Stone-laying retaining wall disease data processing system

Abstract

The invention provides a stone-laying retaining wall disease data processing system, and relates to the technical field of data processing, and the system comprises an early warning area recognition module which collects the three-dimensional point cloud data of the surface of a target stone-laying retaining wall, calculates the spatial distance deviation from each point in the three-dimensional point cloud data to a reference surface, and transmits the spatial distance deviation to a database; identifying disease early-warning areas and calculating a deformation index of each disease early-warning area; the disease cooperative detection module adaptively configures an internal nondestructive detection scheme for each disease early warning area according to the deformation index, and drives corresponding nondestructive detection equipment to perform cooperative detection; the data processing module is used for processing the detection data acquired by the nondestructive detection equipment and identifying an internal disease identification result including the type, the scale and the distribution of the internal disease; and the diagnosis report generation module performs correlation analysis on the internal disease identification result and the deformation index, evaluates the comprehensive risk level of the disease, and generates a disease diagnosis report. According to the method, the accuracy and the comprehensiveness of stone-laying retaining wall disease data processing are improved.

Description

A data processing system for defects in masonry retaining walls Technical Field

[0001] This invention relates to the field of data processing, and in particular to a data processing system for defects in masonry retaining walls. Background Technology

[0002] Masonry retaining walls are widely used in infrastructure construction in various fields such as water conservancy, transportation, municipal engineering, and mining. They are important structures for resisting slope instability, intercepting soil and rock masses, and ensuring the structural safety and normal operation of engineering projects. However, most masonry retaining walls have a long service life. Due to limitations in construction technology, natural environmental erosion, and lateral pressure on the soil and rock masses, they are prone to developing hidden defects such as mortar voids. These defects tend to develop gradually over time, directly threatening the structural stability of the masonry retaining wall.

[0003] Scientific processing and thorough analysis of data related to defects in masonry retaining walls are the core prerequisites for accurately identifying defect status and assessing safety risks, and can provide reliable data support for the safe operation and maintenance management of masonry retaining walls.

[0004] However, existing methods for processing masonry retaining wall defects are mostly single-dimensional analyses with fragmented processes, lacking comprehensive integration and deep correlation of multi-source related data. This results in highly subjective, inaccurate, and incomplete defect risk assessment results, making it difficult to meet the refined and precise requirements of modern engineering for structural safety operation and maintenance. Summary of the Invention

[0005] This invention addresses the technical problems of low accuracy, poor data integration, and strong subjectivity in risk assessment of masonry retaining wall disease data processing in existing technologies by providing a masonry retaining wall disease data processing system.

[0006] The technical solution of this invention to solve the above-mentioned technical problems is as follows: This invention provides a data processing system for masonry retaining wall defects, comprising: a warning area identification module, used to collect three-dimensional point cloud data of the surface of the target masonry retaining wall, and by calculating the spatial distance deviation from each point in the three-dimensional point cloud data to the reference plane, identify local areas where the deformation exceeds a preset threshold as defect warning areas, and quantitatively calculate the deformation index of each defect warning area; a defect collaborative detection module, used to adaptively configure an internal non-destructive testing scheme for each defect warning area according to the deformation index, and drive the corresponding non-destructive testing equipment to perform collaborative testing, wherein the non-destructive testing equipment includes an ultrasonic testing device, a ground penetrating radar, and an acoustic emission sensing device; a data processing module, used to process and fuse the detection data collected by the non-destructive testing equipment in the defect warning area, and identify the internal defect identification results including the internal defect type, scale, and distribution; and a diagnostic report generation module, used to perform correlation analysis between the internal defect identification results and the deformation index, evaluate the comprehensive risk level of the defect, and generate a defect diagnostic report.

[0007] The beneficial effects of this invention are as follows: Compared with the prior art, this application first collects three-dimensional point cloud data of the surface of the target masonry retaining wall through the early warning area identification module. By calculating the spatial distance deviation from each point in the three-dimensional point cloud data to the reference plane, it identifies local areas where the deformation exceeds a preset threshold as disease early warning areas, and quantitatively calculates the deformation index of each disease early warning area, realizing the accurate locking of high-risk disease areas and the multi-dimensional quantitative characterization of the overall deformation severity. Secondly, the disease collaborative detection module adaptively configures the internal non-destructive testing scheme for each disease early warning area according to the deformation index, and drives the corresponding non-destructive testing equipment to perform collaborative testing, avoiding the problem of over-testing in low-risk areas and under-testing in high-risk areas, improving the accuracy, comprehensiveness and efficiency of internal disease detection, and simultaneously optimizing the allocation of testing resources. Thirdly, the data processing module processes and integrates the detection data collected by the non-destructive testing equipment in the disease early warning area, identifies the internal disease identification results including the type, scale and distribution of internal diseases, and fully leverages the specialized detection advantages of each non-destructive testing equipment for different types of diseases, providing comprehensive and reliable data support for subsequent disease risk level assessment. Finally, the diagnostic report generation module correlates the internal disease identification results with the deformation index to assess the comprehensive risk level of the disease and generate a disease diagnostic report, thus achieving a scientific and accurate determination of the comprehensive risk level of the disease.

[0008] Through the aforementioned technical solution, this application achieves a closed-loop process for the entire process of masonry retaining wall disease, from surface deformation identification, adaptive collaborative non-destructive testing, accurate internal disease identification, and comprehensive risk assessment, through the coordinated operation of the early warning area identification module, the disease collaborative detection module, the data processing module, and the diagnostic report generation module. It accurately anchors the disease early warning area and quantifies the deformation index using 3D point cloud data. Based on the deformation index, it adaptively configures a collaborative detection scheme using ultrasonic testing devices, ground-penetrating radar, and acoustic emission sensors. Furthermore, it comprehensively identifies the type, scale, and distribution of internal diseases through multi-source detection data fusion. Finally, it combines the internal disease identification results with the deformation index correlation analysis to obtain a scientific and objective comprehensive risk level and a complete disease diagnosis report. This effectively solves the problems of fragmentation, strong subjectivity, and insufficient accuracy in existing masonry retaining wall disease data processing methods, achieving precise disease detection, comprehensive disease identification, accurate risk assessment, and refined operation and maintenance support, providing reliable and comprehensive technical support for the safe operation and maintenance management of masonry retaining walls. Attached Figure Description

[0009] Figure 1 is a structural schematic diagram of a masonry retaining wall disease data processing system provided by the present invention; Figure 2 is a principle schematic diagram of a masonry retaining wall disease data processing system provided by the present invention.

[0010] In the attached diagram, the components represented by each number are as follows: early warning area identification module 11, disease collaborative detection module 12, data processing module 13, and diagnostic report generation module 14. Detailed Implementation

[0011] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0012] In the description of this invention, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the stated features. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0013] In the description of this invention, the term "for example" is used to mean "used as an example, illustration, or description." Any embodiment described as "for example" in this invention is not necessarily to be construed as being more preferred or advantageous than other embodiments. The following description is provided to enable any person skilled in the art to make and use the invention. Details are set forth in the following description for purposes of explanation. It should be understood that those skilled in the art will recognize that the invention can be made without using these specific details. In other instances, well-known structures and processes will not be described in detail to avoid obscuring the description of the invention with unnecessary detail. Therefore, the invention is not intended to be limited to the embodiments shown, but is consistent with the broadest scope of the principles and features disclosed herein.

[0014] As shown in Figures 1 and 2, this embodiment of the invention provides a data processing system for masonry retaining wall defects, including: a warning area identification module 11, used to collect three-dimensional point cloud data of the surface of the target masonry retaining wall, and by calculating the spatial distance deviation of each point in the three-dimensional point cloud data from the reference surface, identify local areas where the deformation exceeds a preset threshold as defect warning areas, and quantify the deformation index of each defect warning area.

[0015] Traditional methods for detecting defects in masonry retaining walls often rely on manual knocking and cross-sectional demolition, which are subjective, destructive, and inefficient, and make it difficult to fully capture minute surface deformations and regional deformation characteristics.

[0016] Meanwhile, internal defects in masonry retaining walls directly undermine the structural integrity and mechanical stability, leading to an imbalance in stress distribution. This results in observable abnormal deformation on the surface of the retaining wall. Generally, the more severe the internal defects, the more significant the weakening of the structure's load-bearing capacity, and the wider and more concentrated the deformation coverage on the surface of the retaining wall. Conversely, abnormal deformation on the surface of a masonry retaining wall is essentially a macroscopic external manifestation of internal defects, indicating high-risk areas with potential internal defects.

[0017] To address the aforementioned issues, this application uses a warning area identification module 11 to collect three-dimensional point cloud data of the target masonry retaining wall surface. By calculating the spatial distance deviation from each point in the three-dimensional point cloud data to the reference surface, it identifies local areas where the deformation exceeds a preset threshold as disease warning areas and quantifies the deformation index of each disease warning area.

[0018] Specifically, the warning area identification module 11 is used to: collect the current three-dimensional point cloud data of the target masonry retaining wall surface through a three-dimensional laser scanning device; obtain a reference surface that matches the original design surface or historical health status surface of the retaining wall; calculate the spatial Euclidean distance from each point in the three-dimensional point cloud data to the reference surface as a spatial distance deviation value; classify all points whose spatial distance deviation value exceeds a preset deformation threshold as abnormal points, and identify the continuous area formed by spatially adjacent abnormal points as a disease warning area.

