A concrete mesostructure modeling system based on three-dimensional laser scanning

Abstract

The present application relates to the technical field of three-dimensional modeling, and particularly relates to a concrete mesostructure modeling system based on three-dimensional laser scanning, which comprises a scanning posture cooperative acquisition module, a concrete point cloud enhancement fusion module, and a mesostructure segmentation reconstruction and feedback control module; the scanning posture cooperative acquisition module is used for controlling a three-dimensional laser scanner and an electric control turntable to perform multi-posture scanning on a concrete sample, and generating a multi-view concrete original point cloud set; the concrete point cloud enhancement fusion module is used for performing denoising, smoothing, reflection compensation and registration fusion on the original point cloud set, and generating a concrete enhanced fusion point cloud model and a model completeness evaluation result; the mesostructure segmentation reconstruction and feedback control module is used for performing supplementary scanning feedback or three-phase semantic segmentation, and generating a concrete multi-phase mesostructure parameterized model. The present application realizes high-completeness, high-fidelity and automatic parameterized modeling of the concrete mesostructure, and improves the point cloud acquisition precision, model reconstruction quality and numerical calculation applicability of the complex mesostructure region.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional modeling technology, and in particular to a concrete microstructure modeling system based on three-dimensional laser scanning. Background Technology

[0002] Concrete, as a typical multiphase heterogeneous composite material, exhibits a close correlation between its macroscopic mechanical properties, durability, and crack propagation behavior and its internal microstructure. Aggregate morphology, pore distribution, mortar encapsulation state, and interfacial transition zone structure all directly affect the strength, permeability, and service life of concrete materials. Therefore, in concrete material research, numerical simulation analysis, and digital material design, it is necessary to perform high-precision three-dimensional reconstruction of the internal microstructure of concrete to obtain accurate information on aggregate spatial distribution, pore connectivity, and mortar encapsulation boundaries. This provides a reliable data foundation for finite element analysis, discrete element analysis, and digital twin modeling.

[0003] Existing methods for modeling the microstructure of concrete mostly employ two-dimensional slice image reconstruction, industrial CT scanning, or traditional three-dimensional laser scanning to obtain information about the internal structure of concrete. Two-dimensional image analysis struggles to accurately reflect the spatial morphology of aggregates and the connectivity of pores. Industrial CT equipment suffers from high cost, low scanning efficiency, and poor adaptability to large-sized concrete samples. Traditional three-dimensional laser scanning technology is prone to outlier noise, reflection distortion, and localized occlusion when dealing with rough, low-reflection, and porous concrete surfaces, leading to broken point cloud edges and insufficient model integrity. Furthermore, existing technologies generally lack dynamic supplementation mechanisms for occluded areas, low-reflection distortion areas, and broken pore edges, and the point cloud segmentation process relies on manual, empirical thresholds, making it difficult to achieve automated parameter reconstruction of the three-phase structure of aggregates, mortar, and pores. Summary of the Invention

[0004] This invention provides a concrete microstructure modeling system based on three-dimensional laser scanning. By constructing a scanning posture collaborative acquisition module, a concrete point cloud enhancement and fusion module, and a microstructure segmentation, reconstruction, and feedback control module, it achieves multi-posture collaborative scanning of concrete samples, point cloud enhancement and fusion, feedback scanning of defect areas, and semantic segmentation and reconstruction of aggregate, mortar, and pores. This improves the point cloud coverage integrity, reflection stability, and edge continuity of complex microstructure areas, and solves the technical problems of low accuracy, insufficient automation, and difficulty in directly generating numerical calculation models in existing concrete microstructure modeling technologies.

[0005] A concrete microstructure modeling system based on 3D laser scanning includes a scanning posture collaborative acquisition module, a concrete point cloud enhancement and fusion module, and a microstructure segmentation, reconstruction, and feedback control module, wherein; The scanning posture collaborative acquisition module carries the concrete sample to be tested and controls the three-dimensional laser scanner and the electric control turntable to perform horizontal rotation, tilting and lifting linkage according to the outer contour, initial point cloud coverage state and surface reflection response of the concrete sample to be tested, generating multi-view original concrete point cloud sets corresponding to different scanning postures. The multi-view original concrete point cloud sets include point cloud data of the outer surface of the sample, the exposed boundary of aggregate, the rough area of ​​mortar and the pore depression area. The concrete point cloud enhancement and fusion module receives the multi-view original concrete point cloud set, performs outlier noise removal, local plane fitting and smoothing, adaptive compensation of reflection intensity and multi-view registration and fusion on the multi-view original concrete point cloud set, generates a concrete enhanced fused point cloud model, and identifies occlusion and missing areas, low reflection distortion areas and pore edge fracture areas based on the concrete enhanced fused point cloud model to form a model integrity evaluation result. The microstructure segmentation, reconstruction, and feedback control module receives the concrete-reinforced fused point cloud model and the model integrity evaluation result. When the model integrity evaluation result does not meet the preset modeling requirements, it generates supplementary scanning posture instructions based on the positions of the occlusion and missing areas, low-reflection distortion areas, and pore edge fracture areas, and feeds the supplementary scanning posture instructions back to the scanning posture collaborative acquisition module. When the model integrity evaluation result meets the preset modeling requirements, it performs three-phase semantic segmentation of aggregate, mortar, and pores on the concrete-reinforced fused point cloud model, extracts aggregate particle morphology parameters, pore connectivity topology parameters, and mortar wrapping boundary parameters, and generates a parameterized model of the multiphase microstructure of concrete that can be used for numerical calculation.

[0006] Optionally, the scanning posture collaborative acquisition module includes: Contour acquisition: The concrete sample to be tested is fixed at the bearing center of the electrically controlled turntable, and the initial contour point cloud of the concrete sample to be tested is acquired. Based on the initial contour point cloud, the outer boundary of the sample, the change of surface normal and the location of local depressions are determined to form the initial geometric description result of the sample. Region identification: Based on the initial geometric description of the sample, the sample surface is divided into regions, and the point cloud coverage and reflection response of each surface region are statistically analyzed to identify low coverage regions, low reflection regions and concave occlusion regions, and generate a set of scan compensation regions. Attitude acquisition: Based on the set of scanning compensation areas, the target scanning attitude corresponding to each surface area to be compensated is determined, and the electronically controlled turntable is controlled to perform horizontal rotation, tilting and lifting linkage according to the target scanning attitude, so that the three-dimensional laser scanner can acquire multiple views of the outer surface of the sample, the exposed boundary of the aggregate, the rough area of ​​the mortar and the pore depression area, and generate a multi-view set of original concrete points.

