A safety detection method and device for curtain wall deformation conditions based on laser point cloud

Abstract

The present invention provides a safety detection method and device for the deformation condition of curtain walls based on laser point clouds, relating to the technical field of construction engineering. The method includes: Step 1, performing coordinate system normalization processing on the three-dimensional point cloud data set covering the entire surface of the curtain wall to generate a digital twin model of the curtain wall surface, and in the digital twin model, defining a first detection target point and a second detection target point according to the geometric characteristics and material joint characteristics of the curtain wall panels; Step 2, extracting the surface normal vector distribution characteristics according to the digital twin model, and fusing the measured wind pressure load distribution data and temperature field parameters to construct a dynamic flexural deformation coupling calculation model, and outputting global deformation parameters and local deformation parameters. The present invention constructs a digital twin model through point cloud data processing, extracts deformation and displacement parameters, and evaluates the safety factor, realizing high-precision, all-round, and dynamic detection of the flatness of the building exterior wall and safety risk early warning.

Description

Technical Field

[0001] The present invention relates to the technical field of construction engineering, and particularly to a safety detection method and device for curtain wall deformation based on laser point cloud. Background Art

[0002] Traditional methods for detecting the flatness of building exterior walls, such as using a straightedge for detection and total station measurement, have problems such as low detection efficiency, strong dependence on manual labor, and insufficient data accuracy, making it difficult to meet the detection requirements of modern large and complex curtain wall structures. Existing detection methods often only focus on the geometric shape changes on the surface of the curtain wall, simply obtaining the flatness value through point cloud data, but ignoring the dynamic effects of environmental factors such as wind pressure and temperature on the curtain wall structure under long-term action.

[0003] For example, when there is strong wind or drastic temperature change, the actual stress state and deformation trend of the curtain wall cannot be truly reflected by a single geometric parameter, resulting in deviation in the detection results and inability to accurately evaluate the safety status of the curtain wall.

[0004] In addition, in the process of processing and analyzing detection data, it is impossible to effectively integrate multi-source data information, making it difficult to achieve a comprehensive and dynamic assessment of the safety of the curtain wall structure. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a safety detection method and device for curtain wall deformation based on laser point cloud, to solve the problem of comprehensive dynamic assessment, and to improve the accuracy and reliability of detecting the flatness of building exterior walls.

[0006] To solve the above technical problems, the technical solution of the present invention is as follows:

[0007] In the first aspect, a safety detection method for curtain wall deformation based on laser point cloud, the method includes:

[0008] Step 1, perform coordinate system normalization processing on the three-dimensional point cloud data set covering the entire surface of the curtain wall to generate a digital twin model of the curtain wall surface, and in the digital twin model, define the first detection target point and the second detection target point according to the geometric characteristics and material joint characteristics of the curtain wall panels;

[0009] Step 2, extract the surface normal vector distribution characteristics according to the digital twin model, and fuse the measured wind pressure load distribution data and temperature field parameters to construct a dynamic flexural deformation coupling calculation model, and output the global deformation parameters and local deformation parameters;

[0010] Step 3: According to the local deformation parameters, through phase correlation analysis and wavelet transform, extract the vibration displacement components of the first detection target point and the second detection target point to construct the curtain wall vibration mode feature matrix. Meanwhile, analyze the displacement-time series data to identify the over-limit displacement events where the displacement amplitude is greater than or equal to the preset threshold.

[0011] Step 4: Determine the initial coefficient of the curtain wall structure safety degree according to the global deformation parameters and the occurrence frequency of the over-limit displacement events, and calculate the target correction value based on the local deformation parameters and the dynamic displacement components.

[0012] Step 5: Superimpose the target correction value on the initial coefficient of the safety degree to generate the corrected curtain wall structure safety degree coefficient, realize the dynamic adjustment of the initial coefficient of the safety degree, and output a three-dimensional detection report, including the risk level, the distribution of the corrected safety degree coefficient, and the warning mark associated with the target correction value.

[0013] In a second aspect, a safety detection device for the deformation condition of a curtain wall based on laser point cloud includes:

[0014] A data processing module, which is used to perform coordinate system normalization processing on the three-dimensional point cloud data set covering the entire surface of the curtain wall to generate a digital twin model of the curtain wall surface, and define the first detection target point and the second detection target point in the digital twin model according to the geometric characteristics and material joint characteristics of the curtain wall panels.

[0015] A model construction module, which is used to extract the surface normal vector distribution characteristics according to the digital twin model, and fuse the measured wind pressure load distribution data and the temperature field parameters to construct a dynamic flexural deformation coupling calculation model, and output the global deformation parameters and the local deformation parameters.

[0016] A feature extraction module, which is used to extract the vibration displacement components of the first detection target point and the second detection target point according to the local deformation parameters through phase correlation analysis and wavelet transform to construct the curtain wall vibration mode feature matrix. Meanwhile, analyze the displacement-time series data to identify the over-limit displacement events where the displacement amplitude is greater than or equal to the preset threshold.

[0017] A safety degree evaluation module, which is used to determine the initial coefficient of the curtain wall structure safety degree according to the global deformation parameters and the occurrence frequency of the over-limit displacement events, and calculate the target correction value based on the local deformation parameters and the dynamic displacement components.

[0018] A report generation module, which is used to superimpose the target correction value on the initial coefficient of the safety degree to generate the corrected curtain wall structure safety degree coefficient, realize the dynamic adjustment of the initial coefficient of the safety degree, and output a three-dimensional detection report, including the risk level, the distribution of the corrected safety degree coefficient, and the warning mark associated with the target correction value.

[0019] In a third aspect, a computing device includes:

[0020] One or more processors;

[0021] A storage device for storing one or more programs, which when executed by the one or more processors cause the one or more processors to implement the method described above.

[0022] In a fourth aspect, a computer-readable storage medium stores a program that, when executed by a processor, implements the method described above.

[0023] The above solution of the present invention has at least the following beneficial effects:

[0024] By performing coordinate system normalization on the three-dimensional point cloud data set to generate a digital twin model, the consistency and accuracy of the data are ensured, and the surface form of the curtain wall can be truly restored. Defining the detection target points based on the geometric characteristics of the curtain wall panels and the material joint characteristics avoids misjudgment caused by data deviation or improper selection of detection points. When constructing the dynamic flexural deformation coupling calculation model, the distribution characteristics of the surface normal vectors, the measured wind pressure load distribution data, and the temperature field parameters are integrated, changing the traditional mode of analyzing a single factor or a small number of factors. This comprehensive analysis of multiple factors can comprehensively consider various influences on the curtain wall during actual use, and the output global and local deformation parameters are more in line with the actual situation, thus more accurately reflecting the deformation status of the curtain wall.

[0025] Using phase correlation analysis and wavelet transform to extract the vibration displacement components of the detection target points can sensitively capture the subtle vibration characteristics of the curtain wall, and the constructed vibration mode feature matrix provides detailed data for analyzing the dynamic characteristics of the curtain wall. The analysis of the displacement-time series data can timely identify over-limit displacement events, providing a dynamic basis for evaluating the safety of the curtain wall and helping to discover potential safety hazards in advance. Determining the initial safety coefficient based on the global deformation parameters and the occurrence frequency of over-limit displacement events, and calculating the target correction value by combining the local deformation parameters and the dynamic displacement components realizes the scientific evaluation of the structural safety of the curtain wall. Adding the target correction value to the initial coefficient to generate a corrected safety coefficient can be dynamically adjusted according to the actual state of the curtain wall, making the evaluation result more in line with the safety performance of the curtain wall at different stages.

[0026] The output three-dimensional inspection report contains rich information such as risk levels, distribution of corrected safety coefficients, and warning marks, presented in an intuitive three-dimensional visualization form, facilitating managers and technicians to quickly understand the safety status of each part of the curtain wall. The warning marks are associated with the target correction value, which can timely prompt potential risk areas, enabling relevant personnel to quickly take measures for handling, reducing the probability of safety accidents, and ensuring the safety of personnel's lives and property and the normal use of the building. Brief Description of the Drawings

[0027] Figure 1 is a schematic flowchart of a method for safely detecting the deformation condition of a curtain wall based on laser point cloud provided by an embodiment of the present invention.

[0028] Figure 2 is a schematic diagram of a device for safely detecting the deformation condition of a curtain wall based on laser point cloud provided by an embodiment of the present invention. Detailed Description of the Embodiments

[0029] Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art.

[0030] As Figure 1 shown, an embodiment of the present invention proposes a method for safely detecting the deformation condition of a curtain wall based on laser point cloud, and the method includes the following steps:

[0031] Step 1, perform coordinate system normalization processing on the three-dimensional point cloud data set covering the entire surface of the curtain wall to generate a digital twin model of the curtain wall surface, and in the digital twin model, define a first detection target point and a second detection target point according to the geometric characteristics and material joint characteristics of the curtain wall panels;

[0032] Step 2, extract the surface normal vector distribution characteristics according to the digital twin model, and fuse the measured wind pressure load distribution data and temperature field parameters to construct a dynamic flexural deformation coupling calculation model, and output global deformation parameters and local deformation parameters;

[0033] Step 3, according to the local deformation parameters, extract the vibration displacement components of the first detection target point and the second detection target point through phase correlation analysis and wavelet transform to construct a vibration modal characteristic matrix of the curtain wall. At the same time, analyze the displacement-time series data to identify over-limit displacement events where the displacement amplitude is greater than or equal to a preset threshold;

[0034] Step 4, determine the initial coefficient of the curtain wall structure safety degree according to the global deformation parameters and the occurrence frequency of over-limit displacement events, and calculate the target correction value based on the local deformation parameters and dynamic displacement components;

[0035] Step 5, superimpose the target correction value and the initial coefficient of the safety degree to generate a corrected curtain wall structure safety degree coefficient, realize the dynamic adjustment of the initial coefficient of the safety degree, and output a three-dimensional detection report, including the risk level, the distribution of the corrected safety degree coefficient, and the warning mark associated with the target correction value.

[0036] In the embodiment of the present invention, a digital twin model of the curtain wall surface is generated by performing coordinate system normalization on the three-dimensional point cloud data set, and the first and second detection target points are defined based on the geometric characteristics of the curtain wall panels and the material joint characteristics, accurately anchoring the key detection areas and avoiding the blindness of traditional detection. Compared with the traditional method, it can capture the subtle deformation of the curtain wall more carefully, effectively overcome the problem of large manual measurement errors, and improve the accuracy and reliability of the detection data.

[0037] By integrating the measured wind pressure load distribution data and the temperature field parameters, a dynamic flexural deformation coupling calculation model is constructed, breaking through the limitation of the existing technology that only focuses on a single geometric deformation parameter. This model can simulate the dynamic effects of environmental factors such as wind pressure and temperature on the curtain wall, output global deformation parameters and local deformation parameters, truly restore the stress and deformation state of the curtain wall under complex environments, make the detection results more in line with the actual working conditions, effectively avoid the detection deviation caused by ignoring environmental factors, and comprehensively reflect the true safety state of the curtain wall.

