Building inspection method and system based on existing building structure

Abstract

The application discloses a kind of building inspection method and system based on existing building structure, the method includes the following steps: based on multi-period point cloud registration calculation three-dimensional displacement, fusion principal component generates structure deformation stability index, identifies potential instability area;Through multi-scale surface fitting extraction curvature and normal vector evolution gradient, combined with signal-to-noise ratio weighting, construct local distortion sensitivity index;Gradient entropy and roughness are extracted to abnormal area, and material degradation confidence factor is generated by fusing reflection intensity variation;Based on spatial gridding calculation degradation concentration, coupled maintenance frequency and obtained weak degree score by fuzzy mapping, combined with importance weight output risk priority;Deformation rate sequence is extracted to high-risk component, trend is fitted and cumulative score is constructed, and weak score is double-target normalized, and by pareto front and euclidean distance solution, determine the highest position of comprehensive risk and label failure urgency level.The application improves the identification precision of existing building structure deformation.

Description

Technical Field

[0001] This invention relates to the field of construction, and in particular to a building inspection method and system based on existing building structures. Background Technology

[0002] Existing building structures, during long-term service, are continuously subjected to multiple factors such as environmental erosion, repeated loads, and natural material aging, leading to gradual performance degradation. Current mainstream inspection methods still rely primarily on manual visual inspection and localized checks, which suffer from significant problems such as strong subjectivity, high missed detection rates, weak dynamic evolution recognition capabilities, and a lack of spatial continuity. With the widespread adoption of 3D laser scanning and mobile scanning technologies, high-precision, full-coverage point cloud data of building surfaces can be acquired, enabling time-series monitoring through multi-phase scanning. However, most studies focus on displacement and deformation analysis, neglecting material state information such as point cloud reflection intensity and gradient distribution, and often remain at the level of single-phase state recognition, lacking the ability to identify multi-phase evolution trends. Based on this, this invention proposes a building inspection method and system based on existing building structures. Summary of the Invention

[0003] This invention provides a building inspection method based on existing building structures, characterized by comprising: S10. Based on the point cloud data of the building structure surface obtained by multi-phase high-precision laser scanning, after joint registration with control points by the iterative nearest point method, the three-dimensional displacement characteristics and spatial autocorrelation index of the point cloud between adjacent measurements are calculated, and a comprehensive structural deformation stability index is generated by the principal component fusion method. S20. Based on stability indices, identify potential unstable regions, perform multi-scale surface fitting on their local neighborhoods, extract the evolution gradient of curvature and normal vector change rate, and introduce local point cloud signal-to-noise ratio dynamic adjustment weights to generate a local surface distortion sensitivity index. S30. For abnormal regions where the distortion sensitivity index exceeds the threshold, extract the distribution entropy and local roughness of the histogram of the point cloud gradient magnitude, construct the initial degradation response intensity, and integrate the synchronously acquired surface reflection intensity variation coefficient as an environmental interference compensation term to generate the material degradation confidence factor. S40. Based on the spatial rasterization distribution of confidence factors, calculate the degradation concentration coefficient for each structural component and couple it with the maintenance frequency in the historical inspection records. Map the structural weakness score through trapezoidal fuzzy membership degree, sort it in combination with the component functional importance weight, and output the component-level risk priority sequence. S50. For high-risk components in the priority sequence, extract the deformation rate sequence of the most recent measurements, fit a linear trend and construct a historical rate cumulative score. Normalize this score and the weakness score in a dual-objective manner, solve for Pareto optimality through Euclidean distance, determine the list of specific parts with the highest comprehensive risk and mark their failure urgency level.

[0004] The building inspection method based on existing building structures, as described above, utilizes point cloud data of the building structure surface acquired through multiple high-precision laser scans. After joint registration with control points using the iterative nearest point method, the three-dimensional displacement characteristics and spatial autocorrelation index of the point cloud between adjacent scans are calculated. Finally, a comprehensive structural deformation stability index is generated using a principal component fusion method. Specifically, it consists of the following sub-steps: After multiple laser point cloud data are jointly registered with control points using the iterative nearest point method, the three-dimensional displacement modulus of the corresponding points is calculated, and a basic deformation stability index is constructed to quantify the discreteness of the overall structural deformation. Spatial autocorrelation analysis was performed by constructing a spatial weight matrix based on the K-nearest neighbor Gaussian kernel, and the Moran index was calculated to quantify the spatial clustering of displacement values. By combining the two dimensions of displacement dispersion and spatial clustering, a comprehensive structural deformation stability index is generated by integrating the basic deformation stability index and the normalized Moran index through principal component analysis.

[0005] The building inspection method based on existing building structures, as described above, identifies potential instability zones based on stability indices, performs multi-scale surface fitting on their local neighborhoods, extracts the evolution gradient of curvature and normal vector change rate, and introduces dynamic adjustment weights based on the local point cloud signal-to-noise ratio to generate a local surface distortion sensitivity index. Specifically, it consists of the following sub-steps: Based on the distribution of structural deformation stability index, regions below a preset threshold are selected as potential instability zones. The surface of the anomalous region is reconstructed by fitting the multi-scale moving least squares method for the potential unstable region, and the temporal change rate of Gaussian and mean curvature is extracted. The direction identification term is calculated by combining the change rate of normal vector and the consistency of curvature gradient direction. Taking into account the local quality differences of point clouds, a local surface distortion sensitivity index is constructed by introducing a signal-to-noise ratio dynamic adjustment weight based on curvature statistics, integrating multi-scale curvature change rate and directional stability, and combining the second-order curvature derivative to enhance edge response.

[0006] The building inspection method based on existing building structures, as described above, involves extracting the distribution entropy and local roughness of the point cloud gradient magnitude histogram for abnormal areas where the distortion sensitivity index exceeds a threshold. This constructs an initial degradation response intensity, and integrates the synchronously acquired surface reflection intensity variation coefficient as an environmental disturbance compensation term to generate a material degradation confidence factor. Specifically, this method comprises the following sub-steps: After identifying regions with high local surface distortion sensitivity index, the causes of geometric anomalies are analyzed by fusing point cloud reflection intensity and gradient features, distinguishing between deformation caused by structural stress and damage caused by physical degradation of surface materials. Based on the extracted gradient and intensity data, the gradient distribution entropy and enhanced roughness are calculated to achieve quantitative identification of material degradation. To determine whether regions with high geometric distortion are caused by surface material degradation, a material degradation confidence factor is constructed based on texture entropy, enhanced roughness, and the coefficient of variation of reflection intensity as a criterion.