[0019] In this embodiment, the current three-dimensional point cloud data of the target masonry retaining wall surface is first acquired using a three-dimensional laser scanning device. The three-dimensional point cloud data refers to a dataset composed of a large number of three-dimensional spatial coordinate points, each point containing the X, Y, and Z three-dimensional coordinate information of its corresponding position on the target masonry retaining wall surface. The current three-dimensional point cloud data reflects the actual geometric shape of the target masonry retaining wall surface at the time of detection, and can completely reproduce the physical features of the target masonry retaining wall surface, such as unevenness and damage. Specifically, during the acquisition of the three-dimensional point cloud data, the scanning path needs to be planned according to the length, height, and complexity of the target masonry retaining wall to ensure that the scan covers the entire surface of the target masonry retaining wall and avoids scanning blind spots; at the same time, the scanning accuracy is controlled, for example, the scanning point spacing can be set to 5mm to ensure that minute deformation features can be captured.

[0020] Secondly, a reference surface is obtained that matches the original design surface or historical health status surface of the retaining wall. The reference surface refers to the surface geometry corresponding to the normal state of the target masonry retaining wall, free from deformation and defects. This is specifically constructed by matching the original design surface or historical health status surface of the target masonry retaining wall. The original design surface refers to the design geometry surface used during the construction phase of the retaining wall. Its parameters can be obtained by retrieving the original design drawings and construction technical documents of the target masonry retaining wall, reflecting its ideal state before service. The historical health status surface refers to the surface geometry surface of the retaining wall in a healthy state, confirmed by testing during service to be free from defects and significant deformation. Its data can be obtained by retrieving historical 3D laser scan data of the retaining wall and health status detection data from maintenance records, accurately reflecting the normal state of the target masonry retaining wall during its service period.

[0021] Next, the spatial Euclidean distance from each point in the 3D point cloud data to the reference plane is calculated as the spatial distance deviation value. The spatial Euclidean distance refers to the vertical distance from each point in the 3D point cloud data to the reference plane. For example, a perpendicular line can be drawn from each point in the 3D point cloud data to the reference plane; the intersection of this perpendicular line and the reference plane is the corresponding projection point of each point in the 3D point cloud data on the reference plane. Calculating the straight-line distance between these two points yields the spatial Euclidean distance. This is because the reference plane represents the ideal geometric shape of the retaining wall in its original design or historical healthy state. In the ideal case without any deformation, the collected 3D point cloud data should perfectly fit the reference plane, meaning the spatial distance deviation value of each point should be zero. However, when deformation exists, the calculated spatial Euclidean distance directly quantifies the offset of the point cloud position relative to its healthy state; this offset is the spatial distance deviation value. By calculating the spatial distance deviation value, a quantitative comparison can be achieved between the current state of the target masonry retaining wall and its undeformed or normal state.

[0022] For example, the straight-line distance from each point cloud in the three-dimensional point cloud data to the reference plane can be solved one by one by spatial geometric calculation methods to obtain several spatial Euclidean distances. The positive and negative values ​​of the spatial Euclidean distances can reflect the direction of deformation. For example, the positive direction is outward convexity and the negative direction is inward concavity. The absolute value of the spatial Euclidean distances can reflect the degree of deformation.

[0023] Finally, all points whose spatial distance deviation exceeds a preset deformation threshold are classified as anomalies, and the continuous area formed by spatially adjacent anomalies is identified as a disease warning area. The preset deformation threshold is a dynamically determined maximum allowable deformation based on the material properties, service life, and engineering safety level of the target masonry retaining wall, used to define the boundary between normal and abnormal deformation in the 3D point cloud data. For example, the preset deformation threshold can be determined through indoor testing and calibration based on engineering experience. For instance, for a target masonry retaining wall with a service life of over 30 years, the preset deformation threshold can be set to 5 mm; for a target masonry retaining wall with a service life of less than 5 years, the preset deformation threshold can be set to 2 mm.

[0024] Specifically, the process begins by filtering out all points in the 3D point cloud data whose absolute spatial distance deviation exceeds a preset deformation threshold as outliers. These outliers indicate that the surface of the target masonry retaining wall has exceeded the safe deformation range, posing a potential risk of damage. Then, a spatial connectivity analysis algorithm is used to determine the spatial correlation of all outliers. Outliers meeting the spatial adjacency condition are aggregated into continuous spatial regions, serving as early warning areas for accurately locating the abnormal deformation range of the retaining wall surface. This provides a clear target area for subsequent targeted internal non-destructive testing. The spatial adjacency condition can be set to a distance between two points whose Euclidean distance is less than a preset neighborhood threshold. For example, the preset neighborhood threshold can be set to 1.5 times the scanning point spacing of a 3D laser scanning device.

[0025] In summary, this application fully considers the strong positive correlation between surface deformation and internal defects of the target masonry retaining wall: the development of internal defects such as mortar voids, loose masonry, and internal cavities will damage the integrity and stability of the retaining wall structure, thereby directly leading to abnormal deformation of the retaining wall surface that exceeds the safe range; at the same time, surface deformation, as a macroscopic external characteristic, can also inversely indicate the potential distribution and severity of internal defects. Thus, by deeply linking and integrating macroscopic surface deformation characteristics with microscopic internal defect information, the limitations of merely looking at the surface or isolating internal inspections are broken, achieving cross-dimensional collaborative analysis from surface phenomena to internal essence, providing rigorous logical support for subsequent accurate and comprehensive determination of the overall risk level of defects.

[0026] Furthermore, the step of "obtaining a reference surface that matches the original design surface or historical health status surface of the retaining wall" includes: obtaining the original design three-dimensional model or the three-dimensional point cloud data of the target masonry retaining wall under historical health status as reference surface data; using a region growing algorithm to segment the reference surface data into multiple surface segments based on curvature features; for each surface segment, using the RANSAC algorithm to perform planar or quadratic surface fitting; and integrating all the fitted planes or surfaces to form a reference surface that fits the original design or historical health geometry of the retaining wall.

[0027] In this embodiment, the original design 3D model or historical health status 3D point cloud data of the target masonry retaining wall is first acquired as reference surface data. The reference surface data refers to the original reference data used to construct the benchmark surface, including two types: first, the original design 3D model of the target masonry retaining wall, i.e., the digital model provided by the design unit before the retaining wall construction, such as a CAD model or BIM model, containing detailed information such as the design dimensions, geometric shape, and masonry structure of the retaining wall; second, the historical health status 3D point cloud data, i.e., the point cloud data obtained by a 3D laser scanning device when the retaining wall has been inspected and confirmed to be free of defects and significant deformation during its historical service life, which can truly reflect the actual surface morphology of the retaining wall in its healthy state.

[0028] For example, the original design 3D model of the target masonry retaining wall can be obtained by retrieving the construction archives, design drawings or BIM database of the target masonry retaining wall; the 3D point cloud data of the historical health status can be extracted from the operation and maintenance inspection database of the target masonry retaining wall, but it is necessary to ensure that the extracted 3D point cloud data has no known defects, and that the accuracy of the inspection equipment is consistent with the current scanning equipment, so as to avoid the accuracy of the reference plane due to data deviation.

[0029] Secondly, a region growing algorithm is used to segment the reference surface data into multiple surface segments based on curvature characteristics. The region growing algorithm is a segmentation algorithm based on the similarity of local features of the data. Its core idea is to start from a seed point and gradually incorporate adjacent data points with similar features to the seed point into the same region, ultimately forming continuous surface segments. Curvature characteristics refer to the degree of curvature parameters of each point in the reference surface data, which can reflect the geometric changes of the surface. For example, the curvature of a planar region is close to 0, while the curvature of a corner or arc-shaped region is larger.

[0030] Specifically, firstly, a point with stable curvature characteristics is selected from the reference surface data as a seed point. A curvature similarity threshold is set, such as 0.05. Points with curvature differences from the seed point less than the curvature similarity threshold and spatially adjacent are included in the same region, gradually growing to form multiple independent surface segments. Through this segmentation process, regions with different geometric shapes, such as the planar main body of the retaining wall, corner curved surfaces, and top arcs, can be separated to ensure that the subsequent fitting process can accurately match the actual shape of each local region.

[0031] In this way, the complex reference surface data is decomposed into multiple simple local regions, which facilitates accurate fitting in the future.

[0032] Next, for each surface segment, the RANSAC algorithm is used to fit either a plane or a quadratic surface. The RANSAC algorithm (Random Sample Consensus Algorithm) is an iterative algorithm used to extract mathematical model parameters from noisy data. It has advantages such as strong noise resistance and high fitting accuracy, and is suitable for fitting planes or curved surfaces from point cloud data. Plane fitting is suitable for regions where the surface segment is flat, such as the main wall surface or top plane of a retaining wall; quadratic surface fitting is suitable for regions where the surface segment is arc-shaped or curved, such as the corners or arc-shaped tops of a retaining wall.

[0033] Specifically, for each segmented surface fragment, its geometric type is first determined: if the curvature of the surface fragment is close to 0 and the spatial distribution is flat, the RANSAC algorithm is used to fit a planar model, and the equation parameters of the plane are determined through iterative calculation, such as Ax+By+Cz+D=0; if the surface fragment has obvious bending characteristics, such as curvature greater than a preset threshold of 0.1, the RANSAC algorithm is used to fit a quadratic surface model, such as a sphere, cylinder, or general quadratic surface, and the least squares error between the model and the point cloud data of the surface fragment is minimized by optimizing the surface parameters; during the fitting process, the number of iterations, such as 1000, and the interior point threshold, such as 2mm, need to be set to ensure the stability and accuracy of the fitting results.