[0038] Using phase correlation analysis and wavelet transform technology, the vibration displacement components of the detection target points are extracted to construct a curtain wall vibration mode feature matrix, and at the same time, combined with displacement-time series data to identify over-limit displacement events. This process not only realizes the accurate capture of the dynamic response of the curtain wall, but also can timely discover potential safety hazards. Compared with the traditional method that cannot detect dynamic characteristics and the insufficient analysis means of the existing technology, it enhances the monitoring ability of the curtain wall operation state and provides a strong guarantee for preventing accidents.

[0039] Based on the global deformation parameters and the occurrence frequency of over-limit displacement events, the initial safety coefficient is determined, and the target correction value is calculated by combining the local deformation parameters and the dynamic displacement components to dynamically adjust the safety coefficient. This systematic evaluation method effectively integrates multi-source data information. Compared with the defects of the traditional detection lacking an evaluation system and the existing technology being difficult to comprehensively and dynamically evaluate, it can quantitatively evaluate the structural safety of the curtain wall from a global perspective, generate a three-dimensional detection report including risk levels, safety coefficient distributions, and warning marks, provide an intuitive, comprehensive, and scientific decision-making basis for curtain wall maintenance management, and improve the efficiency and accuracy of curtain wall maintenance management.

[0040] In a preferred embodiment of the present invention, in step 1 above, the three-dimensional point cloud data set covering the entire surface of the curtain wall is subjected to coordinate system normalization to generate a digital twin model of the curtain wall surface, and in the digital twin model, according to the geometric characteristics of the curtain wall panels and the material joint characteristics, the first detection target point and the second detection target point are defined; the first detection target point is located in the four corner regions of the curtain wall panel for monitoring the corner stress concentration effect; the second detection target point is located at the junction of the center and the edge of the curtain wall panel for monitoring the temperature difference deformation gradient change, which may include:

[0041] In the embodiments of the present invention, a three-dimensional point cloud dataset covering the entire surface of the curtain wall is obtained, and this task is completed by means of a laser scanning device (such as a three-dimensional laser scanner). The laser scanning device is placed at an appropriate position to scan the curtain wall in all directions. During the scanning, the device emits laser beams and measures the time it takes for them to reflect back, thereby obtaining the three-dimensional coordinates of numerous points on the surface of the curtain wall, and finally forming a three-dimensional point cloud dataset. This dataset covers rich geometric information on the surface of the curtain wall, but due to factors such as the position and attitude of the scanning device, the point cloud data of each part is in different coordinate systems. Since the obtained three-dimensional point cloud dataset comes from different scanning positions or devices, and there are differences in the coordinate systems of the point cloud data of each part, in order to be able to uniformly process and analyze these data, coordinate system normalization processing is required. The specific steps are as follows:

[0042] Extract representative feature points or feature surfaces in the three-dimensional point cloud dataset, such as the corner points and edge lines of the curtain wall. These features will be used as the reference basis for coordinate transformation. Through the iterative closest point algorithm, find a suitable rotation and translation transformation matrix to transform the point cloud data in different coordinate systems into the same coordinate system. This can ensure that all point cloud data is processed in the same reference system. After completing the coordinate system normalization processing, a digital twin model of the curtain wall surface can be generated based on the three-dimensional point cloud dataset in the unified coordinate system. The digital twin model is a digital mapping of the real curtain wall and can accurately reflect the geometric shape and structural information of the curtain wall.

[0043] Specifically, remove the noise points and outliers in the point cloud data to improve the quality and accuracy of the data. Use the point cloud data to reconstruct the surface model of the curtain wall. The surface reconstruction methods include the moving least squares method. Through these methods, the discrete point cloud data can be converted into a continuous surface model to form a digital twin model of the curtain wall surface. The first detection target points are located in the four corner regions of the curtain wall panels and are used to monitor the corner stress concentration effect. The specific process of determining these target points is as follows:

[0044] Divide the point cloud data into regular three-dimensional voxel grids, and realize data downsampling by calculating the average value of the points in each voxel, reducing the data volume while retaining the features. For each point in the point cloud data, construct a covariance matrix in its neighborhood. Perform eigenvalue decomposition on the covariance matrix, and the eigenvector corresponding to the minimum eigenvalue is the normal vector of the marked point. Use the average curvature algorithm to calculate the curvature value of each point in the point cloud , where, and are two eigenvalues of the covariance matrix. The curvature value >set threshold and the curvature change rate of adjacent points is greater than or equal to The marked area is the potential corner area. The curvature change rate of adjacent points The calculation formula is , where is the curvature value of a point within the neighborhood of point , is the distance between two points. By applying the edge detection algorithm, the edge intensity and direction of the point cloud are calculated to identify the edge contour of the slab. The edge intensity , where and are the partial derivatives of the curvature value in the and directions respectively. The edge intersection area is also included in the potential corner area.

[0045] Cluster analysis is performed on all potential corner areas. Clustering algorithms such as DBSCAN are used to merge overlapping areas, and finally the four corner areas of the curtain wall slab are determined. Within each corner area, a spherical range with a radius of is delimited with the regional vertex as the center. In this range, 3 - 5 points that are closest to the vertex and have the largest curvature change rate are selected as the first detection target points. Let the distance from point to the vertex be , , where,( ) and( ) are the coordinates of point and vertex respectively.

[0046] The second detection target point is located at the junction of the center and the edge of the curtain wall slab to monitor the change rate of temperature difference deformation gradient. The specific process of determining these target points is as follows:

[0047] According to the geometric shape and size of the curtain wall slab, the geometric center of the slab is calculated. If the point cloud data of the slab is , then the coordinate calculation formula of the geometric center is , where is the total number of points in the point cloud data of the curtain wall slab, represents the coordinate of the th point in the point cloud data of the curtain wall slab. By constructing a Voronoi diagram to divide the point cloud data of the slab, combined with the distance from the point cloud to the geometric center and the shortest distance from the point cloud to the edge, points within the distance from the slab center within the slab radius 0.6 - 0.8 times, and the distance to the edge is less than a certain threshold Points are determined as the boundary region between the center and the edge. That is, it satisfies: And . Perform grid processing on the point cloud in the boundary region, and divide the region into small grids. Within each small grid, calculate the temperature sensitivity coefficient of each point . According to the type of material used in the curtain wall, the thermal expansion coefficient of building curtain wall materials such as aluminum alloy is about (23.6× - 24.0× ) / ℃. The thermal expansion coefficient of glass varies depending on the type of glass. The thermal expansion coefficient of ordinary flat glass is approximately (8× - 10× ) / ℃, while the thermal expansion coefficient of low-expansion glass-ceramics can be as low as close to 0 / ℃. Determine the accurate value of the material thermal expansion coefficient by obtaining the specific parameters of the curtain wall material, and perform weighted calculation in combination with the point cloud normal vector direction , that is where is the unit vector in the temperature change direction.

[0048] represents the point cloud normal vector direction. Specifically, is the direction vector pointed by the normal vector of the corresponding point in the point cloud data. The normal vector is a vector with magnitude and direction. Here emphasizes the property of direction. Through the cross product operation with the unit vector in the temperature change direction, and in combination with the material thermal expansion coefficient , calculate the temperature sensitivity coefficient . Such a calculation method can take into account the relationship between the point cloud normal vector direction and the temperature change direction, and thus reflect the sensitivity differences of points at different positions to the deformation of the plate under temperature changes.

[0049] After calculating the temperature sensitivity coefficient of each point in each small grid, select the point with the maximum temperature sensitivity coefficient as the second detection target point, so as to ensure that the selected point can sensitively capture the subtle changes in the deformation gradient of the plate under different temperature difference conditions, and thus effectively monitor the influence of temperature difference on the flatness of the curtain wall plate.

[0050] In a preferred embodiment of the present invention, in step 2 above, according to the digital twin model, the surface normal vector distribution characteristics are extracted, and the measured wind pressure load distribution data and the temperature field parameters are fused to construct a dynamic flexural deformation coupling calculation model, and the global deformation parameters and local deformation parameters are output, which may include:

[0051] Step 200: According to the surface point cloud data of the digital twin model, the curtain wall surface is divided into grids and blocks, and for the point cloud data in each divided area, the current target point and the adjacent point set within a preset radius in each block are extracted;

[0052] Step 201: Fit the spatial plane of the adjacent point set by the least square method, calculate the surface normal vector of each target point, and aggregate the normal vectors of each block to generate the surface normal vector distribution characteristics of the curtain wall;

[0053] Step 202: Generate the measured wind pressure load distribution data by mapping the wind pressure sensor data around the building curtain wall to the corresponding spatial coordinates of the digital twin model; according to the data collected by the temperature sensor and the building structure heat transfer parameters, generate the temperature field parameters through finite element heat transfer analysis, including the temperature spatial distribution matrix and the time-varying gradient sequence;

[0054] Step 203: Fuse the surface normal vector distribution characteristics of the curtain wall, the measured wind pressure load distribution data and the temperature field parameters to construct a dynamic flexural deformation coupling calculation model;

[0055] Step 204: Input the wind pressure load distribution data, the temperature field parameters and the material elastic modulus into the dynamic flexural deformation coupling calculation model, and set the fixed constraint conditions for the connection nodes between the curtain wall and the main structure;

[0056] Step 205: Perform dynamic flexural deformation analysis according to the fixed constraint conditions, the wind pressure load distribution data, the temperature field parameters and the material elastic modulus, and extract the global deformation parameters and local deformation parameters; the global deformation parameters include the overall waviness and the regional curvature anomaly value; the local deformation parameters include the corner stress gradient of the first detection target point and the peak value of the dynamic flexural displacement at the edge-center junction of the second detection target point.

[0057] In the embodiment of the present invention, based on the surface point cloud data of the digital twin model, the overall range and size of the curtain wall surface are first determined. According to the preset grid resolution (for example, dividing the curtain wall surface into square grids with a side length of 0.5 meters), the curtain wall surface is divided into multiple regular sub-regions using a spatial grid algorithm (such as octree partitioning or uniform grid partitioning). Each sub-region forms an independent calculation unit. For the point cloud data within each sub-region, target points are selected according to certain rules (which can be all points within the sub-region, or a representative point is selected at a certain distance interval). For each target point, with this point as the center and according to the preset radius (such as 0.3 meters), a spatial search algorithm (such as KD-Tree search) is used to find all neighboring points within the radius range in the sub-region, thereby constructing the neighboring point set of each target point. This neighboring point set contains the point cloud information of the local area around the target point.

[0058] Step 201, for each target point and its corresponding neighboring point set, take the three-dimensional coordinates of the neighboring points ( , , ) as input data and perform spatial plane fitting using the least squares method. The goal of the least squares method is to find a plane equation + + + = 0, such that the sum of the squares of the distances from the neighboring points to this plane is minimized. By solving the corresponding linear equations (matrix equations constructed based on the neighboring point coordinates), the coefficients , , , of the plane equation can be obtained. According to the coefficients of the plane equation, calculate the normal vector of this plane. The normal vector of the plane can be expressed as ( ), and normalize it (that is, divide by the modulus length ) to obtain the unit normal vector, and this unit normal vector is the surface normal vector of the target point. Repeat the above operations for all target points within the sub-region to obtain the surface normal vector of each target point.

[0059] Aggregate the surface normal vectors of all target points within the sub-region. Statistical methods can be used, such as calculating the average of the normal vectors, to obtain the representative normal vector of this sub-region. After performing the same processing on all sub-regions, integrate the representative normal vectors of each sub-region to generate the normal vector distribution characteristics of the entire curtain wall surface, and this characteristic reflects the change of the normal direction at different positions on the curtain wall surface.