[0007] The building inspection method based on existing building structures, as described above, calculates gradient distribution entropy and enhanced roughness based on extracted gradient and intensity data to achieve quantitative identification of material degradation. Specifically, it consists of the following sub-steps: For each point to be analyzed, a local neighborhood window is constructed with that point as the center. The gradient magnitudes of all points are collected within this window, and their normalized histogram distribution is constructed. Based on this distribution, the gradient distribution entropy is calculated. Based on the three-dimensional coordinates of each point in the point cloud, the vertical direction values ​​of all points are extracted into a two-dimensional elevation field within a local neighborhood. The basic roughness is then calculated within the same local window based on the two-dimensional elevation field. Based on the basic roughness, an enhanced roughness variance is constructed to reflect the complex morphology caused by the degradation of the surface material.

[0008] The building inspection method based on existing building structures, as described above, calculates the degradation concentration coefficient for each structural component based on the spatial rasterized distribution of confidence factors. This coefficient is coupled with the maintenance frequency in historical inspection records and mapped to a structural weakness score using trapezoidal fuzzy membership. The score is then ranked based on the functional importance weights of the components, outputting a component-level risk priority sequence. Specifically, this involves the following sub-steps: Based on the point cloud semantic segmentation results, the confidence factor-based distribution map is mapped to the structural components, and their internal degradation statistical features are extracted. Based on the statistical characteristics of component degradation, the aging coupling term is calculated according to the degradation concentration coefficient and historical maintenance frequency to realize dynamic analysis of the health status of structural components; The component aging coupling term is transformed into a standardized weakness score by using a trapezoidal fuzzy membership function, and then combined with functional importance weights for weighted normalization to generate a component-level priority sequence arranged in descending order of risk.

[0009] The building inspection method based on existing building structures, as described above, involves extracting the deformation rate sequence of the most recent measurements for high-risk components in the priority sequence, fitting a linear trend, constructing a historical rate cumulative score, performing bi-objective normalization on this score and the weakness score, solving for Pareto optimality using Euclidean distance, determining a list of specific parts with the highest overall risk, and marking their failure urgency level. Specifically, it consists of the following sub-steps: By using multi-period material degradation and deformation rate sequences, linear trend fitting and acceleration sensitivity analysis are performed on high-risk components to comprehensively determine whether they are in an accelerated deterioration state. By constructing an exponentially decaying weighted historical rate cumulative score to quantify the deformation evolution trend, and combining it with the static weakness score, a list of components with the highest comprehensive risk and urgent failure is determined based on Euclidean distance ranking after bi-objective normalization and Pareto front analysis.

[0010] This invention also provides a building inspection system based on existing building structures, comprising: Structural Deformation Index Module: Based on point cloud data of building structure surface obtained by multi-phase high-precision laser scanning, after joint registration with control points by the iterative nearest point method, the three-dimensional displacement characteristics and spatial autocorrelation index of point cloud between adjacent measurements are calculated, and a comprehensive structural deformation stability index is generated by the principal component fusion method. Distortion Sensitivity Index Module: Based on stability indices, potential unstable regions are identified, multi-scale surface fitting is performed on their local neighborhoods, the evolution gradient of curvature and normal vector change rate is extracted, and the local point cloud signal-to-noise ratio is introduced to dynamically adjust the weights to generate a local surface distortion sensitivity index. Material degradation module: For abnormal regions where the distortion sensitivity index exceeds the threshold, the distribution entropy and local roughness of the point cloud gradient magnitude histogram are extracted to construct the initial degradation response intensity. The synchronously acquired surface reflection intensity variation coefficient is then fused as an environmental disturbance compensation term to generate the material degradation confidence factor. Priority sequence module: Based on the spatial rasterized distribution of confidence factors, the degradation concentration coefficient is calculated for each structural component and coupled with the maintenance frequency in historical inspection records. It is then mapped to the structural weakness score through trapezoidal fuzzy membership degree and sorted by the functional importance weight of the component to output a component-level risk priority sequence. Risk Location List Module: For high-risk components in the priority sequence, extract the deformation rate sequence of the most recent measurements, fit a linear trend and construct a historical rate cumulative score. Normalize this score and the weakness score in a dual objective manner, solve for Pareto optimality through Euclidean distance, determine the specific location list with the highest comprehensive risk and mark its failure urgency level.

[0011] The beneficial effects achieved by this invention are as follows: This invention generates a deformation stability index through principal component fusion, constructs a distortion sensitivity index and a material degradation confidence factor, and innovatively introduces a coupling mechanism between degradation concentration and maintenance frequency to quantify the degree of weakness. Furthermore, by combining dynamic cumulative scores and Pareto optimal analysis, it achieves coordinated ranking of risk priority and failure urgency, thereby improving the accuracy of identifying deformation of existing building structures. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0013] Figure 1 This is a flowchart of a building inspection method based on an existing building structure, provided in Embodiment 1 of this application.

[0014] Figure 2 This is a schematic diagram of a building inspection system based on an existing building structure, provided in Embodiment 2 of this application. Detailed Implementation

[0015] 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, not all, of the embodiments of the present invention. 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.

[0016] Example 1 like Figure 1 As shown, Embodiment 1 of this application provides a building inspection method based on existing building structures, including: S10. Based on the point cloud data of the building structure surface obtained by multi-phase high-precision laser scanning, after joint registration with control points by the iterative nearest point method, the three-dimensional displacement characteristics and spatial autocorrelation index of the point cloud between adjacent measurements are calculated, and a comprehensive structural deformation stability index is generated by the principal component fusion method.

[0017] S11. After the multi-phase laser point cloud data is jointly registered with the control points using the iterative nearest point method, the three-dimensional displacement modulus of the corresponding points is calculated, and a basic deformation stability index is constructed to quantify the discreteness of the overall structural deformation.