[0007] Optionally, the region identification includes: Surface partitioning: Based on the set of boundary points, surface normal variation and local depression points in the initial geometric description of the sample, the surface of the concrete sample to be tested is divided into grids or neighborhood clusters. Point cloud sampling points that are spatially adjacent and have continuous normal variation are grouped into the same surface region to form a set of surface regions. Coverage statistics: The number of point cloud sampling points and the area of ​​each surface region are counted, the point cloud coverage of each surface region is calculated, and the surface regions with point cloud coverage below the coverage density threshold are marked as low coverage regions. Reflection statistics: The reflection response values ​​of point cloud sampling points in each surface region are statistically analyzed, the average reflection response state of each surface region is calculated, and surface regions with average reflection response values ​​lower than the reflection response threshold are marked as low reflection regions; Indentation identification: Based on the local indentation point set in the initial geometric description of the sample, the number of indentation points and the indentation depth in each surface region are counted, the indentation occlusion degree of each surface region is calculated, and the surface region with the indentation occlusion degree higher than the indentation occlusion threshold is marked as the indentation occlusion region. Compensation generation: Low coverage areas, low reflection areas, and recessed occlusion areas are merged, and duplicate areas are removed to generate a scan compensation area set.

[0008] Optionally, the attitude acquisition includes: Attitude matching: Based on the spatial position, surface normal direction and occlusion opening direction of each surface region to be compensated in the scan compensation area set, calculate the corresponding optimal incident direction and generate a target scan attitude set corresponding to each surface region to be compensated. Attitude optimization: Evaluate the coverage and occlusion removal capabilities of each target scanning attitude in the target scanning attitude set, and select the optimal scanning attitude that simultaneously improves regional coverage and reduces occlusion residue; Linkage control: The electronically controlled turntable performs horizontal rotation, tilting and lifting linkage according to the optimal scanning posture, and simultaneously adjusts the scanning direction and scanning distance of the 3D laser scanner to keep the laser incident direction and the local normal direction of the surface area to be compensated within a preset angle range. Point cloud acquisition: Control the 3D laser scanner to perform multi-angle scanning of the concrete sample under various optimal scanning postures to obtain the original point cloud data corresponding to the outer surface of the sample, the exposed boundary of aggregate, the rough area of ​​mortar and the pore depression area. Data fusion: The original point cloud data acquired under each optimal scanning posture are combined according to the scanning order, and a unified spatial coordinate mapping relationship is established based on the corresponding scanning posture parameters to generate a multi-view concrete original point cloud set.

[0009] Optionally, the concrete point cloud enhancement and fusion module includes: Point cloud denoising: Receive the original point cloud set of concrete from multiple perspectives, perform neighborhood distance statistics and local plane fitting analysis on the original point cloud data under each scanning posture, identify and remove outlier noise points caused by rough scattering of concrete surface, low reflection jump and scanning edge misdetection, and at the same time, perform smooth correction on the point cloud sampling points that deviate from the plane according to the local plane normal direction to generate a smooth denoised point cloud set. Compensation fusion: Based on the smoothed and denoised point cloud, neighborhood statistics are performed on the reflection response values ​​of each point cloud sampling point, and adaptive reflection compensation is performed on low reflection sampling points according to the average reflection distribution of the local area. At the same time, according to the horizontal rotation angle, tilt angle and lifting height corresponding to each scanning posture, the point cloud data under different scanning postures are converted to a unified spatial coordinate system and overlapping area registration fusion is performed to generate a concrete-enhanced fused point cloud model. Complete evaluation: Based on the concrete-reinforced fused point cloud model, the point cloud density, reflection response continuity and pore edge geometric continuity in the fused point cloud are jointly analyzed to identify occlusion and missing areas, low reflection distortion areas and pore edge fracture areas, and the model integrity evaluation results are formed according to the proportion of each defect area.

[0010] Optionally, the compensation fusion includes: Reflection statistics: Based on the smoothed and denoised point cloud, neighborhood statistics are performed on the reflection response values ​​of each point cloud sampling point. The average reflection response value and reflection dispersion in the neighborhood of each point cloud sampling point are calculated. Low reflection sampling points are identified based on the reflection distribution characteristics of local areas to form a set of reflection anomaly points. Reflection compensation: Based on the set of reflection anomalies, adaptive reflection compensation is performed on the low-reflection sampling points, and the compensation results are constrained by the neighborhood average reflection response value and the reflection continuity of the local area, generating a cloud of reflection compensation points. Coordinate transformation: Based on the horizontal rotation angle, tilt angle and lifting height corresponding to each scanning posture, construct the corresponding spatial coordinate transformation matrix, and transform the reflection compensation point cloud under different scanning postures to a unified spatial coordinate system; Registration and fusion: After transforming to a unified spatial coordinate system, the overlapping areas of each point cloud are matched, the spatial offset error between adjacent point clouds is calculated, and point cloud registration and fusion are performed according to the principle of minimum error to generate a concrete-reinforced fused point cloud model.

[0011] Optionally, the complete evaluation includes: Density analysis: The concrete-reinforced fusion point cloud model is divided into multiple local evaluation regions. The number of point clouds and the area of ​​each local evaluation region are counted, the point cloud density is calculated, and local evaluation regions with point cloud density lower than the missing density threshold are identified as occlusion missing regions. Reflection analysis: Continuity analysis is performed on the reflection response values ​​of point cloud sampling points in each local evaluation region. The difference in reflection response between adjacent points is calculated, and local evaluation regions with average reflection response below the low reflection threshold or reflection fluctuation exceeding the reflection continuity threshold are identified as low reflection distortion regions. Edge analysis: Based on the spatial continuity of the edge points of the pore region, the adjacency tracking of the pore edge points is performed to calculate the pore edge interruption length and edge closure degree, and the local evaluation area where the pore edge interruption length is greater than the edge breakage threshold or the edge closure degree is lower than the closure threshold is identified as the pore edge breakage zone; Complete evaluation: The occlusion and missing areas, low reflection distortion areas, and broken areas at the edges of pores are summarized, the proportion of each defect area is calculated, and the model integrity evaluation results are formed based on the effective coverage ratio, the reliable reflection ratio, and the edge continuity ratio.

[0012] Optionally, the mesostructure segmentation, reconstruction, and feedback control module includes: Defect Feedback: Receives concrete-reinforced fusion point cloud model and model integrity evaluation results, performs joint analysis on the spatial location, normal direction, and area of ​​occluded and missing areas, low-reflection distortion areas, and pore edge fracture areas. When the model integrity evaluation result is lower than the model integrity threshold, generates corresponding supplementary scanning posture instructions and feeds them back to the scanning posture collaborative acquisition module to control the 3D laser scanner and electronically controlled turntable to perform supplementary scanning on the defect area. Three-phase segmentation: When the model integrity evaluation result is not lower than the model integrity threshold, based on the spatial coordinates, reflection response value and local geometric continuity of the point cloud sampling points in the concrete-enhanced fusion point cloud model, the point cloud is semantically segmented into aggregate, mortar and pore three phases, and the corresponding three-phase point cloud region set is generated, including aggregate point cloud set, mortar point cloud set and pore point cloud set. Parameter reconstruction: Based on the three-phase point cloud region set, the aggregate particle morphology parameters, pore connectivity topology parameters, and mortar-wrapped boundary parameters are extracted, and a parameterized model of the multiphase microstructure of concrete that can be used for numerical calculation is constructed.