[0034] Finally, all fitted planes or curved surfaces are integrated to form a reference surface that conforms to the original design or historical healthy geometry of the retaining wall. Specifically, firstly, the spatial positions of all fitted planes or curved surfaces are calibrated to ensure that the relative positions between each plane or curved surface are consistent with the original positions in the reference surface data; then, a surface stitching algorithm is used to smooth the transition between adjacent plane or curved surface models, eliminating stitching gaps and sharp corners between planes or curved surfaces, ensuring the continuity and integrity of the reference surface; finally, the integrated plane or curved surface is compared and verified with the original design 3D model of the retaining wall or the historical healthy point cloud data, and the fitting error is calculated. If the error exceeds the allowable error range (e.g., the allowable error range is set to 3mm), the previous step is returned to adjust the fitting parameters until the accuracy requirements are met; the final reference surface can completely restore the geometry of the retaining wall in its original design or historical healthy state, providing an accurate and reliable reference standard for subsequent deformation comparison with the current point cloud data.

[0035] Specifically, the early warning area identification module 11 is further configured to: extract the spatial distance deviation values ​​of all abnormal points contained within each disease early warning area, and calculate the average value of the spatial distance deviations of all abnormal points as the overall average spatial distance deviation; calculate the proportion of the number of abnormal points to the total number of points in the disease early warning area as a deformation index; calculate the joint deformation concentration coefficient and the stonework unit motion dispersion based on the spatial distance deviation values ​​and the overall average spatial distance deviation; and normalize and weightedly fuse the deformation index, the joint deformation concentration coefficient, and the stonework unit motion dispersion to obtain a deformation index characterizing the overall deformation severity of the disease early warning area.

[0036] In this embodiment, firstly, for each disease warning area, the spatial distance deviation values ​​of all abnormal points contained within it are extracted, and the average value of the spatial distance deviations of all abnormal points is calculated as the overall average spatial distance deviation. Specifically, the overall average spatial distance deviation can reflect the overall average degree of deformation within the disease warning area. The calculation requires first extracting the spatial distance deviation values ​​of all abnormal points within the disease warning area, and then calculating the overall average spatial distance deviation using the arithmetic mean formula.

[0037] For example, if a disease warning area contains 100 abnormal points and the sum of the spatial distance deviations of the 100 abnormal points is 300mm, then the overall average spatial distance deviation = 300 / 100 = 3mm.

[0038] Secondly, the ratio of the number of abnormal points to the total number of points in the disease warning area is calculated as a deformation index. The deformation index reflects the proportion of deformed areas within the disease warning area; a higher ratio indicates a wider deformation coverage and a broader overall deformation scope. Specifically, the total number of points in the disease warning area refers to the total number of point clouds within that area, including both abnormal points and normal points that do not exceed a preset deformation threshold. The deformation index is obtained by calculating the ratio of the number of abnormal points to the total number of points in the disease warning area. The deformation index ranges from 0 to 1; a larger value indicates a wider deformation coverage.

[0039] Thus, this application first accurately identifies abnormal points exceeding a preset deformation threshold by comparing the current 3D point cloud data of the target masonry retaining wall surface with a reference surface that conforms to the original design or historical healthy geometric shape of the retaining wall. Then, through spatial connectivity analysis, multiple independent disease warning areas are aggregated. Based on this, for each disease warning area, the total number of abnormal points within it is counted, and the proportion of this number to the total number of points in the disease warning area is calculated. This proportion is used as a deformation index to characterize the deformation coverage. This replaces the traditional method of quantifying deformation using deformation deviation amplitude, such as local deformation deviation amplitude of 1mm, 2mm, etc. This traditional method can only reflect the deformation degree of a single point or local area and cannot characterize the overall deformation coverage. This application achieves accurate quantitative characterization of the overall deformation coverage of the disease warning area through a deformation index in the form of a quantity proportion, providing a more comprehensive and targeted reference for subsequent comprehensive assessment of the deformation severity.

[0040] Secondly, based on the spatial distance deviation value and the overall average spatial distance deviation, the deformation concentration coefficient of the masonry joint and the motion dispersion of the masonry unit are calculated. This is because the masonry retaining wall is mainly composed of masonry and mortar, and the two have significant differences in material properties such as strength and elastic modulus. Moreover, the masonry joint is a weak point in the structural connection, and it often shows deformation or defects first under stress and environmental erosion. Therefore, it is necessary to focus on the deformation characteristics of the masonry joint and the masonry unit to calculate the deformation concentration coefficient of the masonry joint and the motion dispersion of the masonry unit.

[0041] Among them, the joint deformation concentration coefficient is a concentration index specifically characterizing the degree of deformation in the joint area relative to the overall deformation degree in the disease warning area. If the joint deformation concentration coefficient is greater than 1, it indicates that the deformation is concentrated in the joint area, and the risk of joint disease is higher. The stonework unit movement dispersion is a quantitative index reflecting the deformation difference between independent stonework units in the disease warning area. The higher the stonework unit movement dispersion, the stronger the uneven deformation of each stonework unit, and the more obvious the overall damage to the retaining wall structure. Both reflect the distribution pattern and concentration of deformation in the disease warning area, such as whether it is concentrated in the joint and the size of the deformation difference between each unit, providing key local characteristic basis for subsequent comprehensive assessment of the severity of deformation.

[0042] Finally, the deformation index, the joint deformation concentration coefficient, and the stonework unit motion dispersion are normalized and weighted to obtain a deformation index characterizing the overall deformation severity of the disease warning area. Specifically, the deformation index, joint deformation concentration coefficient, and stonework unit motion dispersion are first normalized, mapping all three index values ​​to the 0-1 interval to eliminate the influence of dimensions; then, the deformation index is calculated by weighted summation using preset weighting coefficients.

[0043] Normalization can be achieved using min-max normalization, i.e., normalization is performed using the following formula: x′=(x max -x min ) / (xx) min In the formula, x′ is the normalized deformation index or joint deformation concentration coefficient or masonry unit motion dispersion, and x is the original deformation index or joint deformation concentration coefficient or masonry unit motion dispersion. min x is the minimum value of the corresponding indicator in the sample set. max This represents the maximum value of the corresponding indicator in the sample set. The weighting coefficients of the weighted fusion can be dynamically set according to the influence of the deformation index, the joint deformation concentration coefficient, and the motion dispersion of the masonry unit on the severity of deformation. For example, the weight of the deformation index can be set to 0.4, the weight of the joint deformation concentration coefficient to 0.3, and the weight of the motion dispersion of the masonry unit to 0.3.

[0044] For example, if the normalized deformation index is 0.6, the joint deformation concentration coefficient is 0.5, and the stonework unit movement dispersion is 0.4, and the weighting coefficients for the three are 0.4, 0.3, and 0.3 respectively, then the deformation index = 0.6 × 0.4 + 0.5 × 0.3 + 0.4 × 0.3 = 0.51. The deformation index comprehensively considers the deformation index, the joint deformation concentration coefficient, and the stonework unit movement dispersion. It can comprehensively reflect the overall deformation severity of the disease warning area from three dimensions: deformation coverage, deformation concentration characteristics, and deformation uniformity. This avoids the one-sidedness of single-index assessment and makes the characterization of deformation severity more comprehensive and scientific.

[0045] Further, the step of "calculating the joint deformation concentration coefficient and the stonework unit motion dispersion based on the spatial distance deviation value and the overall average spatial distance deviation" includes: calculating the normal vector of each point in each of the disease warning areas in the three-dimensional point cloud data; detecting potential linear groove features based on abrupt changes in the normal vector direction or local maxima of curvature; and, combined with prior knowledge of the layered masonry of the stone retaining wall, selecting linear features whose orientation and the angle between the horizontal direction and the gravity direction are within a preset range as suspected joints; calculating the average spatial distance deviation within the suspected joint distribution area and dividing the average spatial distance deviation by the overall average spatial distance deviation to obtain the joint deformation concentration coefficient; segmenting point cloud clusters with continuous surfaces based on a region growing algorithm; and, according to the typical size range of the stones in the stone retaining wall, performing secondary segmentation on point cloud clusters whose area or volume exceeds the reasonable range of a single stone to obtain multiple unit blocks corresponding to the physical stonework; calculating the average spatial distance deviation of each unit block and calculating the ratio of the standard deviation to the average value of the average spatial distance deviation of all unit blocks as the stonework unit motion dispersion.

[0046] In this embodiment, the mortar joints in a masonry retaining wall are the junctions between the masonry and mortar. Due to differences in material properties and limitations in construction techniques, mortar joints are naturally weak points in the structure, and various defects often preferentially originate and develop in them. Therefore, the degree of deformation in the mortar joint area is a sensitive indicator for assessing the overall stability of the retaining wall. However, in actual engineering, the retaining wall surface often has interfering features such as scratches, local damage, or debris attachment. In the 3D point cloud, the real mortar joints only appear as narrow linear grooves, which are easily confused with these interfering features in terms of morphology. It is difficult to reliably distinguish between interference and real mortar joints by visual observation or simple geometric filtering alone. Therefore, this application first calculates the normal vector of each point in each defect warning area in the 3D point cloud data. Based on the abrupt change in the normal vector direction or the local maximum of curvature, potential linear groove features are detected. Combined with prior knowledge of the layered masonry construction of the masonry retaining wall, linear features whose orientation and the angle between the horizontal direction and the gravity direction are within a preset range are selected as suspected mortar joints.

[0047] Among them, the normal vector is a vector perpendicular to the surface of the point cloud, and its direction change usually corresponds to the change of the surface profile, such as the position of the masonry joint; the a priori knowledge of layered masonry refers to the construction process of masonry retaining walls, which usually adopts horizontal layering or vertical masonry, and the direction of the masonry joint is mostly consistent with the horizontal direction or the direction of gravity.