[0018] To systematically evaluate the long-term changing trends of building geometry during observation periods, high-precision 3D laser scanning point cloud data of the same building structure at different time points are required. Since each scanning period is typically based on different instrument setup positions and coordinate systems, the original point clouds exhibit spatial misalignment. Therefore, an iterative nearest-point algorithm is used for initial registration: this algorithm iteratively finds the closest point pairs between two point clouds and minimizes their sum of squared Euclidean distances, progressively optimizing the rotation matrix and translation vector to achieve automatic point cloud alignment. However, the iterative nearest-point algorithm is susceptible to local minima, and registration accuracy decreases when texture is scarce or overlapping areas are insufficient. Therefore, ground control points are introduced as strong constraints, and their precise positions in the global coordinate system are obtained through total station measurements. The matching error of the control point pairs is added as a strong constraint to the objective function to improve registration accuracy and stability.

[0019] After registration, spatial matching is performed on the point clouds from the two phases. A KD-tree structure is used to find the nearest neighbor of each point in the previous scan for each point in the later scan, forming a set of corresponding point pairs. For each pair of matched points, the coordinate differences in the X, Y, and Z directions are calculated to form a three-dimensional displacement vector. And calculate its Euclidean modulus. The spatial displacement distance of each point is obtained. Within the entire analysis region, the arithmetic mean of the displacement moduli of all matching points is calculated. This reflects the overall average displacement level. Simultaneously, the spatial standard deviation of these displacement moduli is calculated. This characterizes the spatial dispersion of displacement. Based on this, a fundamental deformation stability index is constructed. ,in It is a very small positive number to prevent the denominator from being zero. This indicator reflects the consistency of deformation: when the structure undergoes uniform settlement or overall translation, A higher value indicates that the deformation is consistent and stable; if the structure exhibits localized twisting, tilting, or uneven settlement, A low value indicates the presence of an abnormal deformation area.

[0020] S12. Based on the K-nearest neighbor Gaussian kernel, a spatial weight matrix is ​​constructed to perform spatial autocorrelation analysis, and the Moran index is calculated to quantify the spatial clustering of displacement values.

[0021] Basic deformation stability indices can only reflect the dispersion of overall deformation and are insufficient to capture local anomaly clustering patterns. Spatial autocorrelation analysis is introduced to identify whether displacement anomalies exhibit spatial clustering characteristics. To this end, a spatial weight matrix is ​​constructed. This method is used to quantify the spatial proximity relationships between points in a point cloud: the K-nearest neighbor strategy is used to determine the neighbor set of each point, and a Gaussian kernel function is used. Assign weights, Let be the Euclidean distance between points i and j, and h be the bandwidth parameter. Based on this weight matrix, the global Moran index is calculated to quantify the spatial clustering degree of displacement intensity. In laser scanning, the point cloud density in different regions is significantly affected by the scanning angle, distance, and occlusion, leading to inconsistent Moran index values ​​for the same deformation intensity in dense and sparse regions. Local topological density is used to normalize and correct the Moran index. The normalized Moran index enhances the comparability between different regions, making spatial clustering analysis unaffected by sampling density interference and more realistically reflecting the spatial pattern of structural deformation.

[0022] S13. Combining the two dimensions of displacement dispersion and spatial clustering, a comprehensive structural deformation stability index is generated by integrating the basic deformation stability index and the normalized Moran index through principal component analysis.

[0023] Considering that structural stability depends not only on the dispersion of overall displacement (i.e., whether the deformation is uniform) but also on its spatial distribution pattern (i.e., whether anomalies are concentrated in clusters), these two aspects need to be considered as complementary dimensions for comprehensive evaluation. First, the basic deformation stability index and the normalized Moran's index are standardized to eliminate dimensional differences, resulting in... and Subsequently, a joint covariance matrix is ​​constructed to reflect their spatial variability and covariance. Principal component analysis is then used to extract the two eigenvalues ​​of this matrix: the variance explained by the first principal component. The variance explained by the second principal component By using the proportion of eigenvalues ​​as weights, a comprehensive structural deformation stability index is constructed to more comprehensively and robustly characterize the consistency and stability level of the overall structural deformation of a building. When the structure is stable and the deformation is uniform, and The values ​​are generally high, with S values ​​approaching 1; however, when local clustering anomalies or overall instability occur, the S value decreases significantly.

[0024] S20. Based on the stability index, identify potential unstable regions, perform multi-scale surface fitting on their local neighborhoods, extract the evolution gradient of curvature and normal vector change rate, and introduce local point cloud signal-to-noise ratio dynamic adjustment weight to generate local surface distortion sensitivity index.

[0025] S21. Based on the distribution of structural deformation stability index, regions below a preset threshold are selected as potential instability zones.

[0026] Based on structural deformation stability indices, the entire building surface is statistically analyzed at the component unit level, and the S-value of each region is calculated. A threshold value determined by K-means clustering is set, and regions below the threshold are identified as potential instability zones. These regions correspond to weak points in the structure, including local walls affected by uneven foundation settlement, the periphery of expansion joints affected by temperature stress, or component connection points where damage has already spread. Deformation in these regions typically exhibits inconsistency, i.e., large differences in displacement between adjacent points and inconsistent directions, indicating early-stage defects such as crack development, interface debonding, or local buckling.

[0027] S22. The surface of the abnormal region is reconstructed by fitting the potential unstable region using the multi-scale moving least squares method. The temporal change rate of Gaussian and average curvature is extracted, and the direction identification term is calculated by combining the change rate of normal vector with the consistency of curvature gradient direction.

[0028] Potential instability zones are regional indicators with limited spatial resolution, failing to reveal specific geometric distortion morphologies. It is necessary to leverage the high spatial resolution of point clouds for refined local analysis of these anomalous regions, deeply exploring their microscale geometric evolution characteristics. Since the original point cloud contains noise and lacks continuous surface properties, direct curvature calculation is easily affected by interference. Therefore, the moving least squares method is used to reconstruct multi-scale surfaces in the anomalous regions: multiple neighborhood radii are selected, and a locally fitted surface is constructed around each sampling point to balance detail preservation and noise suppression, capturing sharp edges at a small scale and reflecting overall component morphological changes at a large scale. Based on the fitted smooth surface, the Gaussian curvature K and mean curvature H are calculated point-by-point, reflecting the developability and concavity / convexity of the local surface, respectively, and their absolute rates of change between adjacent measurements are calculated. and As a fundamental indicator of geometric evolution intensity, The time interval between adjacent measurements.

