). The module calculates the initial fit of the joint surface point cloud and determines whether the joint surface meets the preset fit threshold T1 (T1 = 0.95). If the fit is lower than T1, the registration parameters are adjusted and the registration is performed again to obtain a high-fit initial joint surface registration result.

[0203] The joint angle optimization module is used to fit the joint surface normal vector using the least squares method based on the initial joint surface alignment results, calculate the relative angle deviation between the joint surfaces of the two components, optimize the joint angle until the deviation is less than the preset angle threshold T2 (T2 = 0.5 degrees), and obtain the optimized joint angle parameters;

[0204] The stress analysis module is used to simulate the stress distribution of the mortise and tenon structure at the optimized joint angle parameters using the finite element analysis method, extract the stress concentration area and deformation of the mortise and tenon structure, and determine whether the stress distribution meets the preset stress threshold T3 (T3 = 80% of the traditional mortise and tenon stress standard). If not, the joint angle parameters are adjusted and the simulation is repeated to obtain a mortise and tenon structure model that meets the stress distribution characteristics;

[0205] The installation guidance module is used to spatially map the optimized mortise and tenon structure model with the actual component position through augmented reality technology, generate virtual joint guide lines and force distribution visualization images, guide the component installation process, and obtain high-precision component installation results;

[0206] The stability assessment module is used to obtain the actual three-dimensional model of the installed component from the component installation results using structured light scanning technology, perform geometric deviation analysis with the optimized mortise and tenon structure model, calculate the structural stability parameters of the component installation, determine whether the stability meets the preset stability threshold T4 (T4 = 95% design stability), and obtain the final reliability assessment result of the repaired component;

[0207] The process template generation module is used to record the geometric matching data, joint angle parameters and force distribution characteristics of the repaired components based on the final reliability assessment results using digital twin technology, generate digital process guidance files adapted to traditional processes, and obtain process templates suitable for subsequent repairs.

[0208] The above is only a preferred embodiment of the present invention. It should be pointed out that ordinary technicians in this technical field can make several improvements and supplements without departing from the principles of the present invention. These improvements and supplements should also be regarded as the scope of protection of the present invention.

Claims

1. An AR matching repair method for damaged components of ancient buildings, characterized in that: The method comprises: Obtain surface point cloud data of damaged components and generate high-precision three-dimensional point cloud models; For the point cloud model, denoising and surface smoothing are performed to extract the edge features and surface geometric characteristics of the component to obtain a smooth three-dimensional geometric model; Perform initial registration of the joint surfaces between the damaged component and the repaired component from the smoothed 3D geometric model, calculate the initial fit of the joint surface point cloud, and determine whether the joint surface meets the preset fit threshold; According to the initial joint surface registration result, the joint surface normal vector is fitted, the relative angle deviation between the joint surfaces of the two components is calculated, and the joint angle is optimized until the deviation is less than the preset angle threshold to obtain the optimized joint angle parameter; Based on the optimized joint angle parameters, the force distribution of the mortise and tenon structure at the joint angle is simulated, the stress concentration area and deformation of the mortise and tenon structure are extracted, and it is determined whether the force distribution meets the preset force threshold; Using augmented reality technology, the optimized mortise and tenon structure model is spatially mapped to the actual component position, generating virtual joint guide lines and force distribution visualization images to guide the component installation process and obtain high-precision component installation results. From the component installation results, the actual three-dimensional model of the installed component is obtained, and a geometric deviation analysis is performed with the optimized mortise and tenon structure model. The structural stability parameters of the component installation are calculated to determine whether the stability meets the preset stability threshold, and the final reliability assessment result of the repaired component is obtained.

2. The AR matching repair of damaged components of ancient buildings according to claim 1 is characterized in that: The digital expression of the complex geometric shape of the component is obtained by: The surface point cloud data of the damaged component is obtained by using augmented reality technology, and the initial point cloud set is captured using a preset sensor array to obtain a first point cloud data set; If there is noise in the first point cloud data set, the point cloud data is denoised using a mean filtering algorithm to obtain a second point cloud data set; Laser scanning technology is used to perform high-resolution scanning on the damaged components to obtain high-density point cloud data and obtain the third point cloud data set; Capturing image sequences of components at different angles through multi-view stereo vision technology, reconstructing a three-dimensional point cloud, and obtaining a fourth point cloud data set; If the spatial coordinates of the second, third, and fourth point cloud datasets are not aligned, point cloud registration is performed using an iterative closest point algorithm to obtain a unified fifth point cloud dataset; Based on the fifth point cloud data set, a Poisson reconstruction algorithm is used to generate a high-precision three-dimensional point cloud model to obtain a digital geometric expression of the component; The surface details of the three-dimensional point cloud model are enhanced by stereo microscopy technology to obtain the final complex geometric digital model.