[0013] Optionally, the defect feedback includes: Defect localization: Receive the concrete reinforcement fusion point cloud model and model integrity evaluation results, perform spatial cluster analysis on the occlusion and missing areas, low reflection distortion areas and pore edge fracture areas, extract the center position, regional normal direction and regional area of ​​each defect area, and form a defect area feature set; Attitude generation: Based on the center position, average normal direction and area of ​​each defect region, combined with the current scanning attitude and model integrity evaluation results, the corresponding supplementary scanning attitude command is generated, and the target rotation direction, tilt direction and lifting height of the 3D laser scanner and the electronically controlled turntable are determined. Feedback control: The supplementary scanning posture command is fed back to the scanning posture collaborative acquisition module, which controls the electronically controlled turntable to perform horizontal rotation, tilting and lifting linkage, and simultaneously adjusts the scanning direction and scanning range of the 3D laser scanner, so that the scanning beam preferentially covers the occluded and missing areas, low reflection distortion areas and the broken areas at the edges of pores, thereby completing the supplementary scanning of the defective areas.

[0014] Optionally, the three-phase split includes: Feature extraction: When the model integrity evaluation result is not lower than the model integrity threshold, based on the spatial coordinates, reflection response value and local geometric continuity of each point cloud sampling point in the concrete-reinforced fusion point cloud model, the corresponding spatial distribution features, reflection features and curvature continuity features are extracted to construct the semantic feature vector of the point cloud sampling points. Semantic classification: Based on the semantic feature vectors corresponding to each point cloud sampling point, perform three-phase semantic classification of aggregate, mortar and pores on the point cloud sampling points, and output the semantic category results of the point cloud sampling points according to the feature probabilities corresponding to each category; Region generation: Based on the semantic classification results of point cloud sampling points, region clustering is performed on point cloud sampling points of the same category and spatially continuous, generating aggregate point cloud sets, mortar point cloud sets and pore point cloud sets respectively.

[0015] The beneficial effects of this invention are: This invention, through joint analysis of the shape contour, surface normal variation, and local concave areas of the concrete sample to be tested, and combined with the dynamic generation of target scanning posture by scanning compensation area set, enables the three-dimensional laser scanner and the electrically controlled turntable to perform horizontal rotation, tilting, and lifting linkage scanning. This effectively solves the problems of scanning blind spots and missing point clouds caused by occlusion, low reflection, and concave areas in the existing concrete microstructure scanning process, and improves the point cloud coverage integrity and spatial acquisition accuracy of exposed aggregate boundaries, rough mortar areas, and pore concave areas.

[0016] This invention performs outlier removal, local plane fitting smoothing, adaptive reflection compensation, and multi-pose point cloud registration and fusion processing on the original point cloud set of concrete from multiple perspectives. Based on the point cloud density, reflection response continuity, and geometric continuity of pore edges, it forms a model integrity evaluation result. This effectively reduces the point cloud distortion caused by rough scattering of concrete surface, low reflection jumps, and scanning edge misdetection. It enhances the geometric continuity, reflection stability, and edge fidelity of the concrete-enhanced fused point cloud model, and improves the point cloud fusion quality and modeling reliability of complex microstructure regions.

[0017] This invention generates supplementary scanning posture commands and executes feedback scanning when the model integrity evaluation results do not meet the modeling requirements. When the model integrity evaluation results meet the modeling requirements, it performs three-phase semantic segmentation of aggregate, mortar, and pores on the concrete-reinforced fused point cloud model and further extracts aggregate particle morphology parameters, pore connectivity topology parameters, and mortar-wrapped boundary parameters. This achieves a closed-loop automatic reconstruction process of concrete microstructure from point cloud acquisition and defect compensation to parametric modeling. It solves the problems of existing technologies such as reliance on manual threshold segmentation, low automation of modeling, and difficulty in directly generating numerical calculation models. This improves the efficiency of parametric modeling of concrete microstructure and the applicability of subsequent finite element or discrete element numerical analysis. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the system functional modules according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the concrete point cloud enhancement and fusion module according to an embodiment of the present invention. Detailed Implementation

[0020] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. Those skilled in the art may employ other alternative methods to implement some well-known technologies; moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0021] like Figures 1-2 As shown, a concrete microstructure modeling system based on three-dimensional laser scanning includes a scanning posture collaborative acquisition module, a concrete point cloud enhancement and fusion module, and a microstructure segmentation, reconstruction, and feedback control module, wherein; The scanning posture collaborative acquisition module carries the concrete sample to be tested and controls the three-dimensional laser scanner and the electric control turntable to perform horizontal rotation, tilting and lifting linkage according to the outer contour, initial point cloud coverage state and surface reflection response of the concrete sample to be tested, generating multi-view original concrete point cloud sets corresponding to different scanning postures. The multi-view original concrete point cloud sets include point cloud data of the outer surface of the sample, the exposed boundary of aggregate, the rough area of ​​mortar and the pore depression area. The concrete point cloud enhancement and fusion module receives the original concrete point cloud set from multiple perspectives, performs outlier noise removal, local plane fitting and smoothing, adaptive compensation for reflection intensity, and multi-view registration and fusion on the original concrete point cloud set from multiple perspectives, generates a concrete enhanced fused point cloud model, and identifies occlusion and missing areas, low reflection distortion areas and pore edge fracture areas based on the concrete enhanced fused point cloud model, forming a model integrity evaluation result. The microstructure segmentation, reconstruction, and feedback control module receives the concrete-reinforced fusion point cloud model and the model integrity evaluation results. When the model integrity evaluation results do not meet the preset modeling requirements, it generates supplementary scanning posture commands based on the positions of occluded missing areas, low-reflection distortion areas, and pore edge fracture areas, and feeds the supplementary scanning posture commands back to the scanning posture collaborative acquisition module. When the model integrity evaluation results meet the preset modeling requirements, it performs three-phase semantic segmentation of aggregate, mortar, and pores on the concrete-reinforced fusion point cloud model, extracts aggregate particle morphology parameters, pore connectivity topology parameters, and mortar wrapping boundary parameters, and generates a parameterized model of the multiphase microstructure of concrete that can be used for numerical calculation.