[0060] Step 202, according to the structural characteristics and design requirements of the building curtain wall, reasonably arrange wind pressure sensors at key positions such as the windward side, corners, top, and bottom of the curtain wall. For example, for a high-rise curtain wall building with a rectangular plane, sensors can be evenly arranged at the four corners and the middle area of each floor to ensure coverage of different heights and positions of the curtain wall. These sensors collect the magnitude and direction data of wind pressure in real time at a fixed sampling frequency (such as once per second) and transmit the data to the data processing center via wireless communication. The data collected by each sensor includes a timestamp, wind pressure value (unit: Pascal, Pa), wind pressure direction (expressed in degrees, e.g., 0° represents the due east direction), and the unique identification ID of the sensor itself.

[0061] Obtain the actual spatial coordinates of each wind pressure sensor, which can be determined by high-precision GPS positioning or measurement based on the building design drawings, and are represented using a three-dimensional Cartesian coordinate system ( 、 、 axes). At the same time, in the digital twin model, there is also a corresponding coordinate system. To match the sensor coordinates with the digital twin model coordinates, it is necessary to determine the coordinate transformation relationship between the two. If the coordinate systems of the two are inconsistent, operations such as translation, rotation, and scaling may be required. For example, through measurement, it is found that the origin of the coordinate system of the digital twin model is offset by 5 meters in the axis direction and 3 meters in the axis direction, and there is a rotation angle of 15°. Then, the sensor coordinates need to be adjusted according to the corresponding mathematical transformation formula to unify them with the digital twin model coordinates.

[0062] For each point or block area (assuming the block area is a square with a side length of 0.5 meters) in the digital twin model, determine the set of wind pressure sensors within a certain range (such as a radius of 5 meters) around it. For each sensor in the set, calculate the Euclidean distance from the sensor to the target point . Then, according to the inverse distance weighted interpolation principle, calculate the wind pressure value at the target point . Suppose there are sensors, and the wind pressure value of each sensor is , then The calculation formula of = , where is the weight exponent, taking 2, is the index, refers to the The Euclidean distance from each wind pressure sensor to the target point (i.e., the point in the digital twin model where the wind pressure value needs to be calculated). In this formula, the closer the sensor is to the target point, the greater the weight of its wind pressure value in calculating the wind pressure at the target point. For the wind pressure direction, a similar weighted average method is also used for calculation to obtain the wind pressure direction at the target point. After calculating the wind pressure values for all points or sub-regions, the generated measured wind pressure load distribution data is verified. Some known positions that did not participate in the interpolation calculation can be selected (such as positions where additional high-precision wind pressure measurement devices are installed), and the calculated wind pressure values are compared with the actual measured values to calculate the error. If the error exceeds the preset threshold (such as 5%), the reasons are analyzed, which may be unreasonable sensor layout, inaccurate coordinate transformation, or improper parameter settings of the interpolation algorithm, etc. According to the analysis results, adjust the sensor layout, re-check the coordinate transformation relationship, or optimize the interpolation algorithm parameters (such as adjusting the weight index ), and then re-perform the interpolation calculation until the accuracy requirements are met. Finally, accurate and reliable measured wind pressure load distribution data for the entire curtain wall surface is generated, which can detail the magnitude and direction distribution of wind pressure at different positions on the curtain wall surface.

[0063] Collect temperature data using temperature sensors distributed on the curtain wall surface and inside the building, and simultaneously obtain the heat transfer parameters of the building structure (such as the thermal conductivity, specific heat capacity, heat convection coefficient, etc. of the materials). Using these data as inputs, establish a heat transfer analysis model using finite element heat transfer analysis software (such as ANSYS, ABAQUS, etc.). In the model, divide the grid according to the geometric shape and material properties of the building structure, and set the boundary conditions (such as ambient temperature, solar radiation, etc.). By solving the heat transfer equation (such as Fourier's heat conduction equation), perform finite element calculations to obtain the temperature field parameters. The temperature field parameters include the temperature spatial distribution matrix (describing the temperature values at different positions on the curtain wall surface at a certain moment) and the time-varying gradient sequence (describing the change trend and gradient of the curtain wall surface temperature over time).

[0064] Step 203, fuse the normal vector distribution characteristics of the curtain wall surface generated in Step 201, the measured wind pressure load distribution data generated in Step 202, and the temperature field parameters. During the fusion process, first perform format conversion and normalization processing on these data to make them have a unified data format and dimension. Then, based on mechanical principles and numerical calculation methods (such as the finite element method), establish a mathematical model to describe the interaction relationship between wind pressure, temperature, and the curtain wall structure. In the model, use the normal vector distribution characteristics to determine the force direction on the curtain wall surface, use the wind pressure load distribution data as the external load input, and use the temperature field parameters to consider the thermal stress caused by temperature changes. By integrating these factors, construct a dynamic flexural deformation coupling calculation model, which can simulate the dynamic flexural deformation process of the curtain wall under the combined action of wind pressure and temperature.

[0065] Step 204: Input the measured wind pressure load distribution data, the calculated temperature field parameters, and the elastic modulus of the curtain wall materials into the dynamic flexural deformation coupling calculation model. The wind pressure load distribution data is used to apply the external wind load, the temperature field parameters are used to consider the influence of temperature changes on the curtain wall structure, and the material elastic modulus reflects the mechanical properties of the curtain wall materials. At the same time, according to the actual connection method between the curtain wall and the main structure, set the fixed constraint conditions of the connection nodes in the model. For example, if the curtain wall is connected to the main structure by bolts, apply fixed displacement constraints (i.e., restrict the displacements of the nodes in three directions) at the corresponding connection node positions in the model to accurately simulate the stress state of the curtain wall under actual working conditions.

[0066] Step 205: Define the fixed constraint conditions, which are determined based on the actual connection method between the curtain wall and the building main body. For example, at the bolt connection parts, it is necessary to set the displacement limitations in space at this position in the analysis to simulate its fixed characteristics; for the hinged connection nodes, set that they can rotate in a specific direction and the displacements in other directions are restricted.

[0067] The wind pressure load distribution data is collected in real time by the wind pressure sensors installed on the curtain wall surface, recording the wind pressure magnitude and direction information at different times and positions. The temperature field parameters are obtained through the laser point cloud scanning equipment, which can accurately present the spatial distribution of the curtain wall surface temperature and its change over time. The material elastic modulus is determined according to the material properties of the curtain wall materials (such as glass, metal frames) and referring to the material standards, which reflects the ability of the material to resist elastic deformation.

[0068] Obtaining the overall waviness:

[0069] Determine the grid specifications according to the actual size of the curtain wall and the detection accuracy requirements. For example, for a curtain wall with a length of 50 meters and a height of 30 meters, if it is selected to be divided into a 10×10 grid, then the size of each grid in the length direction is 5 meters, and in the height direction is 3 meters. During the division process, take the lower left corner vertex of the curtain wall as the coordinate origin (0 0 0), along the horizontal right direction as the positive direction of the axis, the vertical upward direction as the positive direction of the axis, and the direction perpendicular to the curtain wall surface and outward as the positive direction of the axis to establish a three-dimensional coordinate system. By equally spacing the coordinate intervals of the axis and the axis, determine the boundary positions of each grid to ensure that the grid evenly covers the entire curtain wall surface. Use a three-dimensional laser scanner to scan the curtain wall in all directions. The scanner emits laser beams, and after the laser encounters the curtain wall surface, it is reflected back to the instrument. By measuring the round-trip time of the laser, calculate the distance from each point on the curtain wall surface to the scanner, and combine the spatial position and attitude information of the scanner to determine the three-dimensional coordinates of each measurement point ( ), thus forming point cloud data containing a large number of points. During the scanning process, to ensure data integrity, it is necessary to control the scanning angle and movement path of the scanner to ensure that all parts of the curtain wall, including corners, concave and convex areas, can be scanned. After the scanning is completed, the point cloud data is classified according to the pre-divided grid areas, and the point cloud data within each grid is extracted to form an independent point cloud data set.

[0070] For the point cloud data set of each grid area, the least squares method is used for plane fitting. The core idea of the least squares method is to find a plane such that the sum of the squares of the perpendicular distances from all the point clouds within the grid to this plane is minimized. During the specific operation, first assign initial values to the coefficients in the plane equation + + + =0. Assign 、 、 as relatively small non-zero random numbers, such as taking values between -1 and 1. This can avoid special situations for the initial plane (such as being parallel to the coordinate axes), and at the same time provide a reasonable starting point for subsequent optimization; is initially set to 0 to simplify the initial calculation. By continuously adjusting the coefficients 、 、 、 values, calculate the perpendicular distance from each point cloud to this plane, and sum the squares of the perpendicular distances of all point clouds. During the process of adjusting the coefficients, use mathematical optimization algorithms (such as the gradient descent method), according to the change trend of the objective function (the sum of the squares of the perpendicular distances), and gradually change the coefficient values according to certain rules (such as the gradient direction and step size). For example, when using the gradient descent method, the amount of each coefficient update is related to the partial derivative of the objective function with respect to this coefficient and the set step size. The step size is a key parameter. If the value is too large, it may cause excessive adjustment of the coefficients and miss the final solution; if the value is too small, the calculation convergence speed will be too slow. In actual operation, the step size will first be set to an empirical value (such as 0.01) and dynamically adjusted according to the change of the objective function during the calculation process.

[0071] Continuously repeat the process of adjusting the coefficients and calculating the objective function value to find the 、 、 、 combination that makes this sum of squares reach the minimum value. When the stop condition is met (such as reaching the preset maximum number of iterations), the plane determined by the coefficient combination at this time is the plane representing the benchmark of this grid area.

[0072] After determining the fitting plane, calculate the perpendicular distance from each point cloud within the grid to this plane. The specific method is to use the three-dimensional coordinates of the point cloud ( ) Substitute the distance formula from a point to a plane into , obtaining the deviation value of each point cloud relative to the fitted plane. These deviation values can be positive or negative, with positive values indicating that the point cloud is above the plane and negative values indicating that the point cloud is below the plane. The absolute value reflects the degree to which the point cloud deviates from the plane. After calculating the deviation values for all point clouds, these deviation values are analyzed. You can first calculate the average of the deviation values to understand the overall deviation trend of the point cloud in the grid area relative to the fitted plane. Then, calculate the standard deviation of the deviation values to measure the degree of dispersion of the point cloud data relative to the average value. A larger standard deviation indicates a more dispersed distribution of the point cloud within the grid area and a more uneven surface.

[0073] Regional curvature outlier determination:

[0074] After using a 3D laser scanner to collect point cloud data of a curtain wall surface, the data may contain noise and outliers due to measurement errors, environmental interference, and other factors. First, noise points are identified and removed through statistical analysis. For example, the distance between each point and its neighbors is calculated. If the average distance from a point to its neighbors is significantly greater than that of other points, it is considered a noise point. For outliers, a density-based clustering algorithm can be used to identify and remove points in areas with lower density. Before fitting a NURBS surface, the number and distribution of control points for the surface must be determined. Based on the complexity and accuracy requirements of the local area of the curtain wall, a grid of control points is optimally arranged within the bounding box of the point cloud data. For areas with more complex shapes, such as corners or decorative protrusions, the density of control points is increased; for relatively flat areas, the number of control points is appropriately reduced. The weights of the NURBS surface are determined based on the point cloud data and the specified control points. The weights affect the fit between the surface and the point cloud data. An empirically pre-assigned value is used to assign different initial values to the weights of control points in different areas, depending on the point cloud density and the general shape of the surface. For example, at corners with dense point clouds, the initial value of the weight factor is set to 1.2-1.5, so that the surface is more inclined to be close to the point cloud data in this area during initial construction; in relatively flat areas, the initial value of the weight factor is set to 0.8-1 to avoid excessive fitting of the surface and resulting in shape distortion.