3. The AR matching repair of damaged components of ancient buildings according to claim 1 is characterized in that: The step of obtaining a smooth three-dimensional geometric model comprises: The point cloud model is denoised using a voxel filtering algorithm to generate the first point cloud data set; Calculating a surface normal vector for the first point cloud data set using a normal estimation algorithm to generate a second point cloud data set; If there is a normal vector discontinuity area in the second point cloud data set, the discontinuity point is determined by local curvature analysis to generate an edge point set; According to the edge point set, an initial geometric model is generated by using a surface fitting technique to obtain a first geometric model; Enhance the surface details of the first geometric model by using stereo microscopy technology to generate a second geometric model; If the second geometric model has surface defects, the defects are repaired through the mesh optimization algorithm to obtain the final geometric model; According to the final geometric model, stereo projection technology is used to generate 3D reconstructed views and obtain a 3D view dataset.

4. The AR matching repair of damaged components of ancient buildings according to claim 1 is characterized in that: The initial joint surface registration result with high consistency includes: The iterative closest point algorithm is used to perform initial registration on the joint surface of the damaged component and the repaired component in the smooth geometric model, and the degree of fit of the point cloud data is calculated to obtain the initial registration result. If the degree of fit of the initial registration result is lower than the preset threshold, the registration parameters are adjusted using the least squares method, and the registration is performed again to obtain the optimized registration result; According to the optimized registration results, local geometric features are extracted from the point cloud data of the joint surface to generate a joint surface feature set; The principal component analysis algorithm is used to reduce the dimension of the joint surface feature set, obtain low-dimensional feature vectors, and determine the feature vector set; If the distribution variance of the feature vector set is lower than a preset threshold, the joint surface is locally geometrically adjusted by using a grid subdivision technique to obtain an adjusted joint surface; According to the adjusted joint surface, a three-dimensional mesh model of the joint surface is generated to obtain a final joint surface model; The final joint surface model is rendered using stereoscopic projection technology to generate a three-dimensional view dataset.

5. The AR matching repair of damaged components of ancient buildings according to claim 1 is characterized in that: The optimized bonding angle parameters include: The normal vectors of the damaged component and the repaired component are fitted from the joint surface point cloud data using the least squares method to obtain the initial normal vector set; The relative angle deviation between the joint surface of the damaged component and the repaired component is calculated through the initial normal vector set to obtain the angle deviation value; If the angle deviation value is greater than the preset threshold, the registration parameters are iteratively adjusted through the gradient descent method, the registration position of the joint surface point cloud is updated, and the adjusted angle deviation value is obtained; According to the adjusted angle deviation value, local curvature features are extracted from the joint surface point cloud data to generate a joint surface curvature feature set; The principal component analysis algorithm is used to reduce the dimension of the joint surface curvature feature set, obtain low-dimensional curvature feature vectors, and determine the feature vector set; If the distribution variance of the feature vector set is lower than the preset threshold, the mesh smoothing technology is used to perform geometric optimization on the joint surface point cloud to obtain the optimized joint surface point cloud model; According to the optimized joint surface point cloud model, the three-dimensional mesh structure of the joint surface is generated to obtain the final joint surface mesh model.

6. The AR matching repair of damaged components of ancient buildings according to claim 1 is characterized in that: The determining whether the force distribution meets the preset force threshold comprises: Finite element analysis is used to generate an initial mesh model of the mortise and tenon structure from the optimized joint angle parameters to obtain initial force distribution data. Extract stress concentration area and deformation characteristics from initial stress distribution data to generate stress distribution feature set; The stress distribution feature set is reduced in dimension by the principal component analysis algorithm to obtain low-dimensional stress feature vectors and determine the feature vector set; If the distribution variance of the feature vector set is lower than the preset threshold, the initial mesh model is geometrically adjusted using mesh optimization technology to obtain an optimized mortise and tenon structure mesh model; Extract key stress points from the optimized mortise and tenon structure grid model, generate a stress distribution feature map, and determine whether the stress distribution meets the preset stress threshold T3; If the force distribution does not meet the preset threshold T3, the joint angle parameters are iteratively adjusted using the gradient descent method, the mesh model is updated, and the finite element analysis is re-performed to obtain new force distribution data; The stress concentration area and deformation are extracted from the new force distribution data to generate the final mortise and tenon structure model that meets the force distribution characteristics.