[0048] For example, the normal vector and curvature of each point in the point cloud are first calculated. The normal vector can be estimated using Principal Component Analysis (PCA). For instance, a covariance matrix is ​​constructed by taking the set of neighboring points with a radius of 5 mm for each point, and the eigenvector corresponding to the smallest eigenvalue is the normal vector. Then, dual feature detection is performed: on the one hand, abrupt changes in the normal vector direction are identified by calculating the angle between the normal vectors of adjacent points, and points exceeding a preset angle threshold are classified as edge points. Such abrupt changes often correspond to the intersection edges of the stone surfaces on both sides of the grout joint. On the other hand, local curvature maxima are identified by using a sliding window to calculate curvature and filtering out local extreme points exceeding a preset curvature threshold. These points often correspond to the bottom of the groove. Finally, a spatial clustering algorithm is used to aggregate the two types of spatially continuously distributed feature points to form a series of coherent latent linear groove features.

[0049] For example, further refinement of suspected joints is achieved by incorporating prior knowledge of layered masonry construction: the groove features obtained from the initial screening contain real joints along with interference from scratches, damage, etc., and need to be purified based on structural knowledge. The construction of masonry retaining walls follows the principle of layered masonry, meaning that stones are typically laid in layers horizontally or aligned vertically. This strictly limits the macroscopic orientation of the joints, which is basically consistent with the horizontal or vertical direction. Based on this, by calculating the average orientation vector of each potential linear groove feature and analyzing its angle with the horizontal and vertical directions, only groove features with an angle within a preset tolerance range, such as ±10°, are retained and identified as suspected joints; features with arbitrary orientations are discarded as noise. In this way, by leveraging knowledge from the field of image processing to closely integrate point cloud data with engineering knowledge, robust and automated identification of structural joints in masonry retaining walls is achieved.

[0050] Secondly, the average spatial distance deviation within the suspected joint distribution area is calculated, and then divided by the overall average spatial distance deviation to obtain the joint deformation concentration coefficient. Specifically, the spatial distance deviation values ​​of all points within the suspected joint distribution area are first extracted, and the arithmetic mean is calculated to obtain the average spatial distance deviation; then, the average spatial distance deviation is divided by the overall average spatial distance deviation to obtain the joint deformation concentration coefficient. If the joint deformation concentration coefficient is greater than 1, it indicates that the deformation degree of the joint area is higher than the overall deformation degree, and the deformation is concentrated at the joint; if the joint deformation concentration coefficient is less than 1, it indicates that the deformation degree of the joint area is lower than the overall deformation degree.

[0051] For example, if the calculated average spatial distance deviation within the suspected masonry joint distribution area is 4mm, and the overall average spatial distance deviation is 3mm, the masonry joint deformation concentration coefficient = 4 / 3 ≈ 1.33, indicating that the deformation degree in the masonry joint area is higher than the overall deformation degree, and the deformation is concentrated at the masonry joint.

[0052] Next, point cloud clusters with continuous surfaces are segmented using a region growing algorithm. Then, based on the typical size range of the stones in the masonry retaining wall, point cloud clusters whose area or volume exceeds the reasonable range of a single stone are further segmented to obtain multiple unit blocks corresponding to the physical masonry. Here, a unit block refers to a point cloud cluster corresponding to a single stone in the actual masonry retaining wall. Specifically, the region growing algorithm is used to segment initial point cloud clusters based on the spatial adjacency and surface smoothness of the point clouds. Then, combined with the typical dimensions of the stones in the masonry retaining wall, such as 30-50cm in length, 20-30cm in width, and 20-30cm in height, the initial point cloud clusters are filtered. If the volume of a point cloud cluster exceeds the reasonable range of a single stone volume, such as greater than 0.06m... 3 Then, it is divided into two parts to obtain unit blocks that correspond one-to-one with the actual physical masonry.

[0053] Finally, the average spatial distance deviation of each unit block is calculated, and the ratio of the standard deviation to the mean of the average spatial distance deviation of all unit blocks is calculated as the motion dispersion of the masonry unit. Specifically, the average spatial distance deviation of all points within each unit block is first calculated to obtain the average spatial distance deviation of each unit block; then, the standard deviation and mean of the average spatial distance deviation of all unit blocks are calculated, and the motion dispersion of the masonry unit is obtained by calculating the ratio of the standard deviation to the mean. The larger the motion dispersion of the masonry unit, the greater the difference in deformation among the masonry units, and the worse the stability of the retaining wall structure.

[0054] For example, if the average spatial distance deviation of all unit blocks is calculated to be 3 mm and the standard deviation is 1.2 mm, then the motion dispersion of the masonry unit is 1.2 / 3 = 0.4.

[0055] In summary, compared to existing technologies, this application uses the warning area identification module 11 to collect three-dimensional point cloud data of the target masonry retaining wall surface. By calculating the spatial distance deviation from each point in the three-dimensional point cloud data to the reference plane, it identifies local areas where deformation exceeds a preset threshold as disease warning areas and quantifies the deformation index of each disease warning area. This achieves precise locking of high-risk disease areas and multi-dimensional quantitative characterization of the overall deformation severity, providing precise targeting and reliable data support for subsequent adaptive configuration of internal non-destructive testing schemes and correlation assessment of internal disease risks.

[0056] The disease collaborative detection module 12 is used to adaptively configure an internal non-destructive testing scheme for each of the disease early warning areas according to the deformation index, and drive the corresponding non-destructive testing equipment to perform collaborative testing. The non-destructive testing equipment includes an ultrasonic testing device, a ground penetrating radar, and an acoustic emission sensing device.

[0057] Traditional non-destructive testing of the interior of masonry retaining walls often adopts a single device full-coverage scanning mode, which lacks differentiated consideration of regional risk levels. This not only results in insufficient testing targeting and low efficiency, but also easily leads to serious waste of equipment resources: for example, low-risk areas may increase the amount of ineffective work due to excessive testing, while high-risk areas may have missed detection risks due to insufficient testing depth or accuracy caused by the limited testing capabilities of a single device.

[0058] The disease warning area obtained by the aforementioned warning area identification module 11 has accurately locked the high-risk target range, and the deformation index can quantify the severity of deformation in the disease warning area. Based on this, the traditional full-coverage detection mode can be abandoned, and targeted non-destructive testing can be carried out in the disease warning area. The internal non-destructive testing scheme can be dynamically matched according to the deformation index.

[0059] Meanwhile, the ultrasonic testing device, through the analysis of sound wave propagation characteristics, excels at detecting shallow defects such as mortar voids and is suitable for basic inspection in areas with slight deformation; the ground-penetrating radar, using the principle of electromagnetic pulse reflection, can accurately identify volumetric defects such as internal cavities and is suitable for deep inspection in areas with moderate deformation; the acoustic emission sensor can capture elastic wave signals generated by loose masonry in real time and is suitable for dynamic defect monitoring in areas with severe deformation. The three devices are used in a tiered manner according to the risk level corresponding to the deformation index, which maximizes the specialized advantages of each device and optimizes the allocation of testing resources, thereby simultaneously improving testing efficiency and accuracy.

[0060] To address the aforementioned issues, this application uses a disease collaborative detection module 12 to adaptively configure internal non-destructive testing schemes for each disease early warning area based on the deformation index, and drives corresponding non-destructive testing equipment to perform collaborative testing. The non-destructive testing equipment includes an ultrasonic testing device, a ground-penetrating radar, and an acoustic emission sensor.

[0061] Specifically, the disease collaborative detection module 12 is used to: establish a mapping table between deformation index ranges and non-destructive testing schemes, wherein the non-destructive testing schemes define the combination of non-destructive testing equipment to be called and the corresponding testing parameters; query the mapping table according to the deformation index of the disease warning area to determine the target non-destructive testing scheme corresponding to the disease warning area; and, according to the target non-destructive testing scheme, call one or more of the corresponding ultrasonic testing device, ground penetrating radar, and acoustic emission sensing device, and configure the corresponding testing parameters to perform collaborative detection within the disease warning area.

[0062] In this embodiment, a mapping table is first established between the deformation index range and the non-destructive testing (NDT) scheme. The NDT scheme defines the combination of NDT equipment to be called and the corresponding testing parameters. The mapping table can be established based on experimental data and engineering experience. The testing parameters include at least the probe frequency of the ultrasonic testing device, the transmission frequency of the ground penetrating radar, and the sampling frequency of the acoustic emission sensor.

[0063] For example, the mapping table between deformation index ranges and non-destructive testing (NDT) schemes can first divide the deformation index ranges, such as 0-0.3 for mild deformation, 0.3-0.7 for moderate deformation, and 0.7-1.0 for severe deformation. Then, a corresponding NDT scheme is matched for each deformation index range. For example, the NDT scheme corresponding to the mild deformation range is an ultrasonic testing device with a 2.5MHz probe frequency; the NDT scheme corresponding to the moderate deformation range is an ultrasonic testing device with a 2.5MHz ultrasonic probe frequency and a ground-penetrating radar with a 100MHz radar transmission frequency; and the NDT scheme corresponding to the severe deformation range is an ultrasonic testing device with a 2.5MHz ultrasonic probe frequency, a ground-penetrating radar with a 100MHz radar transmission frequency, and an acoustic emission sensing device with a 1MHz acoustic emission sampling frequency.

[0064] Secondly, based on the deformation index lookup mapping table for disease warning areas, the corresponding target non-destructive testing (NDT) schemes for each disease warning area are determined. Specifically, the deformation index of each disease warning area is first obtained, and then the corresponding NDT scheme is found in the mapping table according to the range to which the deformation index belongs, serving as the target NDT scheme. For example, if the deformation index of a certain disease warning area is 0.5, which falls within the moderate deformation range, the corresponding target NDT scheme obtained from the mapping table would be an ultrasonic testing device with a 2.5MHz ultrasonic probe frequency or a ground-penetrating radar with a 100MHz radar transmission frequency.