[0022] The scanning posture collaborative acquisition module includes: Contour acquisition: The concrete sample to be tested is fixed at the bearing center of the electrically controlled turntable, and the initial contour point cloud of the concrete sample is acquired. Based on the initial contour point cloud, the outer boundary of the sample, the surface normal variation, and the location of local depressions are determined to form the initial geometric description result of the sample, represented as: ; ; ; ; ; in, For the initial outline point cloud, For the first Each point cloud sampling point , , For the first Spatial coordinates of a point cloud sample. For the first The reflection response value of a point cloud sample. This represents the number of sampling points in the initial contour point cloud. This is the set of boundary points of the sample's outer shape. Point cloud sampling points The corresponding spatial gradient intensity, For boundary recognition threshold, Point cloud sampling points The change in surface normal, Point cloud sampling points The neighborhood point set, The number of neighboring points. Point cloud sampling points Surface normal vector, For neighborhood point cloud sampling points Surface normal vector, For a set of local depression points, Point cloud sampling points The height value of the fitting reference plane in the neighborhood. The threshold for indentation recognition. This is the initial geometric description result of the sample; ; ; in, , These represent the mean and standard deviation of the spatial gradient intensity of all sampling points in the initial contour point cloud, respectively. To enhance the adjustment coefficient at the boundary, , These are the mean and standard deviation of the height residuals of all point cloud sampling points, respectively. The adjustment coefficient is used to enhance the concavity. Region identification: Based on the initial geometric description of the sample, the sample surface is divided into regions, and the point cloud coverage and reflection response of each surface region are statistically analyzed to identify low coverage regions, low reflection regions and concave occlusion regions, and generate a set of scan compensation regions. Attitude acquisition: Based on the set of scanning compensation areas, the target scanning attitude corresponding to each surface area to be compensated is determined, and the electronically controlled turntable is controlled to perform horizontal rotation, tilting and lifting linkage according to the target scanning attitude, so that the three-dimensional laser scanner can acquire multiple views of the outer surface of the sample, the exposed boundary of the aggregate, the rough area of ​​mortar and the pore depression area, and generate a multi-view set of original concrete points.

[0023] Region identification includes: Surface partitioning: Based on the initial geometric description of the specimen, including the set of boundary points, surface normal variation, and local depression points, the surface of the concrete specimen is meshed or clustered. Spatially adjacent point cloud sampling points with continuous normal variation are grouped into the same surface region, forming a set of surface regions, represented as: ; in, For a collection of surface regions, For the first A surface area, This represents the number of surface regions. Coverage statistics: The number of point cloud sampling points and the area of ​​each surface region are statistically analyzed to calculate the point cloud coverage of each surface region. Surface regions with point cloud coverage below the coverage density threshold are marked as low-coverage regions, represented as: ; ; in, For the first Point cloud coverage density of a surface region For the first The number of point cloud sampling points in a surface region For the first Surface area of ​​each surface region For the coverage density threshold, A set of low-coverage areas; ; in, , These represent the mean and standard deviation of the point cloud coverage density for the entire surface region, respectively. This is the cover density adjustment coefficient; Reflection statistics: The reflection response values ​​of point cloud sampling points within each surface region are statistically analyzed, the average reflection response state of each surface region is calculated, and surface regions with average reflection response values ​​below the reflection response threshold are marked as low-reflection regions, represented as: ; ; in, For the first The average reflection response value of each surface region For the first Point cloud sampling points within a surface region The reflection response value, The reflection response threshold, A collection of low-reflection regions; ; in, , These represent the mean and standard deviation of the average reflection response values ​​for all surface regions, respectively. This is the reflection response adjustment coefficient; Indentation identification: Based on the local indentation point set in the initial geometric description of the sample, the number and depth of indentations in each surface region are counted, the indentation occlusion degree of each surface region is calculated, and surface regions with an indentation occlusion degree higher than the indentation occlusion threshold are marked as indentation occlusion regions, represented as: ; ; in, For the first The degree of occlusion in the surface area Point cloud sampling points The height value of the fitting reference plane in the neighborhood. Point cloud sampling points The actual height value, For the concave occlusion threshold, A collection of recessed and occluded areas; ; in, , These represent the mean and standard deviation of the degree of depression occlusion across the entire surface area, respectively. This is the concave occlusion adjustment coefficient; Compensation generation: Low-coverage areas, low-reflection areas, and recessed occlusion areas are merged, and duplicate areas are removed to generate a scan compensation area set, represented as: ; in, For scanning the compensation region set.

[0024] Attitude acquisition includes: Attitude matching: Based on the spatial position, surface normal direction, and occlusion opening direction of each surface region to be compensated in the scan compensation area set, the corresponding optimal incident direction is calculated, and a target scan attitude set corresponding to each surface region to be compensated is generated, represented as: ; ; ; in, For the target scan pose set, For the first The target scanning posture corresponding to the surface area to be compensated. The horizontal rotation angle, The tilt angle, For raising and lowering height, For the first The center coordinates of the surface area to be compensated The coordinates of the center of the electronically controlled turntable. The number of surface areas to be compensated; Attitude optimization: The coverage and occlusion removal capabilities of each target scanning pose in the target scanning pose set are evaluated. The optimal scanning pose that simultaneously improves region coverage and reduces occlusion residue is selected, denoted as: ; ; in, For the first The optimal scanning posture for each surface region to be compensated. This is the attitude evaluation value. For scan coverage gain, For occlusion elimination rate, The time required for attitude adjustment , , These are the corresponding weight coefficients; ; ; ; ; ; ; ; ; in, To execute the first The newly acquired effective point cloud coverage area after scanning the pose of each target This is to determine the total area of ​​all areas to be compensated within the scanned compensation area set. To execute the first The cumulative effective coverage area after scanning the pose of each target To execute the first The cumulative effective coverage area before the target scanning posture To count the number of missing points in the point cloud in the occluded area before scanning, To execute the first The number of missing points in the point cloud in the remaining occluded area after scanning the pose of a target. To execute the first The number of points that remain in the scanning blind zone after the target is scanned. To execute the first The number of points that remain in the shadowed area after the target is scanned. Adjust the time for horizontal rotation. To adjust the tilt time, For the adjustment time of lifting and lowering, , , The first The horizontal rotation angle, tilt angle, and vertical rise / fall height corresponding to the scanning posture of each target. , , These represent the horizontal rotation angle, tilt angle, and vertical rise / fall height corresponding to the previous scan posture. The horizontal rotational speed, To adjust the speed by tilting, Adjust the speed for lifting and lowering; Linkage control: The electronically controlled turntable performs horizontal rotation, tilting, and lifting linkage according to the optimal scanning posture, and simultaneously adjusts the scanning direction and scanning distance of the 3D laser scanner to maintain a preset angle range between the laser incident direction and the local normal direction of the surface area to be compensated. This improves the integrity of point cloud acquisition for exposed aggregate boundaries, rough mortar areas, and pore depression areas, as shown below: ; in, For the first The optimal linkage control parameters. To achieve the optimal horizontal rotation angle, To achieve the optimal tilt angle, To achieve the optimal lifting height; Point cloud acquisition: The 3D laser scanner is controlled to perform multi-angle scanning of the concrete sample under various optimal scanning postures to acquire the original point cloud data corresponding to the outer surface of the sample, the exposed aggregate boundary, the rough area of ​​mortar, and the pore depression area, which is represented as: ; in, For the first The original point cloud dataset obtained under each scanning pose. For point cloud sampling points, , , The spatial coordinates of the point cloud sampling points, The reflection response value of the point cloud sampling point. For the first Number of sampling points under each scanning posture; Data fusion: The raw point cloud data acquired under each optimal scanning posture are combined according to the scanning order, and a unified spatial coordinate mapping relationship is established based on the corresponding scanning posture parameters to generate a multi-view concrete raw point cloud set, represented as: ; in, The original points of the concrete are gathered from multiple perspectives.