7. The AR matching repair of damaged components of ancient buildings according to claim 1 is characterized in that: The high-precision component installation result is obtained as follows: The optimized mortise and tenon structure model is scanned in three dimensions using augmented reality technology to obtain the position coordinate data of the model and the actual components. A spatial coordinate mapping matrix M is generated, where M contains the correspondence between the model coordinates (x_m, y_m, z_m) and the component coordinates (x_c, y_c, z_c), and the spatial coordinate mapping result is determined. Extract the relative position deviation between the model and the component from the spatial coordinate mapping matrix M, optimize the deviation using the least squares method, and generate the corrected coordinate mapping matrix M', where M' contains the adjusted coordinate correspondence, to obtain the corrected spatial coordinate mapping; Based on the corrected coordinate mapping matrix M', a virtual joint guide line L is generated by an augmented reality rendering engine, where L represents the three-dimensional path curve of the mortise and tenon joint, and the spatial position of the virtual joint guide line is determined; Extract the force distribution data from the optimized mortise and tenon structure model, combine it with the corrected coordinate mapping matrix M', and generate a force distribution visualization image V through an augmented reality rendering engine. V uses a color gradient to represent the stress intensity, thus obtaining a force distribution visualization result. If the deviation between the virtual joint guide line L and the actual component position exceeds the preset threshold T1, the coordinate mapping matrix M' is adjusted through the iterative optimization algorithm, the virtual joint guide line L is updated, and the force distribution visualization image V is regenerated to determine whether the deviation requirement is met; Based on the virtual joint guide line L and the force distribution visualization image V, the augmented reality display device outputs the installation guidance information, generates the component installation path P, where P includes the installation sequence and angle parameters, and obtains the component installation guidance result; The component position data after installation is extracted from the component installation path P, combined with the spatial coordinate mapping matrix M', and the geometric consistency detection algorithm is used to determine whether the installation accuracy meets the preset threshold T2 to obtain a high-precision installation result.

8. The AR matching repair of damaged components of ancient buildings according to claim 1 is characterized in that: The final reliability evaluation result of the repaired component includes: The structured light scanning technology is used to perform a three-dimensional scan on the installed component to generate an actual three-dimensional model M1, where M1 contains the component surface point cloud data (x1, y1, z1), and obtain the actual three-dimensional model; The actual 3D model M1 is geometrically aligned with the optimized mortise and tenon model M2 through the point cloud registration algorithm, where M2 contains the reference point cloud data (x2, y2, z2), and the alignment matrix T is calculated to obtain the aligned actual 3D model M1'; According to the aligned actual 3D model M1' and the optimized mortise and tenon model M2, the geometric deviation D between the point clouds is calculated using the Euclidean distance, where D represents the average distance difference between the point clouds, and the geometric deviation data is obtained; Extract the deviation distribution feature F from the geometric deviation data D, where F includes the deviation concentration area and the maximum deviation point. Use the principal component analysis algorithm to determine the main direction of the deviation distribution and obtain the deviation distribution feature; According to the deviation distribution characteristics F and the optimized mortise and tenon model M2, the structural stability parameter S is calculated, where S includes the stress concentration factor of the component's load point, and the stability parameter is obtained; If the stability parameter S is lower than the preset threshold value T4, the stress state of the component is simulated by the finite element analysis algorithm to generate the repair adjustment matrix R, where R contains the component geometric adjustment vector, and the repair adjustment data is obtained; Based on the repair adjustment data R and the stability parameter S, the reliability assessment result E is generated, where E includes the stability score of the component after repair, and the final assessment result is obtained.