[0029] Simultaneously, the stability of the surface normal vector direction is analyzed: the rate of change of the cosine value of the angle between the normal vectors at the same location across different measurements is calculated. A larger value indicates a more drastic change in surface orientation, commonly seen in areas where cracks open or bulges appear. Let be the unit normal vector perpendicular to the surface at the i-th point during the previous scan time t. This represents the unit normal vector at the same location in the next scan time t+1. It is the product of the magnitudes of the two normal vectors.

[0030] To enhance the ability to identify persistent deformation modes, a curvature gradient direction consistency factor is introduced. This factor measures the degree to which the direction of the maximum change in curvature is maintained over time. A value of approximately 1 indicates a consistent direction, representing a continuously expanding crack; a value of approximately 0 indicates disordered direction, representing random noise or local disturbances. Let be the Gaussian curvature gradient vector at the i-th point at time t. Let be the Gaussian curvature gradient vector at the same location at time t+1. It is the dot product of the magnitudes of the two curvature gradient vectors.

[0031] Based on the rate of change of normal and the direction consistency factor, a direction identification term is constructed. A certain area simultaneously meets the high ,high ,but A higher level indicates a persistent structural distortion; high ,Low ,but Suppressed, judged as non-persistent disturbance or noise; low ,but Extremely low values ​​are considered a stable region. This indicator can effectively distinguish between true structural distortions and accidental disturbances, significantly improving the robustness of detecting early-stage minor defects.

[0032] S23. Taking into account the local quality differences of point clouds, a local surface distortion sensitivity index is constructed by introducing a signal-to-noise ratio dynamic adjustment weight based on curvature statistics, integrating multi-scale curvature change rate and directional stability, and combining the second-order curvature derivative to enhance edge response.

[0033] The quality of laser-scanned point clouds is significantly affected by the distance to the scanning station, the angle of incidence, and the surface reflection characteristics. Point clouds far from the scanner or in areas with low reflectivity often contain significant noise, leading to distortion in the calculation of higher-order geometric features such as curvature. Therefore, if the reliability of local data is not considered, noise may be misjudged as true distortion. The local point cloud signal-to-noise ratio is then introduced. As a quality metric, a higher value indicates that the geometric features of the region are more stable and reliable.

[0034] To enhance the response to edges of abrupt curvature changes, a local surface distortion sensitivity index is constructed: ,in, The multi-scale average rate of curvature change represents the average rate of curvature evolution at that point between adjacent measurements, reflecting the intensity of geometric deformation. For the adaptive weights of the curvature term, , For direction recognition terms, direction weights , For attenuation parameters, The edge enhancement coefficient, Let be the Laplace curvature modulus, i.e., the second derivative modulus of the Gaussian curvature field, which increases significantly at abrupt changes such as crack edges and corners. All points in the global scope The maximum value. This formula not only integrates the intensity of multi-scale geometric evolution and directional stability, but also suppresses noise interference by dynamically adjusting the weights through the signal-to-noise ratio of the local point cloud, and uses higher-order derivatives to highlight the edge of the lesion, which is used to accurately identify early hidden damage.

[0035] S30. For abnormal regions where the distortion sensitivity index exceeds the threshold, extract the distribution entropy and local roughness of the histogram of the point cloud gradient magnitude, construct the initial degradation response intensity, and integrate the synchronously acquired surface reflection intensity variation coefficient as an environmental interference compensation term to generate the material degradation confidence factor.

[0036] S31. After identifying regions with high local surface distortion sensitivity index, the causes of geometric anomalies are analyzed by fusing point cloud reflection intensity and gradient features, distinguishing between deformation caused by structural stress and damage caused by physical degradation of surface materials.

[0037] Several high-value regions were identified based on the local surface distortion sensitivity index. These regions exhibit significant geometric evolution characteristics, and the causes of such geometric anomalies are multifaceted: they may originate from structural deformation caused by structural stress or be triggered by physical degradation of the surface material. To distinguish between these two causes, it is necessary to move beyond pure geometric analysis and incorporate reflection intensity information acquired simultaneously by laser scanning with point cloud spatial gradient features for fusion and discrimination. Reflection intensity reflects the energy of the laser pulse echo and is affected by factors such as the target surface material, roughness, moisture content, and incident angle. When the material surface deteriorates, diffuse reflection of the laser increases while specular reflection decreases, resulting in a reduced echo intensity and a more uneven distribution. Simultaneously, the point cloud gradient reflects the macroscopic undulations of the surface; material peeling is often accompanied by abrupt changes in local elevation or texture disorder.

[0038] Therefore, in areas with high sensitivity to local surface distortion, the corresponding reflection intensity value and point cloud gradient amplitude are extracted. By analyzing the joint distribution characteristics of these two values, it can be determined whether geometric anomalies are accompanied by signs of material porosity. If a region shows significant deformation but stable reflection intensity and orderly gradient distribution, it strongly suggests a structural response; conversely, if the deformed area simultaneously exhibits a significant decrease in reflection intensity, gradient disorder, or local abrupt changes, it strongly suggests that the surface material has undergone physical degradation.

[0039] S32. Based on the extracted gradient and intensity data, calculate the gradient distribution entropy and enhanced roughness to achieve quantitative identification of material degradation.

[0040] Based on the extracted point cloud gradient magnitude and reflection intensity, high-order feature calculations are performed on local regions to achieve refined identification of surface material degradation. The specific process is as follows: For each point to be analyzed, a local neighborhood window is constructed centered on that point. The gradient magnitudes of all points within this window are collected, and their normalized histogram distribution is constructed. Based on this distribution, the gradient distribution entropy is calculated using the following formula: ,in This represents the probability that the gradient value falls within the b-th interval. Entropy value. It reflects the complexity and disorder of local surface texture. When the material is uniform, the gradient distribution is concentrated and the entropy value is small. When deterioration such as peeling and powdering occurs, the surface undulation pattern is diverse, the gradient distribution is diffuse, and the entropy value increases significantly.