[0025] The concrete point cloud enhancement and fusion module includes: Point cloud denoising: Receives original point clouds of concrete from multiple viewpoints. Performs neighborhood distance statistics and local plane fitting analysis on the original point cloud data under each scanning posture. Identifies and removes outlier noise points caused by rough scattering of the concrete surface, low reflectance jumps, and misdetection of scanning edges. Simultaneously, smooths and corrects point cloud sampling points that deviate from the plane according to the local plane normal direction, thereby reducing surface noise while preserving aggregate edge and pore edge features, generating a smoothed and denoised point cloud, represented as: ; ; ; ; in, Point cloud sampling points The average distance to its neighboring points Point cloud sampling points The neighborhood point set, The number of neighboring points. For the first Original point cloud dataset under scanning pose For the first Denoising point aggregation under each scanning posture The mean of the neighborhood distance. The standard deviation of the neighborhood mean distance This is the outlier noise adjustment coefficient. Point cloud sampling points The deviation distance to the local fitting plane, , , , For locally fitted plane parameters, For local plane normal vectors, For the first Smoothing and denoising of cloud clusters under each scanning posture; Compensation fusion: Based on smooth denoising point cloud, neighborhood statistics are performed on the reflection response values ​​of each point cloud sampling point, and adaptive reflection compensation is performed on low reflection sampling points according to the average reflection distribution of the local area. At the same time, according to the horizontal rotation angle, tilt angle and lifting height corresponding to each scanning posture, the point cloud data under different scanning postures are converted to a unified spatial coordinate system and overlapping area registration fusion is performed to generate a concrete-enhanced fused point cloud model. Complete assessment: Based on the concrete-reinforced fused point cloud model, the point cloud density, reflection response continuity and pore edge geometric continuity in the fused point cloud are jointly analyzed to identify occlusion and missing areas, low reflection distortion areas and pore edge fracture areas, and the model integrity evaluation results are formed according to the proportion of each defect area.

[0026] Compensation fusion includes: Reflection statistics: Based on the smoothed and denoised point cloud, neighborhood statistics are performed on the reflection response values ​​of each point cloud sampling point. The average reflection response value and reflection dispersion within the neighborhood of each point cloud sampling point are calculated. Low-reflection sampling points are identified based on the reflection distribution characteristics of local regions, forming a set of reflection anomalies, represented as: ; ; ; in, Point cloud sampling points The average reflection response value within the neighborhood, Point cloud sampling points Standard deviation of reflection response within the neighborhood Point cloud sampling points The neighborhood point set, For neighborhood point cloud sampling points The reflection response value, For the set of reflection anomalies, The adjustment coefficient is used to identify low reflectivity. Reflection Compensation: Based on the set of reflection anomalies, adaptive reflection compensation is performed on low-reflection sampling points. The compensation result is constrained by the average reflection response value of the neighborhood and the reflection continuity of the local area. This ensures that the compensated point cloud retains the characteristics of concrete aggregate boundaries and pore shadows while reducing point cloud distortion caused by local low reflection. A reflection-compensated point cloud set is generated, represented as follows: ; ; ; in, Point cloud sampling points The compensated reflection response value, The reflection response value before compensation. Point cloud sampling points The corresponding adaptive compensation weights, This represents the average reflection response values ​​of all point cloud sampling points under the current scanning posture. For the first A cluster of reflection compensation points under scanning posture. For the first Smoothing and denoising of cloud clusters under each scanning posture; Coordinate transformation: Based on the horizontal rotation angle, tilt angle, and rise / fall height corresponding to each scanning posture, a corresponding spatial coordinate transformation matrix is ​​constructed, and the reflection compensation point set under different scanning postures is transformed to a unified spatial coordinate system, represented as: ; ; ; in, The coordinates of the sampled points in the converted point cloud. For the first The spatial coordinate transformation matrix corresponding to each scanning posture This is the horizontal rotation transformation matrix. The tilt rotation transformation matrix is... For elevation and translation transformation matrices, For the first A collection of points after scanning pose transformation to a unified spatial coordinate system; Registration and fusion: After transforming the point clouds to a unified spatial coordinate system, overlapping regions are matched, the spatial offset error between adjacent point clouds is calculated, and point cloud registration and fusion are performed according to the principle of minimum error to generate a concrete-reinforced fused point cloud model, represented as: ; ; in, For point cloud registration error, For the point cloud sampling points in the current point cloud, For the corresponding matching points in the overlapping region, For the number of matching points, For concrete-reinforced fusion point cloud model, This represents the number of scan poses.

[0027] A complete assessment includes: Density analysis: The concrete-reinforced fusion point cloud model is divided into multiple local evaluation regions. The number of point clouds and the area of ​​each local evaluation region are counted, the point cloud density is calculated, and local evaluation regions with point cloud densities below the missing density threshold are identified as occlusion missing regions, represented as: ; ; in, For the first Point cloud density of a local evaluation region For the first The number of point clouds within a local evaluation region. For the first The area of ​​each local evaluation region For the first A local evaluation area, The missing density threshold, For the set of occluded and missing regions; ; in, , These represent the mean and standard deviation of the point cloud density for the entire local evaluation region, respectively. This is the missing density adjustment coefficient; Reflection analysis: Continuity analysis is performed on the reflection response values ​​of point cloud sampling points within each local evaluation region. The difference in reflection response between adjacent points is calculated, and local evaluation regions with average reflection response below the low reflection threshold or reflection fluctuation exceeding the reflection continuity threshold are identified as low reflection distortion regions, represented as: ; ; ; in, For the first The average reflection response value of each local evaluation area Point cloud sampling points within a local evaluation area The reflection response value, For the first The reflection response fluctuation of a local evaluation area For low reflection threshold, For the reflection continuity threshold, This is a collection of low-reflection distortion regions; ; ; in, , These represent the mean and standard deviation of the average reflection response values ​​for the entire local evaluation area, respectively. For low reflection adjustment coefficient, , These represent the mean and standard deviation of the reflection response fluctuations across all local evaluation areas, respectively. This is the reflection continuity adjustment coefficient; Edge analysis: Based on the spatial continuity of the edge points in the pore region, adjacency tracing is performed on the pore edge points to calculate the pore edge interruption length and edge closure degree. Local evaluation regions where the pore edge interruption length is greater than the edge breakage threshold or the edge closure degree is less than the closure threshold are identified as pore edge breakage zones, represented as: ; ; ; in, For the first The length of the pore edge interruption within a local evaluation region. For the first The length of the segment interrupted at the segment edge, For the first The number of edge interruption segments within a local evaluation region For the first The degree of edge closure of a local evaluation region For the first The length of the edge of the closed pore within a local evaluation area For the first The total length of the pore edge within a local evaluation region The edge breakage threshold, The edge closure threshold, This is a collection of fracture zones at the pore edges; ; ; in, , These represent the mean and standard deviation of the pore edge interruption length in all local evaluation regions, respectively. This is the edge fracture adjustment coefficient. , These represent the mean and standard deviation of the edge closure degree of all local evaluation regions, respectively. This is the edge closure adjustment coefficient; Complete evaluation: The occlusion and missing areas, low-reflection distortion areas, and porosity edge fracture areas are summarized. The proportion of each defective area is calculated, and the model integrity evaluation result is formed based on the effective coverage ratio, reliable reflection ratio, and edge continuity ratio, expressed as: ; ; ; ; in, To cover the percentage of the missing area, The proportion of the low-reflection distortion region. The percentage of the fractured zone at the pore edge. , , These refer to the number or area of ​​areas with missing occlusion, low-reflection distortion, and fractured areas at the edges of pores. This refers to the total number or total area of ​​all local evaluation regions. For the model completeness evaluation results, , , These are the corresponding evaluation weights.