[0075] Taking the least squares method as an example for optimization, the sum of the squares of the distances from the point cloud to the surface is used as the objective function. During the optimization process, the weight factor is adjusted according to the changing trend of the objective function. Specifically, when adjusting, the weight factor is changed in accordance with a certain step size. The initial value of the step size is set to 0.1. This value can not only ensure obvious changes in the weight factor but also prevent missing the optimal solution due to overly large adjustment amplitudes. During the calculation process, the change of the objective function value is monitored in real time. If the objective function value drops significantly, it indicates that the current step size and adjustment direction are appropriate, and continue to adjust with this step size; if the objective function value drops slowly, then reduce the step size to 0.05 or even smaller to ensure the fineness of the adjustment; if the objective function value rises, then change the adjustment direction and appropriately reduce the step size to readjust. Set double termination conditions. One is to set the maximum number of iterations, such as 200300 times. When the number of iterations reaches this upper limit, regardless of whether the objective function value converges, stop the calculation; the other is to set the change threshold of the objective function value, such as 0.001. When the difference between the objective function values calculated in two adjacent iterations < this threshold, it is considered that the surface has sufficiently approximated the point cloud data. The finally obtained combination of weight factors is the value that can make the NURBS surface fit the point cloud data as much as possible and is used to construct an accurate local surface model of the curtain wall.

[0076] During the fitting process, it is also necessary to determine the degree of the B-spline basis function. The higher the degree of the basis function, the smoother the surface, but the higher the computational complexity. Select an appropriate degree according to actual needs. For example, for a relatively smooth surface such as the curtain wall surface, select the cubic B-spline basis function, which can not only ensure the smoothness of the surface but also effectively control the amount of calculation. By adjusting the control points, weight factors, and the degree of the basis function multiple times and performing fitting repeatedly, an NURBS surface model that can accurately reflect the local surface shape of the curtain wall is finally obtained. For the generated NURBS surface model, a geometric analysis method is used to calculate the curvature values of each point on the surface. First, determine the parametric coordinates of the points on the surface where the curvature is to be calculated. By evenly dividing the grid on the surface, a series of points are selected as the calculation objects. At each selected point, calculate the first derivative and the second derivative of the surface. The first derivative reflects the tangential direction of the surface at this point, and the second derivative is related to the degree of curvature of the surface. By performing derivative operations on the parametric equations of the NURBS surface, these derivative information are obtained.

[0077] Based on the first derivative and the second derivative, further calculate the normal vector of the surface at this point. The normal vector is perpendicular to the surface. Let the first partial derivatives of the surface be and , and the second partial derivatives be , and , then the unit normal vector of the surface is . Using the normal vector and the derivative information, calculate the mean curvature where , , , is the dot product of the first-order partial derivative of the curved surface in the parameter direction with itself, and is the dot product of the first-order partial derivative of the curved surface in the parameter direction and the first-order partial derivative in the parameter direction. Through this formula, the curvature values of each point on the NURBS curved surface can be obtained, so as to evaluate the bending degree of different positions of the curved surface. is the dot product of the first-order partial derivative of the curved surface in the parameter direction with itself. By this formula, the curvature values of each point on the NURBS surface can be obtained, thus evaluating the bending degree of different positions of the surface.

[0078] According to the design drawings and material mechanical properties of the curtain wall, combined with engineering experience and relevant standard specifications, a reasonable curvature range is preset as the judgment standard. Taking a 6mm thick tempered glass flat curtain wall as an example, based on its elastic modulus of 72GPa and the span deformation allowed by the design, through finite element simulation and engineering practice verification, the normal curvature range is set to 0.001 - 0.003 . When comparing the curvature values of each point calculated on the curved surface, in order to accurately identify outliers, a threshold needs to be set. Considering the error of the measuring device ±0.3mm (such as the accuracy of the laser scanner), the fluctuation range of the elastic modulus of the glass material ±5%, and the process error in construction and installation, the threshold is set to 20% of the upper limit of the standard range. Specifically, when the curvature value of a certain point exceeds 0.0036 (i.e., 0.003 ×1.2), it is determined as an outlier of the regional curvature. This threshold setting mechanism can effectively avoid misjudgment caused by measurement errors or small fluctuations in material properties. Once an outlier appears, it indicates that there is bending deformation beyond the normal range in the corresponding local area of the curtain wall, which may be due to fatigue damage caused by long-term wind load on the structure, or deviation from the design accuracy requirements during installation, and targeted detailed inspection and mechanical analysis need to be carried out immediately.

[0079] Calculation of the corner stress gradient of the first detection target point:

[0080] According to the stress distribution results obtained from the finite element calculation model, locate the first detection target point and its adjacent points in the corner area of the curtain wall panel. The selection of adjacent points can be based on actual needs. Taking the target point as the center, set a spherical or cubic region, and extract the stress values of all nodes in this region. The stress values include normal stress and shear stress. Use the finite difference method to calculate the stress in , ,​​​​​​​ The partial derivative in the direction. Taking direction as an example, for the target point and its neighboring point in the and , the partial derivative of the stress in the direction is approximately , where is the distance between the two points in the direction. Similarly, the partial derivatives in the and directions are and . Through the partial derivatives in the three calculated directions, according to vector operations, the corner stress gradient vector is obtained. The magnitude of the stress gradient reflects the degree of stress concentration in the corner area; the direction of the stress gradient indicates the trend of stress change.

[0081] Calculation of the peak value of the dynamic flexural displacement at the edge - center junction of the second detection target point:

[0082] Locate the second detection target point at the center - edge junction of the curtain wall panel from the database of finite element calculation results. At each calculation time step, extract the displacement component data of this target point in the , , three directions, and simultaneously obtain the corresponding timestamps. For example, if the finite element calculation uses 0.1 second as a time step, at the 10th time step, the displacement of the target point in the direction is 2 mm, direction is -1.5 mm, direction is 0.8 mm, and the timestamp is 1 second. Organize this type of data in chronological order and store it in the form of a multi - dimensional array. Use the linear search algorithm to traverse the stored displacement data array. Set the initial maximum displacement value as the combined displacement value of the first time step. The combined displacement calculation formula is , where , , are the three - direction displacement components of the first time step. Starting from the second time step, calculate the combined displacement for each time step and compare it with the current maximum displacement value. After traversing the entire array, the finally obtained maximum combined displacement value is the peak value of the dynamic flexural displacement, and the corresponding time point can reflect the moment when the maximum deformation occurs, so as to evaluate the maximum deformation situation at the junction due to temperature difference and wind force.

[0083] Suppose there is a high-rise office building with a height of 100 meters and 30 floors, and its facade adopts a glass curtain wall structure. When conducting the flatness inspection of the curtain wall:

[0084] Divide the point cloud data on the curtain wall surface in the digital twin model into grids of 1 meter × 1 meter, and a total of 5000 sub-block areas are obtained. For the point cloud data in each sub-block area, select the point at the center of the sub-block as the target point, and use a preset radius of 0.2 meters to find the adjacent point set centered on the target point. For example, in a certain sub-block area, 15 adjacent points are found around the target point. For this target point and its adjacent point set, use the least squares method to fit the spatial plane, obtain the plane equation 2x + 3y - 4z + 5 = 0, and calculate and normalize to obtain the surface normal vector of the target point as (0.37, 0.55, -0.74). After performing the same processing on the target points of all sub-block areas, aggregate to obtain the distribution characteristics of the surface normal vector of the curtain wall.

[0085] A total of 20 wind pressure sensors are installed on the four facades of the building curtain wall. Map the data collected by the sensors to the digital twin model through inverse distance weighted interpolation to generate the measured wind pressure load distribution data. At the same time, use 10 temperature sensors installed on the curtain wall surface and inside the building to collect data, combine the heat transfer parameters of the building structure, and perform finite element heat transfer analysis through ANSYS to obtain the temperature field parameters. For example, the temperature spatial distribution matrix at a certain moment shows that the temperature at the top of the curtain wall is 28°C and the temperature at the bottom is 25°C, and generate a time-varying gradient sequence of temperature changes over time. Integrate the normal vector distribution characteristics, wind pressure load distribution data, and temperature field parameters, and construct a dynamic flexural deformation coupling calculation model based on the finite element method. Input the wind pressure load distribution data, temperature field parameters, and the elastic modulus of the glass material (assumed to be 70 GPa) into the model, and set the fixed constraint conditions for the connection nodes between the curtain wall and the main structure. Through model calculation, obtain the global deformation parameters, such as the overall waviness is 3 mm and there are 3 regional curvature abnormal values; local deformation parameters, such as the corner stress gradient of the first detection target point is 15 MPa / m, and the peak value of the dynamic flexural displacement at the edge-center junction of the second detection target point is 8 mm.

[0086] Through grid-based block processing and extraction of neighboring point sets, the point cloud data on the curtain wall surface is refined. By combining the least squares method to fit a plane and calculate the surface normal vector, the geometric feature information of the curtain wall surface can be accurately obtained. Compared with traditional methods, the accuracy and reliability of data processing are improved. By integrating the measured wind pressure load distribution data and temperature field parameters, the influence of environmental factors on the curtain wall is comprehensively considered. The real-time acquisition and accurate mapping of wind pressure sensor data and temperature sensor data, combined with finite element heat transfer analysis to generate temperature field parameters, enable the model to truly simulate the stress and deformation conditions of the curtain wall under actual working conditions, avoiding the limitations of single-parameter analysis. The constructed dynamic flexural deformation coupling calculation model can comprehensively consider the interaction relationship among wind pressure, temperature, and the curtain wall structure. By setting reasonable constraint conditions and input parameters, accurate simulation of the dynamic flexural deformation of the curtain wall is achieved. The extracted global deformation parameters and local deformation parameters comprehensively describe the deformation of the curtain wall from both the overall and local levels. The overall waviness and regional curvature outliers reflect the overall flatness and local abnormal deformation of the curtain wall; the corner stress gradient and the peak value of the dynamic flexural displacement at the edge-center junction are analyzed in detail for key parts, which helps to detect potential safety hazards in a timely manner and provides specific quantitative indicators for the maintenance and management of the curtain wall.