[0041] Based on the 3D (X,Y,Z) coordinates of each point in the point cloud, the z-values ​​of all points are extracted within a local neighborhood and treated as a 2D elevation field z(x,y). The basic roughness is then calculated based on z(x,y) within the same local window. The variance of the elevations of all points relative to the local mean plane is calculated as the basic roughness, using the following formula: , For local roughness variance, Let z be the z-coordinate of the j-th point within its local neighborhood. is the arithmetic mean of the z values ​​of all points in the local neighborhood, and N is the number of points in the local window.

[0042] Based on the basic roughness, an enhanced roughness variance is constructed to reflect the complex morphology caused by surface material degradation. The formula is as follows: ,in These are adjustable parameters that control the weights of the skewness correction term, kurtosis correction term, and the interaction term between the roughness gradient and curvature gradient, respectively. This is the absolute value of skewness, measuring the asymmetry of elevation distribution within a local area. Positive skewness indicates that local convexities dominate, while negative skewness reflects the dominance of depressions. Control the degree of nonlinearity. Kuroism reflects the thickness of the tail in a local area. High kuroism indicates the presence of many extreme high and low points, which is common on loose or corroded surfaces. The gradient magnitude of the basic roughness, This is the Gaussian curvature gradient.

[0043] Calculations are performed in areas with intact structures and uniform surfaces. mean with standard deviation As a benchmark for health status, These are slight undulations in the normal texture or construction marks. The surface roughness is moderate, indicating the presence of localized weathering or minor flaking. The surface is noticeably rough, strongly suggesting significant degradation such as mortar spalling, honeycomb corrosion, or bulging.

[0044] S33. To determine whether the high geometric distortion region is caused by the degradation of the surface material, a material degradation confidence factor is constructed based on texture entropy, enhanced roughness, and the coefficient of variation of reflection intensity as a criterion.

[0045] After completing the quantitative analysis of geometric distortion, surface roughness, and texture complexity, further discrimination of high-resolution textures is achieved. Determining whether a region's degradation is caused by physical degradation of the surface material requires a deep fusion of geometric features and material responses. However, these two types of information have different physical dimensions, nonlinear response characteristics, and environmental sensitivities, making simple weighting or thresholding insufficient to accurately characterize their coupling relationship. Therefore, to suppress environmental interference, highlight the true degradation signal, and achieve a nonlinear synergistic response of geometric and material features, a material degradation confidence factor C is constructed to quantitatively assess the credibility of surface degradation in a given region. The formula is as follows: , For gradient distribution entropy, For enhanced roughness variance, This is the interference suppression gain coefficient. The coefficient of variation of reflection intensity is the ratio of the standard deviation of reflection intensity to its mean in a local area. For the sensitivity parameter of the variation response, The mean and standard deviation of the gradient distribution entropy of the region background. is an offset constant used to adjust the activation threshold. Regions with significantly increased C values ​​are due to physical degradation of the surface material, rather than structural deformation or scanning noise.

[0046] S40. Based on the spatial rasterized distribution of confidence factors, calculate the degradation concentration coefficient for each structural component and couple it with the maintenance frequency in the historical inspection records. Map the structural weakness score through trapezoidal fuzzy membership degree, sort it in combination with the functional importance weight of the component, and output the component-level risk priority sequence.

[0047] S41. Based on the point cloud semantic segmentation results, map the confidence factor-based distribution map to the structural components and extract their internal degradation statistical features.

[0048] A spatial degradation factor distribution map is generated point by point based on the material degradation confidence factor.

[0049] Since inspection decisions typically revolve around specific structural components rather than scattered points, the conversion from points to components is crucial. Advanced semantic segmentation algorithms are employed to process the building's point cloud data to automatically identify and classify different structural components. The spatial degradation factor distribution map is divided into raster regions, and continuous C-value fields are mapped onto the boundaries of corresponding structural components. For each identified structural component, the spatial degradation statistics of all points within it are calculated, including the maximum value, average value, and standard deviation.

[0050] S42. Based on the statistical characteristics of component degradation, calculate the aging coupling term according to the degradation concentration coefficient and historical maintenance frequency, and realize the dynamic analysis of the health state of structural components.

[0051] For each identified structural component, define the ratio of the maximum value to the average value of the material degradation confidence factor within the component as the degradation concentration coefficient. When the value is about 1, the degradation distribution is uniform and the whole component is aged as a whole. When the value is greater than or equal to 1, the degradation is highly concentrated in local hot spots.

[0052] The degradation of structural components not only depends on the current state, but is also affected by historical use and maintenance activities. According to the historical maintenance frequency and the information of the last maintenance time, construct the comprehensive aging coupling term index formula: , is the degradation concentration coefficient, is the historical maintenance times, is the adjustment coefficient of the time term, which controls the relative weight of the maintenance timeliness in the overall aging assessment, is the current detection time, is the last maintenance time, is the time scale parameter.

[0053] S43. Transform the component aging coupling term into a standardized weakness degree score through a trapezoidal fuzzy membership function, and combine the function importance weight for weighted normalization to generate a component-level priority sequence arranged in descending order of risk.

[0054] To achieve the unified quantification and comparability of multi-source heterogeneous information, take the aging coupling term as the input, and map it to a weakness degree score within the interval [0,1] through designing a trapezoidal fuzzy membership function. This function uses four control points (a < b < c < d) to define the transition interval from normal to severely weak, allowing a complete membership platform to be set in the middle section, which can not only effectively express the fuzzy boundary of the degradation state but also avoid being dominated by extreme values. The closer the weakness degree score is to 1, the worse the structural integrity of the component. Subsequently, introduce the component function importance weight to weight the weakness degree score to reflect the safety level differences of different components in the structural system. Load-bearing columns, shear walls, etc. are key stressed components, secondary beams, floor slabs, etc. are general stressed components, and infill walls, partitions, etc. are non-structural components, with the importance weights decreasing in turn. Finally, calculate the weighted risk score and perform normalization processing, and arrange all components in descending order of the weighted risk score to generate a component-level risk priority sequence. High-risk components will be marked as the key inspection focuses preferentially.

[0055] S50. For high-risk components in the priority sequence, extract the deformation rate sequence of the most recent measurements, fit a linear trend and construct a historical rate cumulative score. Normalize this score and the weakness score in a dual-objective manner, solve for Pareto optimality through Euclidean distance, determine the list of specific parts with the highest comprehensive risk and mark their failure urgency level.