[0028] The detailed structure segmentation and reconstruction and feedback control module includes: Defect Feedback: Receives concrete-reinforced fusion point cloud model and model integrity evaluation results, performs joint analysis on the spatial location, normal direction, and area of ​​occluded and missing areas, low-reflection distortion areas, and pore edge fracture areas. When the model integrity evaluation result is lower than the model integrity threshold, generates corresponding supplementary scanning posture instructions and feeds them back to the scanning posture collaborative acquisition module to control the 3D laser scanner and electronically controlled turntable to perform supplementary scanning on the defect area. Three-phase segmentation: When the model integrity evaluation result is not lower than the model integrity threshold, based on the spatial coordinates, reflection response value and local geometric continuity of the point cloud sampling points in the concrete-enhanced fusion point cloud model, the point cloud is semantically segmented into aggregate, mortar and pore three phases, and the corresponding three-phase point cloud region set is generated, including aggregate point cloud set, mortar point cloud set and pore point cloud set. Parameter Reconstruction: Based on the three-phase point cloud region set, aggregate particle morphology parameters, pore connectivity topology parameters, and mortar-wrapped boundary parameters are extracted, and a parameterized model of the multiphase microstructure of concrete that can be used for numerical calculation is constructed, represented as: ; ; ; ; in, These are the morphological parameters of aggregate particles. The length of the major axis of the aggregate particle. The length of the minor axis of the aggregate particle. For pore connectivity topology parameters, The number of connected pore nodes. This represents the total number of nodes in the pores. For mortar-wrapped boundary parameters, The contact area between mortar and aggregate. The total surface area of ​​the aggregate. For the parameterized model of the multiphase microstructure of concrete, For aggregate parameters, For the set of pore topology parameters, This is the set of boundary parameters for mortar wrapping.

[0029] Defect feedback includes: Defect localization: Receive the concrete-reinforced fusion point cloud model and model integrity evaluation results. Perform spatial cluster analysis on occlusion and missing areas, low-reflection distortion areas, and pore edge fracture areas. Extract the center position, normal direction, and area of ​​each defect region, forming a defect region feature set, represented as: ; ; ; ; in, For the first A set of regional features for each defective region. For the first The center coordinates of the defect area For the first The average normal direction of each defect region For the first The area of ​​each defective region For the first The point cloud set corresponding to each defect region For the first The number of point cloud sampling points in each defect region Point cloud sampling points The normal vector, For the first The first defect area Area of ​​each local grid cell This represents the number of local grid cells; Attitude Generation: Based on the center position, average normal direction, and area of ​​each defect region, combined with the current scanning attitude and model integrity evaluation results, corresponding supplementary scanning attitude commands are generated, and the target rotation direction, tilt direction, and lifting height of the 3D laser scanner and the electronically controlled turntable are determined, as follows: ; ; ; ; ; in, For the model completeness evaluation results, To model the integrity threshold, To supplement the scan attitude instruction set, For the first A supplementary scan attitude command. The horizontal rotation angle, The tilt angle, For raising and lowering height, The coordinates of the center of the electronically controlled turntable. The direction vector is from the center of the electronically controlled turntable to the center of the defect area. This is the current scan height. This is the height adjustment coefficient. This represents the number of defective areas. ; in, , These represent the mean and standard deviation of the historical modeling completeness evaluation results, respectively. This is the integrity adjustment coefficient; Feedback control: The supplementary scanning posture command is fed back to the scanning posture collaborative acquisition module, which controls the electronically controlled turntable to perform horizontal rotation, tilting, and lifting linkage, and simultaneously adjusts the scanning direction and scanning range of the 3D laser scanner, so that the scanning beam preferentially covers the occluded and missing areas, low-reflection distortion areas, and fractured areas at the edges of pores, thereby completing the supplementary scanning of the defective areas, as shown below: ; ; in, For the first The attitude adjustment amount corresponding to each supplementary scan attitude. , , These represent the horizontal rotation angle, tilt angle, and vertical rise / fall height corresponding to the current scanning posture. , , The corresponding weights are adjusted respectively. For the first The scan coverage area corresponding to each defect region For the first The area of ​​each defective region For 3D laser scanners and the first The spatial distance between the centers of the defective areas.

[0030] Three-phase splitting includes: Feature Extraction: When the model integrity evaluation result is not lower than the modeling integrity threshold, based on the spatial coordinates, reflection response values, and local geometric continuity of each point cloud sampling point in the concrete-reinforced fused point cloud model, the corresponding spatial distribution features, reflection features, and curvature continuity features are extracted to construct the semantic feature vector of the point cloud sampling points, represented as: ; ; ; in, Point cloud sampling points The corresponding semantic feature vector, , , The spatial coordinates of the point cloud sampling points, The reflection response value of the point cloud sampling point. The local point cloud density of the point cloud sampling points. The number of neighboring points. Point cloud sampling points The local spatial volume corresponding to the neighborhood. Point cloud sampling points The local geometric continuity parameter, Point cloud sampling points The neighborhood point set, Point cloud sampling points The normal vector, For neighborhood point cloud sampling points The normal vector; Semantic classification: Based on the semantic feature vectors corresponding to each point cloud sampling point, three-phase semantic classification of aggregate, mortar, and porosity is performed on the point cloud sampling points, and the semantic category results of the point cloud sampling points are output according to the feature probabilities corresponding to each category, as shown below: ; ; in, Point cloud sampling points The semantic classification results , , These represent aggregate type, mortar type, and porosity type, respectively. Point cloud sampling points Category The classification probability, For category The corresponding feature weight matrix, For category The corresponding bias parameters; Region Generation: Based on the semantic classification results of the point cloud sampling points, region clustering is performed on the spatially continuous point cloud sampling points of the same category to generate aggregate point cloud sets, mortar point cloud sets, and pore point cloud sets, respectively, as follows: ; ; ; ; in, For aggregate point cloud aggregation, For mortar point cloud collection, For pore point cloud set, This is a collection of three-phase point cloud regions.