9. The AR matching repair method for damaged components of ancient buildings according to claim 1, characterized in that: Also includes: Based on the final reliability assessment results, the geometric matching data, joint angle parameters, and force distribution characteristics of the repaired components are recorded, and a digital process guidance file adapted to the traditional process is generated to obtain a process template suitable for subsequent repairs. The process template suitable for subsequent repair is obtained as follows: Digital twin technology is used to obtain geometric matching data G, joint angle parameters A, and force distribution characteristics F from the repaired component, where G contains the point cloud coordinates of the component surface, A contains the joint surface angle value, and F contains the stress distribution of the force point, thus obtaining digital description data of the component; The component digital description data is processed through a data mapping algorithm, and the geometric matching data G, the joint angle parameter A, and the force distribution characteristics F are matched with the traditional process parameter library to generate an initial process parameter set P, where P includes the processing path and angle constraints. Based on the initial process parameter set P, the stress distribution characteristic F' of the component under the processing path is calculated using the force characteristic analysis method. If F' does not match the expected force distribution threshold T5, P is adjusted through the iterative optimization algorithm to generate the optimized process parameter set P'. By using a digital process generation tool, the optimized process parameter set P' is converted into a digital process guidance file D, where D contains processing instructions and parameter sequences, thereby obtaining a digital process guidance file; Using template adaptation rules, key process parameters are extracted from the digital process guidance file D to generate a process template M suitable for subsequent repair, where M includes a universal processing path and an angle template, thus obtaining a repair process template. According to the repair process template M and the component repair record R, the process parameter optimization method is used to adjust the processing path and angle parameters in M ​​to generate a dedicated process template M' adapted to the specific component, thus obtaining a dedicated process template; The digital archive data S of component repair is generated through the component repair record R and the dedicated process template M', where S contains the repair process parameters and template adaptation record, and the repair archive data is obtained.

10. An AR matching and repair system for damaged components of ancient buildings, characterized in that: include: The point cloud data acquisition module is used to obtain point cloud data on the surface of damaged components through augmented reality technology. It uses laser scanning and multi-view stereo vision fusion methods to generate a high-precision three-dimensional point cloud model and obtain a digital expression of the complex geometric shape of the component; The point cloud processing module is used to perform denoising and surface smoothing on the point cloud model using voxel filtering and normal estimation algorithms, extracting component edge features and surface geometric properties to obtain a smooth three-dimensional geometric model; The initial registration module is used to perform initial registration of the joint surfaces of the damaged component and the repaired component from the smoothed 3D geometric model using the iterative closest point algorithm (ICP). The module calculates the initial fit of the joint surface point cloud and determines whether the joint surface meets the preset fit threshold T1 (T1 = 0.95). If the fit is lower than T1, the registration parameters are adjusted and the registration is performed again to obtain a high-fit initial joint surface registration result. The joint angle optimization module is used to fit the joint surface normal vector using the least squares method based on the initial joint surface alignment results, calculate the relative angle deviation between the joint surfaces of the two components, optimize the joint angle until the deviation is less than the preset angle threshold T2 (T2 = 0.5 degrees), and obtain the optimized joint angle parameters; The stress analysis module is used to simulate the stress distribution of the mortise and tenon structure at the optimized joint angle parameters using the finite element analysis method, extract the stress concentration area and deformation of the mortise and tenon structure, and determine whether the stress distribution meets the preset stress threshold T3 (T3 = 80% of the traditional mortise and tenon stress standard). If not, the joint angle parameters are adjusted and the simulation is repeated to obtain a mortise and tenon structure model that meets the stress distribution characteristics; The installation guidance module is used to spatially map the optimized mortise and tenon structure model with the actual component position through augmented reality technology, generate virtual joint guide lines and force distribution visualization images, guide the component installation process, and obtain high-precision component installation results; The stability assessment module is used to obtain the actual three-dimensional model of the installed component from the component installation results using structured light scanning technology, perform geometric deviation analysis with the optimized mortise and tenon structure model, calculate the structural stability parameters of the component installation, determine whether the stability meets the preset stability threshold T4 (T4 = 95% design stability), and obtain the final reliability assessment result of the repaired component; The process template generation module is used to record the geometric matching data, joint angle parameters and force distribution characteristics of the repaired components based on the final reliability assessment results using digital twin technology, generate digital process guidance files adapted to traditional processes, and obtain process templates suitable for subsequent repairs.

Images