[0065] Finally, according to the target non-destructive testing (NDT) plan, one or more of the following devices—ultrasonic testing equipment, ground-penetrating radar, and acoustic emission sensors—are invoked and configured with corresponding testing parameters for collaborative testing within the disease warning area. Specifically, the corresponding NDT equipment combination is automatically invoked and configured with the appropriate testing parameters based on the target NDT plan. During the testing process, each NDT device collects data within the disease warning area according to the preset testing path and parameters, ensuring that the collected data covers the entire warning area without redundancy.

[0066] For example, when using an ultrasonic detection device with a 2.5MHz ultrasonic probe frequency and a ground-penetrating radar with a 100MHz radar transmission frequency for coordinated detection, the ultrasonic detection device scans along a grid-like path in the disease warning area, and the ground-penetrating radar scans synchronously along the same path, achieving spatial synchronous data acquisition.

[0067] In summary, compared to existing technologies, this application, through the disease collaborative detection module 12, adaptively configures internal non-destructive testing (NDT) schemes for each disease early warning area based on the deformation index, and drives corresponding NDT equipment to perform collaborative testing. The NDT equipment includes an ultrasonic testing device, ground-penetrating radar, and an acoustic emission sensor. Thus, based on the deformation index, differentiated adaptive configuration of NDT schemes for disease early warning areas is achieved. Through collaborative testing using ultrasonic testing devices, ground-penetrating radar, and acoustic emission sensors, the specialized testing advantages of each device are fully utilized. This avoids over-testing in low-risk areas and under-testing in high-risk areas, while improving the accuracy, comprehensiveness, and efficiency of internal disease detection, and optimizing the allocation of testing resources.

[0068] Data processing module 13 is used to process and integrate the detection data collected by the non-destructive testing equipment in the disease warning area, and identify the internal disease identification results, including the internal disease type, scale and distribution.

[0069] Traditional methods for detecting internal defects in masonry retaining walls often rely on isolated data processing from single non-destructive testing (NDT) devices, resulting in a limited perspective and incomplete capture of defect characteristics. The aforementioned module has accurately located defect warning areas through surface deformation analysis. By systematically processing specialized testing data collected from multimodal NDT devices and integrating the testing advantages of each device, it overcomes the limitations of single devices in identifying different types of defects, achieving a comprehensive and accurate characterization of internal defects.

[0070] Due to defects such as mortar depletion, internal voids, and loose stones within masonry retaining walls, the local dielectric properties of the masonry, including density, dielectric constant, and elastic wave propagation velocity, are altered. This results in differentiated characteristics in the propagation of detection signals emitted by various non-destructive testing (NDT) devices, such as increased sound wave attenuation, strong electromagnetic pulse reflection, and increased elastic wave event frequency. Furthermore, different types of defects correspond to unique signal characteristics. The influence range and intensity of the detection data can reflect the scale of the defect, and the location of the signal source can pinpoint the defect distribution. By extracting, analyzing, and fusing this data, the type, scale, and distribution of internal defects can be accurately deduced, yielding complete and reliable internal defect identification results.

[0071] To address the aforementioned issues, this application uses data processing module 13 to process and integrate the detection data collected by the non-destructive testing equipment within the disease warning area, thereby identifying internal disease identification results, including the type, scale, and distribution of internal diseases.

[0072] Specifically, the data processing module 13 is used to: filter and extract waveform features from the waveform data collected by the ultrasonic testing device to identify areas where the sound wave velocity is lower than a preset wave velocity threshold as suspected areas of mortar void defects; filter, adjust gain, and perform offset imaging processing on the radar profile data collected by the ground penetrating radar to identify in-phase axes with strong reflection amplitude or reflection features exhibiting a hyperbolic shape in the radar image as suspected areas of internal void defects; perform event detection, source localization, and event parameter analysis on the continuous time-domain signals collected by the acoustic emission sensing device to identify spatially concentrated areas where the number of events exceeds a preset frequency threshold or the event energy exceeds a preset energy threshold within a unit time as indication areas of masonry loosening defects; and register and fuse the suspected areas of mortar void defects, the suspected areas of internal void defects, and the indication areas of masonry loosening defects in a unified spatial coordinate system to generate and output internal defect identification results including the type, scale, and distribution of internal defects.

[0073] In this embodiment, the waveform data collected by the ultrasonic testing device is first filtered and its features are extracted to identify areas where the sound wave velocity is lower than a preset threshold, which are then identified as suspected areas of mortar voids. The waveform data collected by the ultrasonic testing device refers to the voltage change over time generated by the ultrasonic signal propagating inside the target masonry retaining wall; waveform features include wave velocity, amplitude, and frequency.

[0074] Secondly, the radar profile data acquired by ground-penetrating radar is filtered, gain adjusted, and subjected to offset imaging processing to identify in-phase axes with strong reflection amplitudes or hyperbolic reflection features in the radar images as suspected areas of internal cavity defects. This is because the cavities inside the masonry retaining wall are filled with air, and the dielectric constant of air is approximately 1, much lower than that of masonry (6-8) and mortar (5-7). When an electromagnetic pulse propagates to the cavity interface, strong reflection occurs, and the boundary contour of the cavity causes the reflected signal to exhibit a specific shape: the reflected signal of a planar cavity forms an in-phase axis with strong reflection amplitude, i.e., a continuous line connecting signal points with the same phase; spherical or irregularly shaped cavities, due to electromagnetic pulses reflecting from different angles, form typical hyperbolic reflection features. These two types of features are the key basis for distinguishing cavities from other defects.

[0075] For example, since radar profile data is susceptible to environmental electromagnetic interference and equipment noise during acquisition, resulting in signal contamination, it is necessary to filter the radar profile data acquired by ground-penetrating radar (GPR). For instance, a combination of frequency domain filtering and time domain filtering can be used. Frequency domain filtering converts the time-domain radar signal to the frequency domain using Fourier transform, eliminating high-frequency interference and low-frequency noise unrelated to the GPR's transmission frequency, such as noise above 200MHz and below 10MHz, while retaining the effective signal within the target frequency range. Time domain filtering uses a moving average filtering method, averaging the values ​​of neighboring points around each signal point to smooth signal fluctuations and eliminate instantaneous impulse noise. Finally, clear radar profile data after noise reduction is obtained. In this way, noise interference is eliminated, retaining only clean and effective reflected signals.

[0076] For example, due to energy attenuation of the electromagnetic pulse propagating inside the retaining wall, gain adjustment processing is required on the filtered radar profile data to ensure clear identification of reflected signals in deep areas. For instance, an automatic gain control (AGC) algorithm can be used to dynamically adjust the gain coefficient based on the signal propagation depth: a low gain is used for shallow areas, such as 0-30cm, to avoid signal saturation distortion; for deeper areas, such as above 30cm, the gain is gradually increased to amplify the attenuated reflected signal; simultaneously, a gain upper limit is set, such as 5 times the original signal amplitude, to prevent excessive gain from amplifying noise again, ensuring that the signal amplitude in each depth region is within a reasonable observation range. In this way, the energy attenuation of the electromagnetic pulse propagating inside the retaining wall is compensated, ensuring clear identification of reflected signals in deep areas and preventing the reflection characteristics of deep cavities from being obscured.

[0077] For example, in the radar profile data acquired by ground-penetrating radar, the imaging position of the reflected signal will be horizontally offset due to the pulse propagation angle, i.e., when it is not perpendicular to the incident direction, leading to misjudgment of the defect location. For instance, a cavity with an actual depth of 50cm may be displayed as 45cm in the original image. Therefore, offset imaging processing is required. For example, the average dielectric constant of the target masonry retaining wall can be determined through on-site calibration or empirical values, and the propagation speed of the electromagnetic pulse in the masonry can be calculated using the following formula: Where c is the speed of light and εr is the average dielectric constant; then, using the Kirchhoff migration method, the actual emission point of the reflected signal is calculated backward based on the propagation speed and signal arrival time, and the offset signal is re-imaged to ensure that the reflection features in the radar image accurately correspond to the spatial location of the internal defects. In this way, the propagation path deviation of the electromagnetic pulse is corrected, making the imaged position of the reflected signal consistent with the actual defect location.

[0078] For example, after filtering, gain adjustment, and offset imaging processing, the reflection characteristics of internal cavities in the radar image are clearly presented: the in-phase axis with strong reflection amplitude appears as a continuous, straight, or slightly curved bright line, usually with an amplitude more than three times higher than the surrounding signal, corresponding to the upper and lower interfaces of planar cavities; the hyperbolic reflection characteristics appear as hyperbolic bright areas spreading outwards from the center of the cavity, corresponding to the boundary reflection of spherical or irregular cavities. For example, a reflection amplitude threshold can be set, such as 2.5 times the amplitude of a healthy area, to filter out in-phase axes and hyperbolic features with amplitudes exceeding the reflection amplitude threshold; then, the spatial regions corresponding to these features are extracted using an image segmentation algorithm, combined with the path coordinates of the radar scan, and converted into a three-dimensional spatial range inside the retaining wall, ultimately identifying this area as a suspected area of ​​internal cavity disease.

[0079] Secondly, the continuous time-domain signals collected by the acoustic emission sensing device are subjected to event detection, source localization, and event parameter analysis to identify spatially concentrated areas where the number of events per unit time exceeds a preset frequency threshold or the event energy exceeds a preset energy threshold. These areas serve as indicators of masonry loosening defects. This is because after the bonding between the masonry units and the mortar in the masonry retaining wall fails, the masonry will undergo relative displacement, friction, or collision due to uneven stress. This process releases acoustic emission signals. The more severe and widespread the loosening, the more acoustic emission events are generated per unit time, and the greater the energy released by a single event. Based on this, the loosening area of ​​the masonry can be accurately identified. The acoustic emission time-domain signal refers to the curve of the elastic wave signal generated inside the retaining wall due to masonry loosening and friction over time; an acoustic emission event refers to a local deformation or damage process that can generate an acoustic emission signal.