[0031] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0032] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A concrete microstructure modeling system based on three-dimensional laser scanning, characterized in that, It includes a scanning posture collaborative acquisition module, a concrete point cloud enhancement and fusion module, and a microstructure segmentation, reconstruction, and feedback control module, among which; The scanning posture collaborative acquisition module carries the concrete sample to be tested and controls the three-dimensional laser scanner and the electric control turntable to perform horizontal rotation, tilting and lifting linkage according to the outer contour, initial point cloud coverage state and surface reflection response of the concrete sample to be tested, generating multi-view original concrete point cloud sets corresponding to different scanning postures. The multi-view original concrete point cloud sets include point cloud data of the outer surface of the sample, the exposed boundary of aggregate, the rough area of ​​mortar and the pore depression area. The concrete point cloud enhancement and fusion module receives the multi-view original concrete point cloud set, performs outlier noise removal, local plane fitting and smoothing, adaptive compensation of reflection intensity and multi-view registration and fusion on the multi-view original concrete point cloud set, generates a concrete enhanced fused point cloud model, and identifies occlusion and missing areas, low reflection distortion areas and pore edge fracture areas based on the concrete enhanced fused point cloud model to form a model integrity evaluation result. The microstructure segmentation, reconstruction, and feedback control module receives the concrete-reinforced fused point cloud model and the model integrity evaluation result. When the model integrity evaluation result does not meet the preset modeling requirements, it generates supplementary scanning posture instructions based on the positions of the occlusion and missing areas, low-reflection distortion areas, and pore edge fracture areas, and feeds the supplementary scanning posture instructions back to the scanning posture collaborative acquisition module. When the model integrity evaluation result meets the preset modeling requirements, it performs three-phase semantic segmentation of aggregate, mortar, and pores on the concrete-reinforced fused point cloud model, extracts aggregate particle morphology parameters, pore connectivity topology parameters, and mortar wrapping boundary parameters, and generates a parameterized model of the multiphase microstructure of concrete that can be used for numerical calculation.

2. The concrete microstructure modeling system based on three-dimensional laser scanning according to claim 1, characterized in that, The scanning posture collaborative acquisition module includes: Contour acquisition: The concrete sample to be tested is fixed at the bearing center of the electrically controlled turntable, and the initial contour point cloud of the concrete sample to be tested is acquired. Based on the initial contour point cloud, the outer boundary of the sample, the change of surface normal and the location of local depressions are determined to form the initial geometric description result of the sample. Region identification: Based on the initial geometric description of the sample, the sample surface is divided into regions, and the point cloud coverage and reflection response of each surface region are statistically analyzed to identify low coverage regions, low reflection regions and concave occlusion regions, and generate a set of scan compensation regions. Attitude acquisition: Based on the set of scanning compensation areas, the target scanning attitude corresponding to each surface area to be compensated is determined, and the electronically controlled turntable is controlled to perform horizontal rotation, tilting and lifting linkage according to the target scanning attitude, so that the three-dimensional laser scanner can acquire multiple views of the outer surface of the sample, the exposed boundary of the aggregate, the rough area of ​​the mortar and the pore depression area, and generate a multi-view set of original concrete points.

3. The concrete microstructure modeling system based on three-dimensional laser scanning according to claim 2, characterized in that, The region identification includes: Surface partitioning: Based on the set of boundary points, surface normal variation and local depression points in the initial geometric description of the sample, the surface of the concrete sample to be tested is divided into grids or neighborhood clusters. Point cloud sampling points that are spatially adjacent and have continuous normal variation are grouped into the same surface region to form a set of surface regions. Coverage statistics: The number of point cloud sampling points and the area of ​​each surface region are counted, the point cloud coverage of each surface region is calculated, and the surface regions with point cloud coverage below the coverage density threshold are marked as low coverage regions. Reflection statistics: The reflection response values ​​of point cloud sampling points in each surface region are statistically analyzed, the average reflection response state of each surface region is calculated, and surface regions with average reflection response values ​​lower than the reflection response threshold are marked as low reflection regions; Indentation identification: Based on the local indentation point set in the initial geometric description of the sample, the number of indentation points and the indentation depth in each surface region are counted, the indentation occlusion degree of each surface region is calculated, and the surface region with the indentation occlusion degree higher than the indentation occlusion threshold is marked as the indentation occlusion region. Compensation generation: Low coverage areas, low reflection areas, and recessed occlusion areas are merged, and duplicate areas are removed to generate a scan compensation area set.

4. The concrete microstructure modeling system based on three-dimensional laser scanning according to claim 3, characterized in that, The attitude acquisition includes: Attitude matching: Based on the spatial position, surface normal direction and occlusion opening direction of each surface region to be compensated in the scan compensation area set, calculate the corresponding optimal incident direction and generate a target scan attitude set corresponding to each surface region to be compensated. Attitude optimization: Evaluate the coverage and occlusion removal capabilities of each target scanning attitude in the target scanning attitude set, and select the optimal scanning attitude that simultaneously improves regional coverage and reduces occlusion residue; Linkage control: The electronically controlled turntable performs horizontal rotation, tilting and lifting linkage according to the optimal scanning posture, and simultaneously adjusts the scanning direction and scanning distance of the 3D laser scanner to keep the laser incident direction and the local normal direction of the surface area to be compensated within a preset angle range. Point cloud acquisition: Control the 3D laser scanner to perform multi-angle scanning of the concrete sample under various optimal scanning postures to obtain the original point cloud data corresponding to the outer surface of the sample, the exposed boundary of aggregate, the rough area of ​​mortar and the pore depression area. Data fusion: The original point cloud data acquired under each optimal scanning posture are combined according to the scanning order, and a unified spatial coordinate mapping relationship is established based on the corresponding scanning posture parameters to generate a multi-view concrete original point cloud set.