[0056] S51. By using multi-period material degradation and deformation rate sequences, linear trend fitting and acceleration sensitivity analysis are performed on high-risk components to comprehensively determine whether they are in an accelerated deterioration state.

[0057] To avoid missing rapidly developing but low-risk diseases due to static assessments based solely on a single test, dynamic evolution trend analysis needs to be conducted on the top N high-risk components in the priority sequence. This analysis, combined with multi-period monitoring data, aims to identify whether these components are in the early stages of accelerated deterioration or deformation instability, ensuring precise intervention timing. The specific process is as follows: Extract the time points of the three most recent periodic inspections for each high-risk component. The time interval is The two key indicators are material degradation sequence and geometric deformation sequence. The material degradation sequence includes parameters reflecting the surface physical state, such as material degradation confidence factor, enhanced roughness, and gradient entropy. The geometric deformation sequence is obtained by calculating the deformation rate by the ratio of displacement to time interval after multi-period point cloud registration. It characterizes the displacement change trend of the component.

[0058] Linear regression fitting was performed on the material degradation sequence and deformation rate to establish a function. ,in The values ​​represent the material degradation sequence and deformation rate, where 'a' is the fitted rate of change, 't' is the time variable representing the monitoring moment, and 'b' is the initial offset. This linear regression function is solved using the least squares method. If 'a' > 0, it indicates that the degradation deformation is intensifying.

[0059] The ratio of the current trend to the allowable value is used to measure the degree of deviation of the current degradation trend from the safe threshold of the standard. The formula is as follows: ,in This is the allowable deformation rate limit specified in the current structural health monitoring code for the type corresponding to component j. It approximates the expected change in deformation rate within a monitoring period. This is the multiple by which the deformation rate exceeds the allowable value in the next cycle. When this time comes, it indicates that the limit is about to be exceeded.

[0060] Meanwhile, to further capture sudden acceleration degradation, a second-order difference estimation of the acceleration response is introduced, and the acceleration sensitivity factor is calculated. This indicator is applicable to two types of sequences: material degradation sequences and geometric deformation sequences. It is sensitive to abrupt changes and can effectively identify whether a recent jump in growth has occurred.

[0061] If any of the following conditions are met, it is considered accelerated degradation: and This indicates that material degradation is accelerating. The rate of change of the material degradation sequence. As a sensitive factor for the acceleration of material degradation sequence, Sensitive threshold; and This indicates that the deformation trend is approaching the limit. The deformation rate is the rate of change of velocity. A significant surge, exceeding 50% compared to the previous period, indicates a sudden structural response.

[0062] S52. By constructing an exponentially decaying weighted historical rate cumulative score to quantify the deformation evolution trend, combined with the static weakness score, and through bi-objective normalization and Pareto front analysis, a list of components with the highest comprehensive risk and urgent failure is determined based on Euclidean distance ranking.

[0063] To more accurately integrate the historical deformation evolution trend of components with their current degradation state, and to overcome the shortcomings of traditional linear weighting in reflecting recent dynamic responses, a historical rate cumulative score is constructed to quantify the dynamic risk accumulation level of components. The formula is defined as follows: Where M is the number of measurements, Let be the exceedance ratio of component j in the kth measurement, representing the degree of exceedance relative to the allowable value in the specification, and used to unify the dimensions; It is a non-linear time decay weighting factor, ensuring that more recent data has a greater weight. Control the overall attenuation intensity, Introducing nonlinear memory effect, when The decay is slower over time, preserving more historical memory; when The decay is steeper over time, highlighting the latest changes. This is a rate jump enhancement term, used to amplify the effect of abrupt changes in deformation rate. The discrete difference, approximating acceleration, reflects the drastic change in the deformation rate. , Let be the deformation rates of component j in the k-th and (k+1)-th measurements, respectively. Controlling sensitivity to mutations, The time interval between measurements.

[0064] Calculate the historical rate cumulative score for each high-risk component. And call the weakness score in step S40. Normalize the ranges of both to the [0,1] interval to obtain , Projecting accelerated degradation components into a two-dimensional risk space. For any two components u and g in the set, if and If at least one dimension is strictly greater than a certain value, then component u is said to dominate component g. Iterating through all components, those not dominated by any other component are selected as Pareto optimal frontier components. The Euclidean distance from each frontier component to the ideal point (1,1) is calculated and sorted in ascending order. The component with the smallest distance is the one with the highest overall risk and is added to the list of specific risk locations. Based on the Euclidean distance results, the failure urgency level is classified as urgent, high, or medium. The list includes both structural and material aging problems of the risky component.

[0065] Example 2 like Figure 2 As shown, Embodiment 2 of this application provides a building inspection system based on an existing building structure, including: The structural deformation index module, based on point cloud data of building structure surfaces acquired through multiple high-precision laser scanning sessions, calculates the three-dimensional displacement characteristics and spatial autocorrelation index of the point cloud between adjacent measurements after joint registration with control points using the iterative nearest point method. A comprehensive structural deformation stability index is then generated using principal component fusion. Specifically, it is divided into the following sub-modules: Discreteness submodule: After multi-phase laser point cloud data are jointly registered with control points using the iterative nearest point method, the three-dimensional displacement modulus of the corresponding points is calculated, and a basic deformation stability index is constructed to quantify the discreteness of the overall structural deformation.

[0066] Spatial Clustering Submodule: Based on the K-nearest neighbor Gaussian kernel, a spatial weight matrix is ​​constructed to perform spatial autocorrelation analysis and the Moran index is calculated to quantify the spatial clustering of displacement values.

[0067] The comprehensive submodule integrates the two dimensions of displacement dispersion and spatial clustering. It uses principal component analysis to fuse the basic deformation stability index and the normalized Moran index to generate a comprehensive structural deformation stability index.

[0068] The distortion sensitivity index module identifies potential instability zones based on stability indices, performs multi-scale surface fitting on their local neighborhoods, extracts the evolution gradient of curvature and normal vector change rate, and introduces dynamic weights based on the local point cloud signal-to-noise ratio to generate a local surface distortion sensitivity index. Specifically, it is divided into the following sub-modules: Potential instability submodule: Based on the distribution of structural deformation stability index, regions below a preset threshold are selected as potential instability zones.