[0087] In a preferred embodiment of the present invention, in step 3 above, according to the local deformation parameters, through phase correlation analysis and wavelet transform, the vibration displacement components of the first detection target point and the second detection target point are extracted to construct a curtain wall vibration mode feature matrix. At the same time, the displacement-time series data is analyzed to identify out-of-limit displacement events where the displacement amplitude is greater than or equal to a preset threshold, which may include:

[0088] Step 300, according to the local deformation parameters, determine the dynamically deformed sensitive regions corresponding to the first detection target point and the second detection target point in the time-series point cloud data;

[0089] Step 301, for the corner region point cloud sequence of the first detection target point in the sensitive region, through phase correlation analysis, extract the displacement change amount between adjacent time frames, and use wavelet transform to separate the low-frequency thermal expansion displacement component and the high-frequency wind-induced vibration displacement component; the high-frequency wind-induced vibration displacement component is used as the corner vibration displacement component; for the junction region point cloud sequence of the second detection target point, through wavelet transform, extract the dynamic displacement component within a preset frequency band;

[0090] Step 302, perform spatio-temporal alignment on the corner vibration displacement component of the first detection target point and the junction dynamic displacement component of the second detection target point to construct a curtain wall vibration mode feature matrix;

[0091] Step 303: Analyze the displacement-time series data in the vibration mode feature matrix, and preset a displacement amplitude threshold to identify single overrun events where the displacement amplitude is greater than or equal to the preset displacement amplitude threshold.

[0092] In the embodiment of the present invention, with the first detection target point as the center, a spatial range including the stress concentration area is delimited according to the distribution range of the corner stress gradient. For example, if the corner stress gradient varies significantly within a range of 0.5 meters from the target point, a spherical area with the target point as the center of the sphere and 0.5 meters as the radius is delimited as the dynamic deformation sensitive area of the first detection target point. For the second detection target point, a rectangular or irregular-shaped area is delimited at the junction of the center and the edge according to the influence range of its dynamic flexural displacement peak value. Assuming that the displacement peak value at the junction where the second detection target point is located mainly affects the surrounding range of 1 meter × 0.8 meters, a rectangular area with corresponding dimensions is delimited. In the time-series point cloud data, the point cloud data sequence falling within the sensitive area is filtered out by matching the coordinate information of the point cloud with the delimited sensitive area. Since the point cloud data carries timestamp information, these filtered point cloud sequences can reflect the deformation of the sensitive area over time.

[0093] Step 301: For the point cloud sequence in the corner area, the point cloud data of adjacent time frames are regarded as images (which can be transformed into two-dimensional images by projecting the point cloud onto a specific plane). Phase correlation analysis is based on the Fourier transform. First, perform a two-dimensional Fourier transform on the images corresponding to the point cloud of two adjacent frames to obtain the amplitude and phase information in the frequency domain. According to the phase correlation theorem, calculate the phase correlation function of the two frames of images where and are the Fourier transform results of the two frames of images respectively, is 's complex conjugate, is the frequency coordinate of the corresponding image in the horizontal direction, is the frequency coordinate of the corresponding image in the vertical direction. By finding the peak position of the phase correlation function, the translation amount between two frames of point clouds is determined, that is, the displacement change amount between adjacent time frames. Taking the extracted displacement change amount sequence as the input, a suitable wavelet basis function (such as Daubechies wavelet) is selected for discrete wavelet transform. The wavelet transform decomposes the displacement sequence into different frequency sub-bands. The displacement change caused by thermal expansion usually has a lower frequency, while the displacement change caused by wind vibration has a higher frequency. The frequency threshold is set based on historical data statistical analysis. For example, through the spectral analysis of past curtain wall displacement data, it is found that the thermal expansion displacement frequency is mainly between 0 - 0.1Hz, and the wind vibration displacement frequency is above 0.1Hz, then the frequency threshold can be set to 0.1Hz. In the result of discrete wavelet transform, the detail components of different layers correspond to different frequency ranges. The higher the layer number, the lower the corresponding frequency. According to the frequency threshold and the frequency characteristics of wavelet transform, it is determined which detail components and approximation components belong to the thermal expansion displacement component and which belong to the corner vibration displacement component. According to the set frequency threshold, the approximation component and detail components belonging to the thermal expansion displacement component are reconstructed to obtain the thermal expansion displacement component sequence; the detail components belonging to the corner vibration displacement component are reconstructed to obtain the corner vibration displacement component sequence.

[0094] Perform wavelet transform directly on the point cloud sequence in the junction area. According to the frequency range affected by factors such as wind force under actual working conditions of the curtain wall, a preset frequency band is set (for example, 3 - 10Hz, which is a common frequency range for general building curtain walls affected by wind vibration). After decomposing the point cloud sequence into different frequency sub-bands through wavelet transform, the dynamic displacement component within the preset frequency band is extracted, and this component contains the displacement change information generated by the dynamic loads such as wind force acting on the junction.

[0095] Step 302, since there may be a slight difference in the displacement data acquisition time between the first detection target point and the second detection target point, and their spatial positions are different, spatio-temporal alignment is required. In terms of time, taking the time series of one of the target points as the benchmark, the displacement data of the other target point is time-calibrated through methods such as linear interpolation to ensure that the displacement data of the two target points corresponds to the same moment on the time axis. In terms of space, according to the three-dimensional coordinates of the two target points in the digital twin model, the displacement data is converted to a unified coordinate system. The corner vibration displacement component of the first detection target point and the junction dynamic displacement component of the second detection target point after spatio-temporal alignment are arranged in chronological order to form a two-dimensional matrix. Each row of the matrix corresponds to a time step, and each column corresponds to the displacement components of the two target points in different directions ( 、 、 ). For example, the first column of the matrix is the vibration displacement component of the first detection target point in the direction, and the second column is the vibration displacement component of the second detection target point in the The dynamic displacement components in the direction, and so on. Finally, the vibration modal characteristic matrix of the curtain wall is constructed.

[0096] Step 303, according to the curtain wall design specifications and safety standards, combined with historical detection data and actual usage conditions, preset the displacement amplitude threshold. For example, for a certain type of curtain wall, according to its material strength and structural design, the displacement amplitude threshold is set to 5 mm. Analyze the displacement data at each time step in the vibration modal characteristic matrix, and calculate the combined displacement amplitude of the displacement components in each direction at each time step: , where 、 、 are the displacement components in three directions respectively. If the combined displacement amplitude at a certain time step is greater than or equal to the preset threshold, then this time step is identified as an over-limit displacement event, and information such as the time of event occurrence and displacement amplitude is recorded. By analyzing the entire displacement-time series data, characteristics such as the occurrence frequency and duration of over-limit displacement events can be statistically obtained, providing a basis for evaluating the safety of the curtain wall.

[0097] Suppose there is a high-rise commercial building with a unitized glass curtain wall structure on its exterior facade. In a certain detection:

[0098] Through preliminary calculation, it is obtained that the corner stress gradient of the first detection target point changes significantly in the area with a radius of 0.4 meters centered on the target point. Therefore, this spherical area is designated as the dynamic deformation sensitive area of the first detection target point; the second detection target point is at the junction of the center and the edge, and the influence range of its dynamic flexural displacement peak value is a rectangular area of 1.2 m × 0.6 m. This rectangular area is designated as the dynamic deformation sensitive area of the second detection target point. In the time-series point cloud data, filter out the point cloud sequences falling into these two areas to obtain the data for subsequent analysis. For the first detection target point, project the adjacent time frames of the corner area point cloud sequence onto the XY plane and convert them into images, perform phase correlation analysis, and calculate the displacement change amount between two adjacent frames of point clouds. Then, perform discrete wavelet transform through Daubechies wavelet, set the frequency threshold to 1 Hz, regard the displacement components below 1 Hz as thermal expansion displacement components, and extract the displacement components above 1 Hz as corner vibration displacement components. For the second detection target point, perform wavelet transform on the junction area point cloud sequence and extract the dynamic displacement components in the frequency band of 3 - 8 Hz. Based on the time series of the first detection target point, perform time interpolation calibration on the displacement data of the second detection target point, and convert the displacement data of the two target points to the same coordinate system. Then, arrange the corner vibration displacement components of the first detection target point and the dynamic displacement components at the junction of the second detection target point in chronological order to construct a two-dimensional matrix, completing the construction of the vibration modal characteristic matrix of the curtain wall. <{

[0099] According to the design requirements of the curtain wall, the preset displacement amplitude threshold is 4 mm. Analyze the displacement data in the vibration mode characteristic matrix. At a certain time step, the comprehensive displacement amplitude of the first detection target point is calculated to be 5.2 mm, exceeding the threshold. It is determined that this time step is an over-limit displacement event, and the relevant information is recorded.

[0100] Through phase correlation analysis and wavelet transform, the vibration displacement components of the curtain wall under different environmental factors can be accurately extracted from complex point cloud data. Phase correlation analysis can accurately calculate the displacement changes between adjacent time frames, and wavelet transform effectively separates the low-frequency thermal expansion and high-frequency wind vibration displacements, avoiding the problem that traditional methods are difficult to distinguish the influences of different factors, and providing a powerful means for in-depth analysis of the dynamic response of the curtain wall. After spatially and temporally aligning the displacement components of the first and second detection target points, a vibration mode characteristic matrix is constructed, integrating the dynamic information of key positions of the curtain wall. By presetting the threshold to identify over-limit displacement events, abnormal displacement situations that occur during the use of the curtain wall can be detected in a timely manner. Compared with traditional manual inspections, this data-driven automatic identification method is more sensitive and accurate, can early warn of potential safety risks, provide a scientific basis for the maintenance and repair of the curtain wall, reduce the probability of safety accidents, and ensure the safe operation of the building.

[0101] In a preferred embodiment of the present invention, in step 4 above, according to the global deformation parameter and the occurrence frequency of over-limit displacement events, determine the initial coefficient of the curtain wall structure safety degree, and calculate the target correction value based on the local deformation parameter and the dynamic displacement component, which may include:

[0102] Step 400, obtain the historical cumulative term of wind pressure according to the wind pressure from the starting moment to the current moment and the cumulative attenuation of wind pressure over time; analyze the surface temperature field data of the curtain wall obtained by laser point cloud scanning, and combine the spatial coordinate information of the curtain wall to obtain the temperature gradient reflecting the local thermal stress condition of the curtain wall; determine the fatigue cumulative damage according to the number of cycles of the curtain wall under different stress levels and the corresponding fatigue life.

[0103] Step 401, fuse the historical cumulative term of wind pressure, the gradient of the temperature field at the spatial position, and the fatigue cumulative damage to determine the static mechanics related term;

[0104] Step 402, determine the maximum displacement at the junction of the curtain wall under dynamic load, that is, the peak value of dynamic deflection displacement, and the maximum vibration displacement of the corner of the curtain wall in high-frequency wind vibration after frequency band filtering processing, that is, the peak value of the vibration displacement of the filtered corner, to obtain a comprehensive parameter reflecting the degree of mutual influence between deformation and vibration of the curtain wall structure under dynamic load.

[0105] Step 403: Determine the relationship between the static risk and the deformation-vibration sensitivity of the curtain wall structure based on the static-related terms and the comprehensive parameter of the mutual influence degree between the deformation and vibration of the curtain wall structure under dynamic loads; obtain the relationship representing the vibration displacement synergy between the corner and the junction according to the peak value of the filtered corner vibration displacement and the peak value of the filtered dynamic displacement at the same frequency band at the curtain wall junction, that is, the dynamic-related terms.

[0106] Step 404: Determine the contributions of the static factors and the dynamic factors to the target correction value respectively according to the static-related terms and the dynamic-related terms, so as to obtain the final target correction value.