[0080] For example, since continuous time-domain signals contain a large amount of invalid noise, event detection needs to be performed on the continuous time-domain signals first. For example, a dual-threshold method can be used for event detection: first, set a trigger threshold, such as 200μV, and a latching threshold, such as 70% of the trigger threshold. When the signal amplitude exceeds the trigger threshold, it is determined that an event has started; second, continuously monitor the signal until the amplitude drops below the latching threshold, which is determined to be the end of the event; third, set the minimum and maximum duration of the event, such as 50μs and 5ms, to eliminate transient noise and abnormal interference, such as noise shorter than 50μs and longer than 5ms, to ensure that the filtered events are valid acoustic emission events caused by the loosening of masonry; at the same time, record the basic parameters of each event, such as the start time, peak time, peak amplitude, and duration. In this way, valid acoustic emission events are separated from the continuous time-domain signal, and environmental interference is eliminated.

[0081] For example, since a single acoustic emission sensor cannot locate the event source, a multi-sensor array is needed for source localization. For instance, a time-difference localization method can be used. At least three acoustic emission sensors are deployed at preset intervals, such as 20cm x 20cm, on the surface of the target masonry retaining wall. The three-dimensional coordinates of each sensor are recorded in a unified spatial coordinate system. Since the elastic waves generated by the same acoustic emission event propagate at a fixed wave speed, there is a certain time difference in their arrival at different sensors. Therefore, a localization equation can be established based on the three-dimensional coordinates of the acoustic emission sensors, the elastic wave speed, and the time difference. The formula for the localization equation is: Δt ij =∣r i- r j | / v, where Δt ij Let r be the time difference between acoustic emission sensors i and j. i r j Let be the distance from the event source to the acoustic emission sensing devices i and j, and v be the wave velocity. By solving the positioning equation, the three-dimensional spatial coordinates of each acoustic emission event can be obtained, thus determining the location of the masonry loosening. In this way, the three-dimensional spatial coordinates of each valid acoustic emission event inside the retaining wall are determined.

[0082] For example, key parameters related to the degree of masonry loosening are extracted, such as the number of events per unit time and event energy, which together characterize the activity and severity of loosening. For instance, the energy value of a single event can be obtained by integrating the square of the signal amplitude of the acoustic emission event; the higher the energy, the more severe the masonry loosening. The number of events per unit time can be counted by setting a time window, such as 5 minutes, counting the number of valid events within each time window, and then calculating the number of events per unit time; the more events per unit time, the more active the loosening in the area. A preset frequency threshold can be set by collecting the acoustic emission signal of a healthy retaining wall as background data, setting three times the frequency of the background events as the preset frequency threshold, and a preset energy threshold can be set five times the average energy of the background events as the preset energy threshold, ensuring that the thresholds can effectively distinguish between normal and loose states.

[0083] For example, since masonry loosening usually manifests as a concentrated problem in a localized area rather than an isolated single-point loosening, effective acoustic emission events will exhibit spatial clustering characteristics. For instance, the internal space of the retaining wall is divided into a three-dimensional grid according to a preset size, such as 5cm×5cm×5cm. The number of events per unit time and the average event energy within each three-dimensional grid are calculated. Then, three-dimensional grids with a number of events per unit time exceeding a preset frequency threshold or an average event energy exceeding a preset energy threshold are selected. Through spatial connectivity analysis, adjacent three-dimensional grids that meet the conditions are aggregated into a continuous spatial region, and isolated qualified grids are eliminated. Finally, this continuous region is determined as the indicator area of ​​masonry loosening, and its three-dimensional coordinate range is the corresponding spatial distribution range.

[0084] Finally, the suspected areas of mortar void defects, suspected areas of internal void defects, and indicated areas of masonry loosening defects are registered and fused under a unified spatial coordinate system to generate and output internal defect identification results that include the type, scale, and distribution of internal defects. The unified spatial coordinate system refers to a spatial coordinate system established based on the coordinate system of the 3D point cloud data, ensuring that the data collected by each detection device has the same spatial reference. Specifically, the spatial coordinates of the suspected areas of mortar void defects, suspected areas of internal void defects, and indicated areas of masonry loosening defects are first transformed to a unified coordinate system, and then fused through spatial intersection analysis and redundancy elimination processing to obtain the final internal defect identification results.

[0085] For example, if a region is simultaneously identified as a suspected area of ​​mortar void defects and a suspected area of ​​internal void defects under a unified spatial coordinate system, the region is actually an internal void through fusion analysis. Its size is determined by spatial volume calculation in the unified spatial coordinate system, and the corresponding spatial coordinate range is determined, forming an internal defect identification result that includes the type, scale and distribution of internal defects.

[0086] Furthermore, the step of "filtering and extracting waveform features from the waveform data collected by the ultrasonic testing device to identify areas where the sound wave velocity is lower than a preset wave velocity threshold as suspected areas of mortar void defects" includes: obtaining the masonry structure parameters corresponding to the defect warning area, wherein the masonry structure parameters include typical sound wave velocity values ​​of the stone and mortar; based on the typical sound wave velocity values ​​of the stone and mortar, combined with the theoretical volume ratio of stone and mortar in the masonry, calculating the theoretical comprehensive wave velocity value of the defect warning area under mortar fullness using a composite material equivalent wave velocity model; setting the theoretical comprehensive wave velocity value as a preset wave velocity threshold; filtering and extracting waveform features from the waveform data collected by the ultrasonic testing device to obtain the actual sound wave velocity distribution; and identifying continuous areas in the actual sound wave velocity distribution that are lower than the preset wave velocity threshold as suspected areas of mortar void defects.

[0087] In this embodiment, the masonry structure parameters corresponding to the disease warning area are first obtained. These parameters include the typical acoustic velocity values ​​of the stone and the mortar. Specifically, the masonry structure parameters can be obtained by consulting the construction drawings of the target masonry retaining wall, conducting on-site sampling tests, etc.

[0088] For example, if the stone used in the disease warning area is granite and the mortar is cement mortar, query prior data or collect stone and mortar samples from the site, conduct sound wave velocity tests in the laboratory, and then obtain the typical sound wave velocity value of granite as 5000-6000m / s and the typical sound wave velocity value of cement mortar as 3000-4000m / s.

[0089] Secondly, based on the typical acoustic wave velocity values ​​of stone and mortar, and combined with the theoretical volume ratio of stone and mortar in the masonry, the theoretical comprehensive wave velocity value of the disease warning area under mortar fullness is calculated using a composite material equivalent wave velocity model. The theoretical volume ratio of stone and mortar in the masonry can be obtained by retrieving the original design documents, construction technical briefing materials, or industry masonry standards of the target masonry retaining wall. This theoretical volume ratio directly affects the bond strength between the masonry and mortar, the overall density of the masonry, and its mechanical stability. For example, if the mortar volume ratio is too low, it will lead to incomplete filling of the joints, reducing the overall structural integrity and increasing the risk of mortar voids and loosening of the masonry. If the mortar volume ratio is too high, it may cause cracks due to shrinkage deformation, weakening the load-bearing capacity of the retaining wall. Simultaneously, this ratio also affects the propagation characteristics of ultrasonic detection signals.

[0090] The composite material equivalent wave velocity model is a mathematical model that calculates the equivalent wave velocity based on the volume proportion of each medium, thereby determining the equivalent wave velocity of sound wave propagation in multi-medium composite materials. Specifically, assuming the volume proportion of stone is V1, the volume proportion of mortar is V2, and V1 + V2 = 1, the typical sound wave velocity of stone is v1, and the typical sound wave velocity of mortar is v2, then the theoretical comprehensive wave velocity value v of the disease warning area under mortar full-saturation state can be calculated using the composite material equivalent wave velocity model formula: v = V1 × v1 + V2 × v2. For example, if the volume proportion of stone is 70% and the typical sound wave velocity of stone is 5500 m / s, and the volume proportion of mortar is 30% and the typical sound wave velocity of mortar is 3500 m / s, then the theoretical comprehensive wave velocity value v of the disease warning area under mortar full-saturation state is v = 0.7 × 5500 + 0.3 × 3500 = 4900 m / s.

[0091] Secondly, the theoretical comprehensive wave velocity value is set as a preset wave velocity threshold. Specifically, the theoretical comprehensive wave velocity value is the normal sound wave propagation speed of the target masonry retaining wall under the theoretical filling state of the mortar. If the detected wave velocity is lower than this value, it indicates that there is mortar voiding, that is, air medium replaces mortar, which leads to a decrease in wave velocity.

[0092] Furthermore, the waveform data acquired by the ultrasonic testing device is filtered and its features are extracted to obtain the actual sound wave velocity distribution. For example, a wavelet denoising algorithm can be used to filter the original waveform data to remove interference from environmental and equipment noise. Then, based on the time difference between the ultrasonic signal transmission and reception times, combined with the distance between the ultrasonic testing probes, the actual sound wave velocity of each ultrasonic testing device is calculated using the formula: v = distance / time difference. Finally, interpolation is used to obtain the actual sound wave velocity distribution of the entire disease warning area.

[0093] Finally, continuous areas in the actual acoustic wave velocity distribution that are below a preset wave velocity threshold are identified as suspected areas of mortar voiding. Specifically, the actual acoustic wave velocity distribution is divided into regions, and areas where the actual acoustic wave velocity is below the preset wave velocity threshold are selected. Then, through spatial connectivity analysis, continuous areas below the preset wave velocity threshold are identified as suspected areas of mortar voiding, and their spatial coordinate range is recorded.

[0094] In summary, compared to existing technologies, this application processes and integrates the detection data collected by the non-destructive testing equipment within the disease warning area through the data processing module 13, identifying internal disease identification results, including the type, scale, and distribution of internal diseases. This fully leverages the specialized detection advantages of each non-destructive testing equipment for different types of diseases, accurately identifying the type, scale, and distribution of internal diseases, and providing comprehensive and reliable data support for subsequent disease risk level assessment.