5. The concrete microstructure modeling system based on three-dimensional laser scanning according to claim 1, characterized in that, The concrete point cloud enhancement and fusion module includes: Point cloud denoising: Receive the original point cloud set of concrete from multiple perspectives, perform neighborhood distance statistics and local plane fitting analysis on the original point cloud data under each scanning posture, identify and remove outlier noise points caused by rough scattering of concrete surface, low reflection jump and scanning edge misdetection, and at the same time, perform smooth correction on the point cloud sampling points that deviate from the plane according to the local plane normal direction to generate a smooth denoised point cloud set. Compensation fusion: Based on the smoothed and denoised point cloud, neighborhood statistics are performed on the reflection response values ​​of each point cloud sampling point, and adaptive reflection compensation is performed on low reflection sampling points according to the average reflection distribution of the local area. At the same time, according to the horizontal rotation angle, tilt angle and lifting height corresponding to each scanning posture, the point cloud data under different scanning postures are converted to a unified spatial coordinate system and overlapping area registration fusion is performed to generate a concrete-enhanced fused point cloud model. Complete evaluation: Based on the concrete-reinforced fused point cloud model, the point cloud density, reflection response continuity and pore edge geometric continuity in the fused point cloud are jointly analyzed to identify occlusion and missing areas, low reflection distortion areas and pore edge fracture areas, and the model integrity evaluation results are formed according to the proportion of each defect area.

6. The concrete microstructure modeling system based on three-dimensional laser scanning according to claim 5, characterized in that, The compensation fusion includes: Reflection statistics: Based on the smoothed and denoised point cloud, neighborhood statistics are performed on the reflection response values ​​of each point cloud sampling point. The average reflection response value and reflection dispersion in the neighborhood of each point cloud sampling point are calculated. Low reflection sampling points are identified based on the reflection distribution characteristics of local areas to form a set of reflection anomaly points. Reflection compensation: Based on the set of reflection anomalies, adaptive reflection compensation is performed on the low-reflection sampling points, and the compensation results are constrained by the neighborhood average reflection response value and the reflection continuity of the local area, generating a cloud of reflection compensation points. Coordinate transformation: Based on the horizontal rotation angle, tilt angle and lifting height corresponding to each scanning posture, construct the corresponding spatial coordinate transformation matrix, and transform the reflection compensation point cloud under different scanning postures to a unified spatial coordinate system; Registration and fusion: After transforming to a unified spatial coordinate system, the overlapping areas of each point cloud are matched, the spatial offset error between adjacent point clouds is calculated, and point cloud registration and fusion are performed according to the principle of minimum error to generate a concrete-reinforced fused point cloud model.

7. The concrete microstructure modeling system based on three-dimensional laser scanning according to claim 6, characterized in that, The complete assessment includes: Density analysis: The concrete-reinforced fusion point cloud model is divided into multiple local evaluation regions. The number of point clouds and the area of ​​each local evaluation region are counted, the point cloud density is calculated, and local evaluation regions with point cloud density lower than the missing density threshold are identified as occlusion missing regions. Reflection analysis: Continuity analysis is performed on the reflection response values ​​of point cloud sampling points in each local evaluation region. The difference in reflection response between adjacent points is calculated, and local evaluation regions with average reflection response below the low reflection threshold or reflection fluctuation exceeding the reflection continuity threshold are identified as low reflection distortion regions. Edge analysis: Based on the spatial continuity of the edge points of the pore region, the adjacency tracking of the pore edge points is performed to calculate the pore edge interruption length and edge closure degree, and the local evaluation area where the pore edge interruption length is greater than the edge breakage threshold or the edge closure degree is lower than the closure threshold is identified as the pore edge breakage zone; Complete evaluation: The occlusion and missing areas, low reflection distortion areas, and broken areas at the edges of pores are summarized, the proportion of each defect area is calculated, and the model integrity evaluation results are formed based on the effective coverage ratio, the reliable reflection ratio, and the edge continuity ratio.

8. The concrete microstructure modeling system based on three-dimensional laser scanning according to claim 1, characterized in that, The microstructure segmentation, reconstruction, and feedback control module includes: Defect Feedback: Receives concrete-reinforced fusion point cloud model and model integrity evaluation results, performs joint analysis on the spatial location, normal direction, and area of ​​occluded and missing areas, low-reflection distortion areas, and pore edge fracture areas. When the model integrity evaluation result is lower than the model integrity threshold, generates corresponding supplementary scanning posture instructions and feeds them back to the scanning posture collaborative acquisition module to control the 3D laser scanner and electronically controlled turntable to perform supplementary scanning on the defect area. Three-phase segmentation: When the model integrity evaluation result is not lower than the model integrity threshold, based on the spatial coordinates, reflection response value and local geometric continuity of the point cloud sampling points in the concrete-enhanced fusion point cloud model, the point cloud is semantically segmented into aggregate, mortar and pore three phases, and the corresponding three-phase point cloud region set is generated, including aggregate point cloud set, mortar point cloud set and pore point cloud set. Parameter reconstruction: Based on the three-phase point cloud region set, the aggregate particle morphology parameters, pore connectivity topology parameters, and mortar-wrapped boundary parameters are extracted, and a parameterized model of the multiphase microstructure of concrete that can be used for numerical calculation is constructed.

9. A concrete microstructure modeling system based on three-dimensional laser scanning according to claim 8, characterized in that, The defect feedback includes: Defect localization: Receive the concrete reinforcement fusion point cloud model and model integrity evaluation results, perform spatial cluster analysis on the occlusion and missing areas, low reflection distortion areas and pore edge fracture areas, extract the center position, regional normal direction and regional area of ​​each defect area, and form a defect area feature set; Attitude generation: Based on the center position, average normal direction and area of ​​each defect region, combined with the current scanning attitude and model integrity evaluation results, the corresponding supplementary scanning attitude command is generated, and the target rotation direction, tilt direction and lifting height of the 3D laser scanner and the electronically controlled turntable are determined. Feedback control: The supplementary scanning posture command is fed back to the scanning posture collaborative acquisition module, which controls the electronically controlled turntable to perform horizontal rotation, tilting and lifting linkage, and simultaneously adjusts the scanning direction and scanning range of the 3D laser scanner, so that the scanning beam preferentially covers the occluded and missing areas, low reflection distortion areas and the broken areas at the edges of pores, thereby completing the supplementary scanning of the defective areas.

10. A concrete microstructure modeling system based on three-dimensional laser scanning according to claim 9, characterized in that, The three-phase split includes: Feature extraction: When the model integrity evaluation result is not lower than the model integrity threshold, based on the spatial coordinates, reflection response value and local geometric continuity of each point cloud sampling point in the concrete-reinforced fusion point cloud model, the corresponding spatial distribution features, reflection features and curvature continuity features are extracted to construct the semantic feature vector of the point cloud sampling points. Semantic classification: Based on the semantic feature vectors corresponding to each point cloud sampling point, perform three-phase semantic classification of aggregate, mortar and pores on the point cloud sampling points, and output the semantic category results of the point cloud sampling points according to the feature probabilities corresponding to each category; Region generation: Based on the semantic classification results of point cloud sampling points, region clustering is performed on point cloud sampling points of the same category and spatially continuous, generating aggregate point cloud sets, mortar point cloud sets and pore point cloud sets respectively.

Images