[0107] In the embodiment of the present invention, high-precision wind pressure sensors are installed at key positions such as the windward side, corners, top and bottom of the curtain wall, and wind pressure data is collected in real time at a fixed frequency (such as once per second). The wind pressure data collected by the sensors contains size and direction information, and is transmitted to the data center through a non-communication method to form a sequence of wind pressure varying with time , where is the time variable. The attenuation coefficient is determined according to the climatic conditions of the region where the building is located and the historical wind pressure data, and its value range is between 0.02 and 0.1, which is used to describe the attenuation characteristics of the wind pressure with time. The wind pressure farther from the current moment has less influence on the structure.

[0108] Use a laser point cloud scanning device to periodically scan the curtain wall surface. The scanning frequency is set according to the change of the ambient temperature. For example, the scanning is encrypted when the day-night temperature difference is large or during the season alternation. The scanning device combines a high-precision positioning system to obtain the temperature value of each point and its three-dimensional spatial coordinates . By performing finite difference calculation on the temperature values of adjacent points, the temperature gradient is obtained, which reflects the local thermal stress condition of the curtain wall. Consult the curtain wall design documents and material performance reports to obtain the theoretical fatigue life of the curtain wall structure under different stress levels . At the same time, stress sensors are installed at the key stress-bearing parts of the curtain wall to monitor the stress change in real time. When the stress exceeds the set threshold, a cycle is recorded, and the actual cycle times under each stress level are accumulated .

[0109] Install high-precision displacement sensors at the curtain wall junctions and corners to monitor the displacement changes under dynamic loads (such as wind force). By setting a band-pass filter, the displacement data in the high-frequency wind vibration frequency band (3-10 Hz) is screened out, and then the maximum displacement of the curtain wall junction under dynamic loads (i.e., the peak value of the dynamic flexural displacement), and the maximum vibration displacement of the curtain wall corner after frequency band filtering (i.e., the peak value of the vibrational displacement at the corner after filtering), and the peak value of the dynamic displacement at the junction of the curtain wall after filtering in the same frequency band 。

[0110] Comprehensively determine the initial safety factor:

[0111] Monitor curtain walls of different types (such as unit curtain walls, framed curtain walls, point-supported curtain walls, etc.) and different scales (small building curtain walls, large commercial complex curtain walls, high-rise building curtain walls, etc.). These curtain walls cover different geographical locations, service life, and environmental conditions to obtain rich and diverse data samples.

[0112] For each data sample, accurately record the following key information. Use high-precision laser scanning equipment to obtain the point cloud data of the curtain wall surface, and calculate the standard deviation of the overall waviness through processing, accurate to 0.01 mm. Combine the point cloud data through advanced 3D modeling technology to calculate and count the number of regional curvature outliers, and record the degree of deviation of each outlier from the normal range, accurate to 0.001 。Use high-precision displacement sensors installed at key parts of the curtain wall to count the occurrence frequency of over-limit displacement events, and the statistical period can be set to one month. For each data sample, combine the design data and historical maintenance records of the curtain wall to preliminarily determine the corresponding initial safety factor, with the value range set between 0 and 1. The larger the value, the higher the safety level.

[0113] After collecting a large number of data samples, first clean and preprocess the data. Check the integrity of the data, fill in missing values, verify and correct or eliminate abnormal data points. Standardize the data such as the standard deviation of the overall waviness, the number and deviation degree of regional curvature outliers, and the occurrence frequency of over-limit displacement events to make them comparable. Use statistical methods, such as correlation analysis, to study the linear or non-linear relationship between each factor and the initial safety factor, and preliminarily judge which factors have a greater impact on the initial safety factor. Taking multiple linear regression as an example, use the standard deviation of the overall waviness, the number and deviation degree of regional curvature outliers, and the occurrence frequency of over-limit displacement events as independent variables, and the initial safety factor as the dependent variable to construct a regression model. By continuously adjusting the model parameters and optimizing the model fitting effect, make the model accurately reflect the internal relationship between each factor and the initial safety factor, so as to construct an accurate correspondence table.

[0114] Determination of the initial safety factor:

[0115] Use a measuring device and a standardized measuring method to accurately obtain the standard deviation of the overall waviness of the curtain wall, the number and deviation degree of regional curvature outliers again, and precisely count the occurrence frequency of over-limit displacement events through displacement sensors. Substitute the data obtained from actual measurement and statistics into the corresponding relationship table established through the above analysis. Use the weighted average method for comprehensive calculation to determine the coefficients and weights corresponding to each factor.

[0116] For example, the coefficient corresponding to the standard deviation of the overall waviness is determined by dividing the interval according to its numerical value. When the standard deviation is less than 0.5 mm, the corresponding coefficient is 0.8 - 1.0; when it is between 0.5 - 1.5 mm, the corresponding coefficient is 0.6 - 0.8; when it is greater than 1.5 mm, the corresponding coefficient is 0.4 - 0.6. The correlation coefficient of regional curvature outliers is determined according to the number and deviation degree of outliers. When the number is small and the deviation degree is low, the coefficient is 0.8 - 1.0; when the number and deviation degree are moderate, the coefficient is 0.6 - 0.8; when the number is large and the deviation degree is high, the coefficient is 0.4 - 0.6. The coefficient corresponding to the occurrence frequency of over-limit displacement events is determined according to the occurrence frequency. When the occurrence frequency within one month is less than 5 times, the coefficient is 0.8 - 1.0; when it is between 5 - 15 times, the coefficient is 0.6 - 0.8; when it is greater than 15 times, the coefficient is 0.4 - 0.6.

[0117] In terms of weights, the weight range of the standard deviation of the overall waviness is 0.3 - 0.5. For curtain walls with high flatness requirements, such as glass curtain walls, this weight can be biased towards 0.5; the weight of the relevant data of regional curvature outliers is 0.2 - 0.4. If the curtain wall structure is complex and local stress problems are likely to occur, the weight can be biased towards 0.4; the weight of the occurrence frequency of over-limit displacement events is 0.2 - 0.4. In the curtain wall scenarios with large wind loads or long service life, the weight can be biased towards 0.4.

[0118] Initial safety factor = coefficient corresponding to the standard deviation of the overall waviness × its weight + correlation coefficient of regional curvature outliers × its weight + coefficient corresponding to the occurrence frequency of over-limit displacement events × its weight. Considering the influence of each factor comprehensively, finally determine the accurate initial safety factor of the curtain wall structure. 。

[0119] Weight coefficient setting:

[0120] Used to adjust the weight of the historical cumulative term of wind pressure in the static-related terms, with a value range of 0.3 - 0.5. When the building is located in a windy area, a relatively larger value can be appropriately taken;

[0121] Used to adjust the weight of the temperature gradient term, with a value range of 0.2 - 0.4. For curtain wall materials with high thermal sensitivity, the value is on the larger side;

[0122] Reflect the importance degree of the fatigue cumulative damage term, with a value range of 0.2 - 0.3;

[0123] Used to quantify the relative importance of the static-related terms (including the cumulative wind pressure history, temperature gradient, fatigue cumulative damage) when calculating the target correction value, with a value range of 0.6 - 0.8, emphasizing the dominant influence of static factors on the safety of the curtain wall structure;

[0124] Adjust the weight of the dynamics-related term (vibration displacement coordination), with a value range of 0.2 - 0.4.

[0125] Positive number Take , used to avoid the situation where the denominator of the formula is zero.

[0126] Using the numerical integration method (such as the trapezoidal integration method), divide the time interval , into multiple small intervals, perform weighted calculation on the wind pressure values within each small interval, and then multiply by the weight coefficient to obtain the value of the cumulative wind pressure history term . Multiply the obtained temperature gradient by the weight coefficient to obtain the value of the temperature gradient term .

[0127] Add the ratio of the actual number of cycles at different stress levels to the corresponding fatigue life , and then multiply by the weight coefficient to obtain the value of the fatigue cumulative damage term .

[0128] Add the cumulative wind pressure history term, temperature gradient term, and fatigue cumulative damage term calculated above to obtain the numerator of the static-related term , combine with the obtained dynamic displacement parameter and to calculate the denominator; finally, divide the numerator by the numerator of the static-related term to obtain the value of the static-related term .

[0129] Divide the peak value of the filtered corner vibration displacement by the peak value of the filtered dynamic displacement at the curtain wall junction in the same frequency band , and then multiply by the weight coefficient to obtain the value of the dynamics-related term, which characterizes the coordination of the vibration displacements at the corner and junction of the curtain wall.

[0130] Multiply the calculated static-related terms by a weight coefficient , and add them to the dynamic-related terms to obtain the final target correction value , that is . This value is used to correct the initial coefficient of the curtain wall structure safety factor , comprehensively reflecting the influence degrees of static and dynamic factors on the safety of the curtain wall structure . The larger the value, the higher the risk faced by the curtain wall structure

[0131] The formula incorporates static factors such as historical wind pressure accumulation, temperature gradient, and fatigue cumulative damage, as well as dynamic factors such as the deformation and vibration relationship of different parts of the curtain wall under dynamic loads, changing the limitations of traditional single-factor or few-factor evaluations, being able to comprehensively reflect the actual conditions of the curtain wall structure under complex environments and loads, and improving the accuracy and reliability of the evaluation. It includes considerations of static long-term acting factors (such as long-term wind pressure accumulation and fatigue damage), and also incorporates the structural responses under dynamic loads (such as peak vibration displacement and synergy), realizing an evaluation method that combines static and dynamic aspects. It can more realistically simulate the stress and deformation conditions of the curtain wall during actual use, and timely discover potential safety hazards

[0132] By setting different weight coefficients ( , , , , ), the contribution degrees of various factors to the target correction value can be flexibly adjusted according to the specific conditions such as the material properties, structural forms, and usage environments of the curtain wall. This makes the evaluation model have stronger adaptability and is applicable to the safety evaluations of different types of curtain wall structures

[0133] It clearly determines the relationships between static risks and the deformation-vibration sensitivity of the curtain wall structure, as well as the vibration displacement synergy at corners and junctions, etc., and obtains the target correction value through quantitative calculation. It provides specific quantitative indicators for the safety evaluation of the curtain wall structure, facilitating engineers and managers to intuitively understand the safety state of the curtain wall structure and providing a scientific basis for decision-making. Through the calculation of the fatigue cumulative damage term, the fatigue state of the curtain wall structure during long-term use can be tracked, early warning of the structural failure risk caused by fatigue can be given, which helps to formulate reasonable maintenance and inspection plans, extend the service life of the curtain wall, and reduce the probability of safety accidents

[0134] In a preferred embodiment of the present invention, in step 5 above, the target correction value is superimposed on the initial safety factor to generate a corrected curtain wall structure safety factor, realizing the dynamic adjustment of the initial safety factor, and outputting a three-dimensional inspection report, including the risk level, the distribution of the corrected safety factor, and the warning marks associated with the target correction value, which may include:

[0135] Step 500, incorporate the dynamic adjustment information reflected by the target correction value, including local deformation and dynamic displacement components, into the initial safety factor to generate a corrected curtain wall structure safety factor;

[0136] Step 501, according to the corrected curtain wall structure safety factor, determine the risk level of the curtain wall by referring to the preset risk level classification standard;

[0137] Step 502, map the corrected safety factor onto the digital twin model of the curtain wall surface, process the distribution of the corrected safety factor in each part of the curtain wall, and obtain the distribution status of the corrected safety factor in each part of the curtain wall;

[0138] Step 503, add corresponding warning marks in the digital twin model according to the magnitude and change trend of the target correction value, and the correlation with the distribution status of the safety factor in each part of the curtain wall;

[0139] Step 504, integrate the risk level of the curtain wall, the distribution status of the safety factor in each part of the curtain wall, the warning marks, the basic information of the curtain wall, and the inspection process description, including the inspection time, inspection equipment, and inspection method, to generate a three-dimensional inspection report.