[0095] The diagnostic report generation module 14 is used to perform correlation analysis between the internal disease identification results and the deformation index, assess the comprehensive risk level of the disease, and generate a disease diagnostic report.

[0096] The aforementioned modules have obtained a deformation index characterizing the severity of deformation through surface deformation analysis, and obtained internal defect identification results including the type, scale, and distribution of internal defects through the fusion of multi-source non-destructive testing data. Since there is a strong positive correlation between deformation and internal defects, a correlation analysis can be performed based on the two. A defect risk correlation assessor integrates the impact of surface deformation on structural stability, the severity and development trend of internal defects, and then assesses the comprehensive risk level, providing reliable support for subsequent retaining wall maintenance decisions.

[0097] To address the aforementioned issues, this application uses a diagnostic report generation module 14 to perform correlation analysis between the internal disease identification results and the deformation index, assess the comprehensive risk level of the disease, and generate a disease diagnostic report.

[0098] Specifically, the diagnostic report generation module 14 is used to: construct a disease risk association evaluator based on a machine learning classification algorithm; input the internal disease identification results and the deformation index into the disease risk association evaluator and output a comprehensive risk level; and generate a structured disease diagnostic report based on the comprehensive risk level and the internal disease identification results.

[0099] In this embodiment, a disease risk association evaluator is first constructed based on a machine learning classification algorithm. This disease risk association evaluator is a model used to establish the mapping relationship between internal disease identification results, deformation index, and comprehensive risk level.

[0100] Secondly, the internal disease identification results and deformation index are input into the disease risk correlation assessment device, and the output is the comprehensive risk level. For example, the comprehensive risk level can be divided into three levels: mild risk, moderate risk, and severe risk, each corresponding to different treatment priorities. For example, if the internal disease identification results and deformation index are: internal cavity disease, scale 0.1m 3 Distribution density is 2 per m 2 The deformation index is 0.8. Both are input into the disease risk association assessment tool, and the output shows a comprehensive risk level of severe risk.

[0101] Finally, a structured disease diagnosis report is generated based on the comprehensive risk level and the internal defect identification results. Specifically, the structured disease diagnosis report may include basic information about the target masonry retaining wall, an overview of the inspection process, details of internal defect identification, deformation index test results, comprehensive risk level assessment conclusions, and treatment recommendations, ultimately outputting the disease diagnosis report in document form. The treatment recommendations can be dynamically formulated based on the comprehensive risk level; for example, mild risk recommends regular monitoring, moderate risk recommends partial repair, and severe risk recommends overall reinforcement or demolition and reconstruction.

[0102] Furthermore, the construction process of the "disease risk association evaluator" includes: collecting multiple historical cases of masonry retaining walls with known safety states; for each historical case of masonry retaining walls, obtaining the historical deformation index and historical internal disease identification results to form a model training input feature set; based on the actual safety state of each historical case of masonry retaining walls, labeling each sample in the model training input feature set with a true historical comprehensive risk level to form a model training label set; using a machine learning classification algorithm to construct the disease risk association evaluator; and using the model training input feature set and the corresponding model training label set to perform supervised learning training on the disease risk association evaluator until verification convergence, thus obtaining the trained disease risk association evaluator.

[0103] In this embodiment, multiple historical cases of masonry retaining walls with known safety conditions are first collected. For each historical case, the historical deformation index and historical internal defect identification results are obtained to form the model training input feature set. Specifically, historical cases of masonry retaining walls from different regions, with different service years and different defect types are collected by reviewing engineering archives, inspection reports, and maintenance records to ensure that the cases cover various safety conditions such as mild, moderate, and severe risks. The corresponding historical deformation index and historical internal defect identification results are extracted from the archive data of each historical case. These data are then organized and cleaned to remove missing and outlier values. Finally, the relevant data of each historical case are combined into a feature vector, and all feature vectors together constitute the model training input feature set.

[0104] Secondly, based on the actual safety status of each historical case of a masonry retaining wall, a true historical comprehensive risk level is assigned to each sample in the model training input feature set, forming a model training label set. Specifically, based on the actual safety status of each historical case of a masonry retaining wall, such as whether a collapse has occurred, whether reinforcement treatment has been carried out, and the operation and maintenance assessment conclusions, a unified comprehensive risk level assessment standard is formulated by those skilled in the art, and then a corresponding comprehensive risk level label is assigned to each sample in the model training input feature set accordingly.

[0105] For example, the comprehensive risk level assessment criteria can be set as follows: a situation with only minor mortar voids and localized minor cracks that remain stable without long-term reinforcement and show no obvious deformation trend is marked as a mild risk; a situation with an expansion of localized mortar voids, a small amount of loose masonry, or a slow deformation that requires targeted monitoring is marked as a moderate risk; and a situation with localized collapses, large-area masonry loosening, or large-scale complete mortar voids that seriously threaten structural safety and require emergency treatment is marked as a severe risk.

[0106] For example, based on the aforementioned comprehensive risk level assessment criteria, if a historical case of a masonry retaining wall exhibits serious structural safety issues such as partial collapse or large-area masonry instability, it is marked as severely risky; if a retaining wall in a case only has minor mortar voids and localized micro-cracks, and remains stable without long-term reinforcement and shows no obvious signs of deterioration, it is marked as mildly risky; if a retaining wall in a case has localized mortar voids covering 10%-30% of its surface and 3-5 masonry stones slightly loosening, requiring regular monitoring, it is marked as moderately risky. The labeled tags of each sample in the model training input feature set together constitute the model training label set.

[0107] Next, a disease risk association evaluator is constructed using a machine learning classification algorithm. For example, leveraging the advantages of the random forest algorithm—its resistance to overfitting, strong robustness, and adaptability to multi-dimensional features—it can be used to construct the disease risk association evaluator. The main structure is as follows: 100 CART binary classification decision trees serve as the base learner, with a maximum tree depth of 10. The Gini coefficient is used as the node splitting criterion, and pre-pruning limits the tree complexity. During training, a bootstrap sampling method is used to generate an independent training subset for each tree. During node splitting, some input features are randomly selected. Finally, the prediction results of the 100 trees are fused through a hard voting mechanism to output three risk levels: mild, moderate, and severe, balancing prediction accuracy and generalization ability.

[0108] Finally, the disease risk association evaluator is trained under supervised learning using the model training input feature set and the corresponding model training label set until the verification convergence, thus obtaining the trained disease risk association evaluator.

[0109] For example, the model training input feature set and the corresponding model training label set are divided into a training set and a validation set in a 7:3 ratio; the disease risk association evaluator is trained using the training set through supervised learning, and the model parameters are continuously adjusted through the backpropagation algorithm; after each training round, the classification accuracy of the disease risk association evaluator is evaluated using the validation set; when the accuracy of the validation set no longer improves after 5 consecutive rounds, it is considered that the validation has converged, the training is stopped, and the trained disease risk association evaluator is obtained.

[0110] In summary, compared to existing technologies, this application uses the diagnostic report generation module 14 to perform correlation analysis between the internal disease identification results and the deformation index, assesses the comprehensive risk level of the disease, and generates a disease diagnosis report. Thus, by deeply correlating the internal disease identification results with the deformation index, the one-sidedness of single-dimensional assessment is avoided, achieving a scientific and accurate determination of the comprehensive risk level of the disease. Furthermore, the generated disease diagnosis report can provide direct and comprehensive technical support for the precise maintenance and safety management of masonry retaining walls.

[0111] In summary, the embodiments of this application have at least the following technical effects: Compared with the prior art, this application first collects three-dimensional point cloud data of the surface of the target masonry retaining wall through the early warning area identification module 11. By calculating the spatial distance deviation from each point in the three-dimensional point cloud data to the reference plane, local areas where the deformation exceeds a preset threshold are identified as disease early warning areas, and the deformation index of each disease early warning area is quantitatively calculated. In this way, the accurate locking of high-risk disease areas and the multi-dimensional quantitative characterization of the overall deformation severity are achieved, providing accurate targeting and reliable data support for subsequent adaptive configuration of internal non-destructive testing schemes and correlation assessment of internal disease risks.

[0112] Secondly, this application uses the disease collaborative detection module 12 to adaptively configure internal non-destructive testing (NDT) schemes for each disease early warning area based on the deformation index, and drives corresponding NDT equipment to perform collaborative testing. The NDT equipment includes an ultrasonic testing device, ground-penetrating radar, and an acoustic emission sensor. Thus, based on the deformation index, differentiated adaptive configuration of NDT schemes for disease early warning areas is achieved. Through collaborative testing using ultrasonic testing devices, ground-penetrating radar, and acoustic emission sensors, the specialized testing advantages of each device are fully utilized. This avoids over-testing in low-risk areas and under-testing in high-risk areas, improves the accuracy, comprehensiveness, and efficiency of internal disease detection, and optimizes the allocation of testing resources.

[0113] Furthermore, this application processes and integrates the detection data collected by the non-destructive testing equipment within the disease warning area through the data processing module 13, identifying internal disease identification results including the type, scale, and distribution of internal diseases. In this way, the specialized detection advantages of each non-destructive testing equipment for different types of diseases are fully utilized, accurately identifying the type, scale, and distribution of internal diseases, and providing comprehensive and reliable data support for subsequent disease risk level assessment.

[0114] Finally, this application uses the diagnostic report generation module 14 to perform correlation analysis between the internal defect identification results and the deformation index, assesses the comprehensive risk level of the defects, and generates a defect diagnosis report. In this way, by deeply correlating the internal defect identification results with the deformation index, the one-sidedness of single-dimensional assessment is avoided, achieving a scientific and accurate determination of the comprehensive risk level of the defects. Furthermore, the generated defect diagnosis report can provide direct and comprehensive technical support for the precise maintenance and safety management of masonry retaining walls.