[0069] Direction recognition submodule: The surface of the abnormal region is reconstructed by fitting the multi-scale moving least squares method for the potential unstable region, the temporal change rate of Gaussian and average curvature is extracted, and the direction recognition term is calculated by combining the change rate of normal vector and the consistency of curvature gradient direction.

[0070] Local Distortion Submodule: Taking into account the local quality differences of point cloud, it introduces a signal-to-noise ratio dynamic adjustment weight based on curvature statistics, integrates multi-scale curvature change rate and directional stability, and combines the second-order curvature derivative to enhance edge response, thus constructing a local surface distortion sensitivity index.

[0071] Material Degradation Module: For abnormal regions where the distortion sensitivity index exceeds a threshold, the distribution entropy and local roughness of the point cloud gradient magnitude histogram are extracted to construct the initial degradation response intensity. The synchronously acquired surface reflectance intensity variation coefficient is then integrated as an environmental disturbance compensation term to generate a material degradation confidence factor. Specifically, it is divided into the following sub-modules: Differentiation prompt submodule: After identifying regions with high local surface distortion sensitivity index, the cause of geometric anomalies is analyzed by fusing point cloud reflection intensity and gradient features, distinguishing between deformation caused by structural stress and damage caused by physical degradation of surface materials.

[0072] Degradation identification submodule: Based on the extracted gradient and intensity data, it calculates the gradient distribution entropy and enhanced roughness to achieve quantitative identification of material degradation.

[0073] Confidence Factor Submodule: To determine whether regions with high geometrical distortion are caused by surface material degradation, a material degradation confidence factor is constructed based on texture entropy, enhanced roughness, and the coefficient of variation of reflection intensity as a criterion.

[0074] Priority sequence module: Based on the spatial rasterized distribution of confidence factors, it calculates the degradation concentration coefficient for each structural component and couples it with the maintenance frequency in historical inspection records. It then maps this coefficient to a structural weakness score using trapezoidal fuzzy membership, and sorts the components based on their functional importance weights, outputting a component-level risk priority sequence. Specifically, it is divided into the following sub-modules: Degradation statistics submodule: Based on the point cloud semantic segmentation results, the confidence factor-based distribution map is mapped to the structural components, and its internal degradation statistics features are extracted.

[0075] Aging Coupling Submodule: Based on the statistical characteristics of component degradation, the aging coupling term is calculated according to the degradation concentration coefficient and historical maintenance frequency to realize dynamic analysis of the health status of structural components.

[0076] The component sequence submodule transforms the component aging coupling term into a standardized weakness score using a trapezoidal fuzzy membership function, and then combines it with functional importance weights for weighted normalization to generate a component-level priority sequence arranged in descending order of risk.

[0077] Risk Location List Module: For high-risk components in the priority sequence, extract the deformation rate sequence from the most recent measurements, fit a linear trend, and construct a historical rate cumulative score. This score is then normalized with the weakness score using a bi-objective method. Pareto optimality is obtained through Euclidean distance to determine the specific locations with the highest overall risk and to label their failure urgency level. This module is further divided into the following sub-modules: Deterioration Judgment Submodule: By using multi-period material degradation and deformation rate sequences, linear trend fitting and acceleration sensitivity analysis are performed on high-risk components to comprehensively determine whether they are in an accelerated deterioration state.

[0078] The component list submodule quantifies the deformation evolution trend by constructing an exponentially decaying weighted historical rate cumulative score. Combined with the static weakness score, and after bi-objective normalization and Pareto front analysis, the component list with the highest comprehensive risk and urgent failure is determined based on Euclidean distance ranking.

[0079] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made on the basis of the technical solution of the present invention should be included within the scope of protection of the present invention.

Claims

1. A building inspection method based on existing building structures, characterized in that, include: S10. Based on the point cloud data of the building structure surface obtained by multi-phase high-precision laser scanning, after joint registration with control points by the iterative nearest point method, the three-dimensional displacement characteristics and spatial autocorrelation index of the point cloud between adjacent measurements are calculated, and a comprehensive structural deformation stability index is generated by the principal component fusion method. S20. Based on stability indices, identify potential unstable regions, perform multi-scale surface fitting on their local neighborhoods, extract the evolution gradient of curvature and normal vector change rate, and introduce local point cloud signal-to-noise ratio dynamic adjustment weights to generate a local surface distortion sensitivity index. S30. For abnormal regions where the distortion sensitivity index exceeds the threshold, extract the distribution entropy and local roughness of the histogram of the point cloud gradient magnitude, construct the initial degradation response intensity, and integrate the synchronously acquired surface reflection intensity variation coefficient as an environmental interference compensation term to generate the material degradation confidence factor. S40. Based on the spatial rasterization distribution of confidence factors, calculate the degradation concentration coefficient for each structural component and couple it with the maintenance frequency in the historical inspection records. Map the structural weakness score through trapezoidal fuzzy membership degree, sort it in combination with the component functional importance weight, and output the component-level risk priority sequence. S50. For high-risk components in the priority sequence, extract the deformation rate sequence of the most recent measurements, fit a linear trend and construct a historical rate cumulative score. Normalize this score and the weakness score in a dual-objective manner, solve for Pareto optimality through Euclidean distance, determine the list of specific parts with the highest comprehensive risk and mark their failure urgency level.

2. The building inspection method based on existing building structures as described in claim 1, characterized in that, Based on point cloud data of building structure surfaces acquired through multiple high-precision laser scanning sessions, the three-dimensional displacement characteristics and spatial autocorrelation index of the point cloud between adjacent measurements are calculated after joint registration with control points using the iterative nearest point method. A comprehensive structural deformation stability index is then generated using the principal component fusion method. The process is divided into the following sub-steps: After multiple laser point cloud data are jointly registered with control points using the iterative nearest point method, the three-dimensional displacement modulus of the corresponding points is calculated, and a basic deformation stability index is constructed to quantify the discreteness of the overall structural deformation. Spatial autocorrelation analysis was performed by constructing a spatial weight matrix based on the K-nearest neighbor Gaussian kernel, and the Moran index was calculated to quantify the spatial clustering of displacement values. By combining the two dimensions of displacement dispersion and spatial clustering, a comprehensive structural deformation stability index is generated by integrating the basic deformation stability index and the normalized Moran index through principal component analysis.