[0140] In the embodiment of the present invention, first, the target correction value is analyzed to clarify the local deformation and dynamic displacement component information it contains. For example, the target correction value may include dynamic displacement information such as the peak value of high-frequency vibration displacement at the corner of the curtain wall and the peak value of dynamic flexural displacement at the junction, as well as information related to local deformation such as temperature gradient and fatigue cumulative damage. Classify and quantify this information according to certain rules. Determine the fusion method of the initial safety factor and the target correction value. The fusion strategy is weighted summation, that is, the corrected curtain wall structure safety factor + × , where is the initial safety factor, It is the fusion weight, with a value range between 0.5 and 1, which can be adjusted according to actual engineering experience and the degree of emphasis on various factors. This formula indicates that, based on the initial safety factor, the coefficient is dynamically adjusted according to the magnitude of the target correction value and the fusion weight. If the target correction value reflects a higher risk, after calculation by this formula, the corrected safety factor will increase accordingly; otherwise, it will decrease. For each detection point (or grid unit) of the curtain wall structure, obtain its initial safety factor and the corresponding target correction value respectively, and calculate according to the above fusion formula to obtain the corrected safety factor of the curtain wall structure for each point (or grid unit).

[0141] Step 501: Before detection, according to the design specifications, material properties, service life of the curtain wall, and relevant industry standards, formulate a preset risk level classification standard. For example, divide the risk levels into four levels: low risk, medium risk, high risk, and extremely high risk, and set specific safety factor ranges for each level. Assume that the safety factor range corresponding to low risk is 0.8 - 1.0, medium risk is 0.6 - 0.79, high risk is 0.4 - 0.59, and extremely high risk is 0 - 0.39.

[0142] Compare the corrected safety factor of the curtain wall structure for each detection point (or grid unit) calculated in Step 500 with the preset risk level classification standard. Determine the risk level to which each point (or grid unit) belongs. Then, by counting the number of points (or grid units) in different risk levels and the proportion they account for in the total curtain wall area, comprehensively evaluate the risk level of the entire curtain wall. For example, if the area proportion of high-risk and extremely high-risk areas exceeds 30%, then determine that the risk level of the entire curtain wall is high risk.

[0143] Step 502: Perform grid processing on the digital twin model of the curtain wall, and each grid corresponds to an actual area on the curtain wall surface. Map the corrected safety factor of each detection point (or grid unit) calculated in Step 500 to the corresponding grid in the digital twin model. Ensure that the safety factor corresponds accurately to the actual curtain wall part.

[0144] Render the digital twin model using professional 3D visualization software (such as Unity, Unreal Engine, etc.). Assign different colors or transparencies to each grid according to the size of the corrected safety factor. For example, areas with a high safety factor are represented by green, and as the factor decreases, the color gradually transitions to yellow, orange, and finally red for high-risk areas. At the same time, different transparencies can be set to make the distribution of the safety factor more intuitive. In addition, contour lines or isopleths can be added to further show the changing trend of the safety factor on the curtain wall surface. Smooth the mapped safety factor data to avoid discontinuous or abnormal displays caused by detection errors or data fluctuations. A filtering algorithm (such as Gaussian filtering) can be used to process the data to make the distribution of the safety factor smoother and more natural. At the same time, compress and optimize the data to improve the efficiency and performance of the visual display.

[0145] Step 503: Conduct an in-depth analysis of the magnitude and changing trend of the target correction value. Calculate the change rate of the target correction value within a certain time period. For example, calculate the ratio of the difference between the target correction values of two adjacent detections to the previous target correction value to obtain the change rate of the target correction value. At the same time, combine the distribution of the safety factor in each part of the curtain wall to analyze the correlation between the target correction value and the safety factor. For example, observe whether there is an obvious increasing trend in the target correction value in areas with a low safety factor. Based on the analysis results, formulate warning rules. For example, if the target correction value in a certain area is greater than the preset high-risk threshold and the safety factor is in the medium-low risk range, or the change rate of the target correction value exceeds a certain percentage (such as 20%), then it is determined that a warning mark needs to be added to this area. The warning rules can be flexibly adjusted according to the actual situation to ensure that potential risks can be discovered promptly and accurately. In the digital twin model, add warning marks to the corresponding areas according to the warning rules. The warning marks can take different forms, such as flashing icons, prominent text prompts, etc. For example, add a red exclamation mark icon in the high-risk area and display a brief risk prompt message beside it, such as "There is a high safety risk here, please check promptly". At the same time, set different priorities for the warning marks so that users can quickly pay attention to the most urgent risk areas.

[0146] Step 504: Collect the basic information of the curtain wall, including the type of the curtain wall (such as unitized curtain wall, framed curtain wall), area, service life, design unit, etc.; the information describing the detection process, such as detection time, detection equipment (such as laser point cloud scanner, model of wind pressure sensor, etc.), detection methods (such as finite element analysis method, wavelet transform analysis method, etc.); the risk level of the curtain wall determined in Step 501; the distribution of the corrected safety factor obtained in Step 502; the warning marks added in Step 503 and other data. Classify and organize these data to ensure the accuracy and integrity of the data. Design the template of the 3D detection report and determine the structure and content layout of the report. The report can include parts such as cover, table of contents, overview, detailed detection results, conclusions and suggestions.

[0147] Fill the organized data into the corresponding positions according to the requirements of the report template. For the risk level, display the area proportion of different risk level regions in the form of a chart; for the distribution of the safety factor, insert a 3D visualization image and provide an interactive operation function to facilitate users to view from different angles; for the warning marks, list the detailed warning area locations, risk descriptions and recommended measures and other information. Finally, review and proofread the report. After ensuring that the content is accurate, generate the final 3D detection report. The report can be output in PDF format or 3D interactive file format (such as HTML5 format) for easy viewing and sharing by users.

[0148] Suppose there is a high-rise commercial building with a unitized glass curtain wall structure on its facade. In a certain regular inspection:

[0149] The initial safety factor is obtained through preliminary calculation = 0.7, the target correction value = 0.15, and the set fusion weight = 0.8. According to the formula + × , the corrected safety factor S of the curtain wall structure is calculated as S = 0.7 + 0.8×0.15 = 0.82. Comparing with the preset risk level classification standard, the low-risk range is 0.8 - 1.0. Since 0.8 ≤ 0.82 ≤ 1.0, the risk level of the curtain wall is determined to be low risk. The digital twin model is evenly divided into 1000 grid units, and the corrected safety factor corresponding to each grid is recorded and marked one by one. According to the preset rules, the grids with safety factors in the range of 0.8 - 0.9 are marked as "light green", and the grids in the range of 0.9 - 1.0 are marked as "dark green". The grid coefficients are processed through a data smoothing algorithm to eliminate outliers caused by detection errors or data fluctuations, and finally a detailed distribution list of the safety factors of each part of the curtain wall is formed. It can be clearly seen from the list that the safety factors in most areas are in the green marked range, indicating that the overall safety condition is good.

[0150] In the comparative analysis of the detection data, it is found that in the grid area numbered A-123 in the southeast corner of the curtain wall, the target correction value has increased from 0.1 in the previous detection to 0.2 in this detection. After calculation, the change rate is = 100%, significantly exceeding the preset threshold of 20%. At the same time, the safety factor of this area is 0.65, which is in the medium-risk range. According to the established early warning rules, a special identification symbol "△" is added at the corresponding grid position in the digital twin model and set to a red flashing state. At the same time, a prompt message "The target correction value in the area A-123 in the southeast corner has changed too much, there is a potential risk" is added next to the grid entry in the detection data record form.

[0151] Collect the basic information of the curtain wall, including the type of the curtain wall is unitized glass curtain wall, the total area is 5000 square meters, and the service life is 5 years, etc.; detailed record the information of the detection process, such as the detection time is October 1, 2024, and the detection equipment used includes a certain type of laser point cloud scanner, wind pressure sensor, etc., and the detection methods adopted cover technical means such as finite element analysis and wavelet transform. Arrange the risk level assessment results, the distribution list of the safety factors of each grid, the early warning mark positions and risk descriptions, etc. according to the standard report template. In the finally generated three-dimensional detection report, the risk level distribution ratio is presented in the form of a data table, the safety factor situation of each area is described in the way of combining text with data lists, and the grid numbers, specific positions and risk contents of all early warning marks are listed in detail.

[0152] By superimposing the target correction value on the initial safety factor, the impact of various factors such as local deformation and dynamic displacement on the curtain wall safety is comprehensively considered. It can more accurately evaluate the actual safety condition of the curtain wall, avoid the limitations of single-factor evaluation, and improve the accuracy and reliability of the evaluation results. The dynamic adjustment of the initial safety factor is realized, which can timely reflect the changes in the safety performance of the curtain wall caused by factors such as environmental changes and load effects during the use process. This dynamic adjustment mechanism makes the safety assessment more in line with the actual situation and helps to timely discover potential safety hazards. The generated 3D inspection report visually displays the distribution of the corrected safety factor through the digital twin model and adds warning marks. Managers and technicians can quickly and comprehensively understand the safety conditions of various parts of the curtain wall through the 3D visualization image, locate high-risk areas, and improve the efficiency of problem detection and decision-making. According to the correlation between the target correction value and the safety factor distribution, warning marks are added to give early warnings of potential risks. Users can take corresponding maintenance and repair measures in a timely manner according to the warning information, reduce the probability of safety accidents, ensure the safety of personnel's lives and property, and at the same time help to extend the service life of the curtain wall and reduce maintenance costs. The 3D inspection report integrates various aspects such as the basic information of the curtain wall, the description of the inspection process, the risk level, the safety factor distribution, and warning marks, providing comprehensive and detailed information for the maintenance, management, and supervision of the curtain wall. It is convenient for different personnel to understand the safety situation of the curtain wall from different perspectives, promoting information sharing and communication.

[0153] As Figure 2 shown, an embodiment of the present invention further provides a safety detection device for the deformation condition of a curtain wall based on laser point cloud, including:

[0154] A data processing module, configured to perform coordinate system normalization processing on the three-dimensional point cloud data set covering the entire surface of the curtain wall, generate a digital twin model of the curtain wall surface, and define a first detection target point and a second detection target point in the digital twin model according to the geometric characteristics and material joint characteristics of the curtain wall panels;

[0155] A model construction module, configured to extract the surface normal vector distribution characteristics according to the digital twin model, and fuse the measured wind pressure load distribution data and temperature field parameters to construct a dynamic flexural deformation coupling calculation model, and output global deformation parameters and local deformation parameters;

[0156] A feature extraction module, configured to extract the vibration displacement components of the first detection target point and the second detection target point through phase correlation analysis and wavelet transform according to the local deformation parameters to construct a curtain wall vibration mode feature matrix. At the same time, analyze the displacement-time series data to identify over-limit displacement events with displacement amplitudes greater than or equal to a preset threshold;

[0157] A safety assessment module, configured to determine an initial coefficient of the curtain wall structure safety degree according to the global deformation parameters and the occurrence frequency of over-limit displacement events, and calculate a target correction value based on the local deformation parameters and the dynamic displacement components;

[0158] A report generation module, configured to superimpose the target correction value on the initial coefficient of the safety degree, generate a corrected safety coefficient of the curtain wall structure, realize dynamic adjustment of the initial coefficient of the safety degree, and output a three-dimensional detection report, including a risk level, a distribution of the corrected safety coefficient, and a warning mark associated with the target correction value.