[0115] Through the aforementioned technical solution, this application achieves a closed-loop process for the entire process of masonry retaining wall disease, from surface deformation identification, adaptive collaborative non-destructive testing, accurate internal disease identification, and comprehensive risk assessment, through the coordinated operation of the early warning area identification module, the disease collaborative detection module, the data processing module, and the diagnostic report generation module. It accurately anchors the disease early warning area and quantifies the deformation index using 3D point cloud data. Based on the deformation index, it adaptively configures a collaborative detection scheme using ultrasonic testing devices, ground-penetrating radar, and acoustic emission sensors. Furthermore, it comprehensively identifies the type, scale, and distribution of internal diseases through multi-source detection data fusion. Finally, it combines the internal disease identification results with the deformation index correlation analysis to obtain a scientific and objective comprehensive risk level and a complete disease diagnosis report. This effectively solves the problems of fragmentation, strong subjectivity, and insufficient accuracy in existing masonry retaining wall disease data processing methods, achieving precise disease detection, comprehensive disease identification, accurate risk assessment, and refined operation and maintenance support, providing reliable and comprehensive technical support for the safe operation and maintenance management of masonry retaining walls.

[0116] It should be noted that the descriptions of each embodiment in the above embodiments have different focuses. For parts that are not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0117] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0118] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in one or more blocks of the flowchart illustrations and / or one or more blocks of the block diagrams.

[0119] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in one or more flowcharts and / or one or more block diagrams.

[0120] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions specified in one or more flowcharts and / or one or more block diagrams.

[0121] Although preferred embodiments of the invention have been described, those skilled in the art, once they have learned the basic inventive concept, can make other changes and modifications to these embodiments.

[0122] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of this invention and its equivalents, this invention also intends to include these modifications and variations.

Claims

1. A data processing system for defects in masonry retaining walls, characterized in that, The system includes: a warning area identification module, used to collect three-dimensional point cloud data of the surface of the target masonry retaining wall, and by calculating the spatial distance deviation from each point in the three-dimensional point cloud data to the reference plane, identify local areas where the deformation exceeds a preset threshold as disease warning areas, and quantify the deformation index of each disease warning area; a disease collaborative detection module, used to adaptively configure an internal non-destructive testing scheme for each disease warning area according to the deformation index, and drive the corresponding non-destructive testing equipment to perform collaborative testing, wherein the non-destructive testing equipment includes an ultrasonic testing device, a ground penetrating radar, and an acoustic emission sensor; a data processing module, used to process and fuse the detection data collected by the non-destructive testing equipment in the disease warning area, and identify the internal disease identification results including the internal disease type, scale, and distribution; and a diagnostic report generation module, used to perform correlation analysis between the internal disease identification results and the deformation index, evaluate the comprehensive risk level of the disease, and generate a disease diagnostic report.

2. The data processing system for masonry retaining wall defects according to claim 1, characterized in that, The warning area identification module is specifically used to: collect the current three-dimensional point cloud data of the target masonry retaining wall surface through a three-dimensional laser scanning device; and obtain a reference surface that matches the original design surface or historical health status surface of the retaining wall. Calculate the spatial Euclidean distance from each point in the three-dimensional point cloud data to the reference surface, and use it as the spatial distance deviation value; All points whose spatial distance deviation exceeds a preset deformation threshold are classified as abnormal points, and the continuous area formed by spatially adjacent abnormal points is identified as a disease warning area.

3. The data processing system for masonry retaining wall defects according to claim 2, characterized in that, Obtaining a reference surface that matches the original design surface or historical health status surface of the retaining wall includes: acquiring the original design 3D model or the 3D point cloud data of the target masonry retaining wall under historical health status as reference surface data; using a region growing algorithm to segment the reference surface data into multiple surface segments based on curvature features; performing planar or quadratic surface fitting on each surface segment using the RANSAC algorithm; and integrating all fitted planes or surfaces to form a reference surface that fits the original design or historical health geometry of the retaining wall.

4. The data processing system for masonry retaining wall defects according to claim 1, characterized in that, The warning area identification module is further specifically used to: extract the spatial distance deviation values ​​of all abnormal points contained within each disease warning area, and calculate the average value of the spatial distance deviations of all abnormal points as the overall average spatial distance deviation. The proportion of abnormal points to the total number of points in the disease warning area is calculated as a deformation index. Based on the spatial distance deviation value and the overall average spatial distance deviation, calculate the joint deformation concentration coefficient and the motion dispersion of the masonry unit; The deformation index, the joint deformation concentration coefficient, and the motion dispersion of the masonry unit are normalized and weighted to obtain the deformation index that characterizes the overall deformation severity of the disease warning area.

5. The data processing system for masonry retaining wall defects according to claim 4, characterized in that, Based on the spatial distance deviation value and the overall average spatial distance deviation, the deformation concentration coefficient of the masonry joint and the motion dispersion of the masonry unit are calculated, including: calculating the normal vector of each point in each of the disease warning areas in the three-dimensional point cloud data; detecting potential linear groove features based on the abrupt change in the normal vector direction or the local maximum of the curvature; and, combined with the prior knowledge of the layered masonry of the masonry retaining wall, selecting linear features whose orientation and the angle between the horizontal direction and the gravity direction are within a preset range as suspected masonry joints; calculating the average spatial distance deviation in the distribution area of ​​suspected masonry joints, and dividing the average spatial distance deviation by the overall average spatial distance deviation to obtain the deformation concentration coefficient of the masonry joint; segmenting point cloud clusters with continuous surfaces based on the region growing algorithm, and, according to the typical size range of the blocks in the masonry retaining wall, performing secondary segmentation on point cloud clusters whose area or volume exceeds the reasonable range of a single block to obtain multiple unit blocks corresponding to the physical masonry; calculating the average spatial distance deviation of each unit block, and calculating the ratio of the standard deviation to the average value of the average spatial distance deviation of all unit blocks as the motion dispersion of the masonry unit.

6. The data processing system for masonry retaining wall defects according to claim 1, characterized in that, The disease collaborative detection module is specifically used for: establishing a mapping table between deformation index ranges and non-destructive testing (NDT) schemes, wherein the NDT schemes define the combination of NDT equipment to be called and the corresponding detection parameters; querying the mapping table based on the deformation index of the disease warning area to determine the target NDT scheme corresponding to the disease warning area; and calling one or more of the corresponding ultrasonic testing device, ground penetrating radar, and acoustic emission sensing device according to the target NDT scheme, configuring the corresponding detection parameters, and performing collaborative detection within the disease warning area.

7. The data processing system for masonry retaining wall defects according to claim 1, characterized in that, The data processing module is specifically used for: filtering and extracting waveform features from the waveform data collected by the ultrasonic testing device to identify areas where the sound wave velocity is lower than a preset wave velocity threshold, as suspected areas of mortar void defects; filtering, gain adjustment, and offset imaging of radar profile data collected by ground penetrating radar to identify in-phase axes with strong reflection amplitude or reflection features exhibiting a hyperbolic shape in the radar image, as suspected areas of internal void defects; and performing event detection, source localization, and event parameter analysis on continuous time-domain signals collected by the acoustic emission sensing device to identify spatially concentrated areas where the number of events exceeds a preset frequency threshold or the event energy exceeds a preset energy threshold within a unit time, as indication areas of masonry loosening defects. The suspected areas of mortar void defects, the suspected areas of internal void defects, and the indicated areas of masonry loosening defects are registered and fused in a unified spatial coordinate system to generate and output internal defect identification results including the type, scale, and distribution of internal defects.

8. The data processing system for masonry retaining wall defects according to claim 7, characterized in that, The waveform data acquired by the ultrasonic testing device is filtered and its features are extracted to identify areas where the sound wave velocity is lower than a preset velocity threshold, which are then identified as suspected areas of mortar voiding. This process includes: acquiring the masonry structure parameters corresponding to the warning area, wherein the masonry structure parameters include typical sound wave velocity values ​​of the stone and mortar; based on the typical sound wave velocity values ​​of the stone and mortar, and combined with the theoretical volume ratio of stone and mortar in the masonry, calculating the theoretical comprehensive velocity value of the warning area under mortar fullness using a composite material equivalent velocity model; setting the theoretical comprehensive velocity value as a preset velocity threshold; filtering and extracting waveform features from the waveform data acquired by the ultrasonic testing device to obtain the actual sound wave velocity distribution; and identifying continuous areas in the actual sound wave velocity distribution that are lower than the preset velocity threshold as suspected areas of mortar voiding.

9. The data processing system for masonry retaining wall defects according to claim 1, characterized in that, The diagnostic report generation module is specifically used for: constructing a disease risk association assessor based on a machine learning classification algorithm; inputting the internal disease identification results and the deformation index into the disease risk association assessor, and outputting a comprehensive risk level; A structured disease diagnosis report is generated based on the comprehensive risk level and the internal disease identification results.

10. The data processing system for masonry retaining wall defects according to claim 9, characterized in that, The construction process of the disease risk association assessor includes: collecting multiple historical cases of masonry retaining walls with known safety states; for each historical case of masonry retaining walls, obtaining the historical deformation index and historical internal disease identification results to form a model training input feature set; based on the actual safety state of each historical case of masonry retaining walls, labeling each sample in the model training input feature set with a true historical comprehensive risk level to form a model training label set; using a machine learning classification algorithm to construct the disease risk association assessor; and using the model training input feature set and the corresponding model training label set to perform supervised learning training on the disease risk association assessor until verification convergence, thus obtaining the trained disease risk association assessor.

Images