3. The building inspection method based on existing building structures as described in claim 1, characterized in that, Potential instability zones are identified based on stability indices. Multi-scale surface fitting is performed on their local neighborhoods to extract the evolution gradients of curvature and normal vector change rates. Furthermore, a local point cloud signal-to-noise ratio is introduced to dynamically adjust the weights, generating a local surface distortion sensitivity index. Specifically, this involves the following sub-steps: Based on the distribution of structural deformation stability index, regions below a preset threshold are selected as potential instability zones. The surface of the anomalous region is reconstructed by fitting the multi-scale moving least squares method for the potential unstable region, and the temporal change rate of Gaussian and mean curvature is extracted. The direction identification term is calculated by combining the change rate of normal vector and the consistency of curvature gradient direction. Taking into account the local quality differences of point clouds, a local surface distortion sensitivity index is constructed by introducing a signal-to-noise ratio dynamic adjustment weight based on curvature statistics, integrating multi-scale curvature change rate and directional stability, and combining the second-order curvature derivative to enhance edge response.

4. The building inspection method based on existing building structures as described in claim 1, characterized in that, For anomalous regions where the distortion sensitivity index exceeds a threshold, the distribution entropy and local roughness of their point cloud gradient magnitude histogram are extracted to construct an initial degradation response intensity. This is then fused with the synchronously acquired surface reflectance intensity variation coefficient as an environmental disturbance compensation term to generate a material degradation confidence factor. Specifically, this involves the following sub-steps: After identifying regions with high local surface distortion sensitivity index, the causes of geometric anomalies are analyzed by fusing point cloud reflection intensity and gradient features, distinguishing between deformation caused by structural stress and damage caused by physical degradation of surface materials. Based on the extracted gradient and intensity data, the gradient distribution entropy and enhanced roughness are calculated to achieve quantitative identification of material degradation. To determine whether regions with high geometric distortion are caused by surface material degradation, a material degradation confidence factor is constructed based on texture entropy, enhanced roughness, and the coefficient of variation of reflection intensity as a criterion.

5. A building inspection method based on existing building structures as described in claim 4, characterized in that, Based on the extracted gradient and intensity data, the gradient distribution entropy and enhanced roughness are calculated to achieve quantitative identification of material degradation. This is specifically divided into the following sub-steps: For each point to be analyzed, a local neighborhood window is constructed with that point as the center. The gradient magnitudes of all points are collected within this window, and their normalized histogram distribution is constructed. Based on this distribution, the gradient distribution entropy is calculated. Based on the three-dimensional coordinates of each point in the point cloud, the vertical direction values ​​of all points are extracted into a two-dimensional elevation field within a local neighborhood. The basic roughness is then calculated within the same local window based on the two-dimensional elevation field. Based on the basic roughness, an enhanced roughness variance is constructed to reflect the complex morphology caused by the degradation of the surface material.

6. The building inspection method based on existing building structures as described in claim 1, characterized in that, Based on the spatial rasterized distribution of confidence factors, a degradation concentration coefficient is calculated for each structural component and coupled with the maintenance frequency in historical inspection records. This coefficient is then mapped to a structural weakness score using trapezoidal fuzzy membership. Finally, the components are ranked according to their functional importance weights, outputting a component-level risk priority sequence. The specific steps are as follows: Based on the point cloud semantic segmentation results, the confidence factor-based distribution map is mapped to the structural components, and their internal degradation statistical features are extracted. Based on the statistical characteristics of component degradation, the aging coupling term is calculated according to the degradation concentration coefficient and historical maintenance frequency to realize dynamic analysis of the health status of structural components; The component aging coupling term is transformed into a standardized weakness score by using a trapezoidal fuzzy membership function, and then combined with functional importance weights for weighted normalization to generate a component-level priority sequence arranged in descending order of risk.

7. A building inspection method based on existing building structures as described in claim 1, characterized in that, For high-risk components in the priority sequence, the deformation rate sequence of the most recent measurements is extracted, a linear trend is fitted, and a historical rate cumulative score is constructed. This score is then normalized with the weakness score using a bi-objective method. Pareto optimality is obtained through Euclidean distance to determine the list of specific parts with the highest overall risk and to label their failure urgency level. The specific steps are as follows: By using multi-period material degradation and deformation rate sequences, linear trend fitting and acceleration sensitivity analysis are performed on high-risk components to comprehensively determine whether they are in an accelerated deterioration state. By constructing an exponentially decaying weighted historical rate cumulative score to quantify the deformation evolution trend, and combining it with the static weakness score, a list of components with the highest comprehensive risk and urgent failure is determined based on Euclidean distance ranking after bi-objective normalization and Pareto front analysis.

8. A building inspection system based on existing building structures, characterized in that, include: Structural Deformation Index Module: Based on point cloud data of building structure surface obtained by multi-phase high-precision laser scanning, after joint registration with control points by the iterative nearest point method, the three-dimensional displacement characteristics and spatial autocorrelation index of point cloud between adjacent measurements are calculated, and a comprehensive structural deformation stability index is generated by the principal component fusion method. Distortion Sensitivity Index Module: Based on stability indices, potential unstable regions are identified, multi-scale surface fitting is performed on their local neighborhoods, the evolution gradient of curvature and normal vector change rate is extracted, and the local point cloud signal-to-noise ratio is introduced to dynamically adjust the weights to generate a local surface distortion sensitivity index. Material degradation module: For abnormal regions where the distortion sensitivity index exceeds the threshold, the distribution entropy and local roughness of the point cloud gradient magnitude histogram are extracted to construct the initial degradation response intensity. The synchronously acquired surface reflection intensity variation coefficient is then fused as an environmental disturbance compensation term to generate the material degradation confidence factor. Priority sequence module: Based on the spatial rasterized distribution of confidence factors, the degradation concentration coefficient is calculated for each structural component and coupled with the maintenance frequency in historical inspection records. It is then mapped to the structural weakness score through trapezoidal fuzzy membership degree and sorted by the functional importance weight of the component to output a component-level risk priority sequence. Risk Location List Module: For high-risk components in the priority sequence, extract the deformation rate sequence of the most recent measurements, fit a linear trend and construct a historical rate cumulative score. Normalize this score and the weakness score in a dual objective manner, solve for Pareto optimality through Euclidean distance, determine the specific location list with the highest comprehensive risk and mark its failure urgency level.

Images