[0159] It should be noted that this device corresponds to the above method. All implementation manners in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0160] An embodiment of the present invention further provides a computing device, including: a processor and a memory storing a computer program. When the computer program is run by the processor, the above-mentioned method is executed. All implementation manners in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0161] An embodiment of the present invention further provides a computer-readable storage medium storing instructions. When the instructions are run on a computer, the computer is caused to execute the above-mentioned method. All implementation manners in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0162] The above is the preferred embodiment of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the principle of the present invention, several improvements and refinements can be made, and these improvements and refinements should also be regarded as the protection scope of the present invention.

Claims

1. A safety detection method for the deformation condition of a curtain wall based on laser point cloud, characterized in that, The method includes: Step 1: Perform coordinate system normalization on the three-dimensional point cloud data set covering the entire surface of the curtain wall to generate a digital twin model of the curtain wall surface. In the digital twin model, define the first detection target point and the second detection target point according to the geometric characteristics and material joint characteristics of the curtain wall panels. The first detection target point is located in the four corner regions of the curtain wall panel and is used to monitor the corner stress concentration effect. The second detection target point is located at the junction of the center and the edge of the curtain wall panel and is used to monitor the change in the temperature difference deformation gradient. Step 2: According to the digital twin model, extract the surface normal vector distribution characteristics, and fuse the measured wind pressure load distribution data and the temperature field parameters to construct a dynamic flexural deformation coupling calculation model, and output the global deformation parameters and local deformation parameters. The global deformation parameters include the overall waviness and the regional curvature anomaly value. The local deformation parameters include the corner stress gradient of the first detection target point and the peak value of the dynamic flexural displacement at the edge-center junction of the second detection target point. Step 3: According to the local deformation parameters, extract the vibration displacement components of the first detection target point and the second detection target point through phase correlation analysis and wavelet transform to construct a curtain wall vibration mode feature matrix. At the same time, analyze the displacement-time series data to identify the over-limit displacement events with the displacement amplitude ≥ the preset threshold. Step 4: Determine the initial coefficient of the curtain wall structure safety degree according to the global deformation parameters and the occurrence frequency of the over-limit displacement events, and calculate the target correction value based on the local deformation parameters and the dynamic displacement components. Step 5: Superimpose the target correction value on the initial coefficient of the safety degree to generate a corrected curtain wall structure safety degree coefficient, realize the dynamic adjustment of the initial coefficient of the safety degree, and output a three-dimensional detection report, including the risk level, the distribution of the corrected safety degree coefficient, and the warning marks associated with the target correction value.

2. The safety detection method for curtain wall deformation conditions based on laser point cloud according to claim 1, characterized in that According to the digital twin model, extract the surface normal vector distribution characteristics, and fuse the measured wind pressure load distribution data and the temperature field parameters to construct a dynamic flexural deformation coupling calculation model, and output the global deformation parameters and local deformation parameters, including: Perform grid-based block processing on the curtain wall surface according to the surface point cloud data of the digital twin model, and for the point cloud data in each block area, extract the set of neighboring points within the preset radius of each target point in the current block. Fit the spatial plane of the neighboring point set by the least squares method, calculate the surface normal vector of each target point, and aggregate the normal vectors of each block to generate the surface normal vector distribution characteristics of the curtain wall. Generate the measured wind pressure load distribution data by mapping the wind pressure sensor data around the building curtain wall to the corresponding spatial coordinates of the digital twin model. Generate the temperature field parameters, including the temperature spatial distribution matrix and the time-varying gradient sequence, through finite element heat transfer analysis based on the data collected by the temperature sensors and the building structure heat transfer parameters. Fuse the surface normal vector distribution characteristics of the curtain wall, the measured wind pressure load distribution data, and the temperature field parameters to construct a dynamic flexural deformation coupling calculation model. Input the wind pressure load distribution data, temperature field parameters, and material elastic modulus into the dynamic flexural deformation coupling calculation model, and set the fixed constraint conditions for the connection nodes between the curtain wall and the main structure; Conduct dynamic flexural deformation analysis based on the fixed constraint conditions, wind pressure load distribution data, temperature field parameters, and material elastic modulus, and extract the global deformation parameters and local deformation parameters.

3. The safety detection method for curtain wall deformation conditions based on laser point cloud according to claim 2, characterized in that, According to the local deformation parameters, through phase correlation analysis and wavelet transform, extract the vibration displacement components of the first detection target point and the second detection target point to construct the curtain wall vibration mode feature matrix. At the same time, analyze the displacement-time series data to identify the over-limit displacement events where the displacement amplitude ≥ the preset threshold, including: According to the local deformation parameters, determine the dynamically deformed sensitive areas corresponding to the first detection target point and the second detection target point in the time-series point cloud data; For the corner area point cloud sequence of the first detection target point in the sensitive area, extract the displacement change amount between adjacent time frames through phase correlation analysis, and use wavelet transform to separate the low-frequency thermal expansion displacement component and the high-frequency wind vibration displacement component; the high-frequency wind vibration displacement component is used as the corner vibration displacement component; for the junction area point cloud sequence of the second detection target point, extract the dynamic displacement component within the preset frequency band through wavelet transform; Perform spatio-temporal alignment on the corner vibration displacement component of the first detection target point and the junction dynamic displacement component of the second detection target point to construct the curtain wall vibration mode feature matrix; Analyze the displacement-time series data in the vibration mode feature matrix and preset the displacement amplitude threshold to identify the single over-limit events where the displacement amplitude ≥ the preset displacement amplitude threshold.

4. The safety detection method for the deformation condition of the curtain wall based on laser point cloud according to claim 3, wherein, Determine the initial coefficient of the curtain wall structure safety factor based on the global deformation parameters and the occurrence frequency of the over-limit displacement events, and calculate the target correction value based on the local deformation parameters and the dynamic displacement components, including: Obtain the wind pressure historical cumulative term according to the wind pressure from the starting moment to the current moment and the cumulative attenuation of the wind pressure over time; analyze the curtain wall surface temperature field data obtained by laser point cloud scanning, and combine the spatial coordinate information of the curtain wall to obtain the temperature gradient reflecting the local thermal stress condition of the curtain wall; determine the fatigue cumulative damage according to the number of cycles and the corresponding fatigue life of the curtain wall under different stress levels; Fuse the wind pressure historical cumulative term, the gradient of the temperature field at the spatial position, and the fatigue cumulative damage to determine the static-related term; Determine the maximum displacement of the curtain wall junction under dynamic load, that is, the peak value of the dynamic flexural displacement, and the maximum vibration displacement of the curtain wall corner in high-frequency wind vibration after frequency band filtering, that is, the peak value of the filtered corner vibration displacement, to obtain the comprehensive parameter reflecting the mutual influence degree of deformation and vibration of the curtain wall structure under dynamic load; Determine the relationship between the static risk and the deformation-vibration sensitivity of the curtain wall structure according to the static-related term and the comprehensive parameter reflecting the mutual influence degree of deformation and vibration of the curtain wall structure under dynamic load; obtain the relationship characterizing the vibration displacement synergy between the corner and the junction according to the peak value of the filtered corner vibration displacement and the peak value of the dynamic displacement of the curtain wall junction after the same frequency band filtering, that is, the dynamic-related term; Determine the contributions of static factors and dynamic factors to the target correction value according to the static-related terms and dynamic-related terms, respectively, to obtain the final target correction value.

5. The safety detection method for curtain wall deformation condition based on laser point cloud according to claim 4, characterized in that, Superimpose the target correction value on the initial safety factor to generate a corrected curtain wall structure safety factor, realizing the dynamic adjustment of the initial safety factor, and output a three-dimensional inspection report, including the risk level, the distribution of the corrected safety factor, and the warning marks associated with the target correction value, including: Integrate the dynamic adjustment information reflected by the target correction value, including local deformation and dynamic displacement components, into the initial safety factor to generate a corrected curtain wall structure safety factor; Determine the risk level of the curtain wall according to the corrected curtain wall structure safety factor and the preset risk level classification standard; Map the corrected safety factor onto the digital twin model of the curtain wall surface, process the distribution of the corrected safety factor in each part of the curtain wall, and obtain the distribution status of the corrected safety factor in each part of the curtain wall; Add corresponding warning marks in the digital twin model according to the magnitude and change trend of the target correction value, as well as the correlation with the distribution status of the safety factor in each part of the curtain wall; Integrate the risk level of the curtain wall, the distribution status of the safety factor in each part of the curtain wall, the warning marks, the basic information of the curtain wall, and the description of the inspection process, including the inspection time, inspection equipment, and inspection method, to generate a three-dimensional inspection report.

6. A safety detection device for the deformation condition of a curtain wall based on laser point cloud, which implements the method described in any one of claims 1 to 5, characterized in that, Including: A data processing module for performing coordinate system normalization on the three-dimensional point cloud data set covering the entire surface of the curtain wall to generate a digital twin model of the curtain wall surface, and defining a first detection target point and a second detection target point in the digital twin model according to the geometric characteristics and material joint characteristics of the curtain wall panels; the first detection target point is located in the four corner regions of the curtain wall panel for monitoring the corner stress concentration effect; the second detection target point is located at the junction of the center and the edge of the curtain wall panel for monitoring the change of the temperature difference deformation gradient; A model construction module for extracting the surface normal vector distribution characteristics according to the digital twin model, and fusing the measured wind pressure load distribution data and the temperature field parameters to construct a dynamic flexural deformation coupling calculation model, and outputting global deformation parameters and local deformation parameters; the global deformation parameters include the overall waviness and the regional curvature anomaly value; the local deformation parameters include the corner stress gradient of the first detection target point and the peak value of the dynamic flexural displacement at the edge-center junction of the second detection target point; A feature extraction module for extracting the vibration displacement components of the first detection target point and the second detection target point through phase correlation analysis and wavelet transform according to the local deformation parameters to construct a curtain wall vibration mode feature matrix, and at the same time, analyzing the displacement-time series data to identify the over-limit displacement events with the displacement amplitude ≥ the preset threshold; A safety evaluation module for determining the initial curtain wall structure safety factor according to the global deformation parameters and the occurrence frequency of the over-limit displacement events, and calculating the target correction value based on the local deformation parameters and the dynamic displacement components; A report generation module is used to superimpose the target correction value on the initial safety factor, generate a corrected curtain wall structure safety factor, realize the dynamic adjustment of the initial safety factor, and output a three-dimensional detection report, including the risk level, the distribution of the corrected safety factor, and a warning mark associated with the target correction value.

7. A computing device, characterized in that, It includes: One or more processors; A storage device for storing one or more programs, which when executed by the one or more processors, cause the one or more processors to implement the method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, A program is stored in the computer-readable storage medium, and when the program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

Images