A method and system for generating detailed curtain wall models

Abstract

A method and system for generating a refined curtain wall model, relating to the field of architectural model generation technology, involves collecting multi-source heterogeneous data of the target building, generating curtain wall edge lines based on the building outline, dynamically adjusting the segmentation parameters of the curtain wall edge lines bidirectionally and adaptively using piecewise spline interpolation to generate the basic curtain wall surface, constructing a basic three-dimensional geometric model, generating an initial lattice surface, constructing a thermal response model by combining the basic three-dimensional geometric model, material property data, and thermal response data, acquiring thermal feature data for each region, optimizing the lattice surface of the basic three-dimensional geometric model based on the thermal feature data, generating a three-dimensional geometric model, and performing automated grouping and panel generation operations on the three-dimensional geometric model; fusing the three-dimensional geometric model, thermal feature data, and material property data through multimodal features to generate a refined three-dimensional curtain wall model, and performing multiphysics coupling verification on the refined three-dimensional curtain wall model, significantly improving the efficiency and accuracy of curtain wall model generation.

Description

Technical Field

[0001] This invention relates to the field of architectural model generation technology, specifically a method and system for generating a refined curtain wall model. Background Technology

[0002] Chinese Patent Publication No. CN114781030A discloses a method and system for generating a refined curtain wall model based on the curtain wall edge line, including the following steps: generating a curtain wall edge line based on preset modeling elements; creating a curtain wall dot matrix surface based on the curtain wall edge line; dividing the dots in the curtain wall dot matrix surface into multiple dot matrix groups by grouping; and generating a curtain wall model by combining preset adaptive panels and multiple dot matrix groups.

[0003] Chinese Patent Publication No. CN112149211A discloses a modeling method and system for curtain walls based on BIM models, including the following steps: S11. Identifying curtain wall nodes using node recognition software and uploading the identified curtain wall nodes to curtain wall modeling software; S12. Establishing a 3D facade mesh for the curtain wall based on the curtain wall modeling software; S13. Selecting curtain wall segments of unit panels in the established 3D facade mesh and placing the curtain wall nodes identified in step S11 at the node positions of the selected curtain wall segments to generate different standard panel template models; S14. Establishing the entire curtain wall model based on the generated standard panel template models.

[0004] In the construction industry, curtain walls, as the external envelope of buildings, are crucial to the aesthetics, performance, and safety of buildings due to the precision of their design and construction. Traditional curtain wall model generation methods often suffer from insufficient accuracy and an inability to fully reflect the various characteristics of curtain walls. For example, in the construction of geometric shapes, it is difficult to accurately simulate the subtle changes of complex curved curtain walls; in terms of the representation of material and thermal characteristics, there is a lack of effective integration methods, resulting in generated models that cannot meet the needs of actual engineering projects for curtain wall performance analysis, construction guidance, and post-maintenance. With the development of the construction industry, the demand for refined curtain wall models is increasing, and there is an urgent need for a method that can comprehensively consider multiple factors and generate high-precision curtain wall models with rich information. Summary of the Invention

[0005] To address the aforementioned technical problems, the present invention aims to provide a method for generating a detailed curtain wall model, comprising the following steps:

[0006] Step s1: Collect multi-source heterogeneous data of the target building and user intervention commands, and perform data preprocessing on the multi-source heterogeneous data, which includes high-precision point cloud data, building outline, environmental data, material property data, thermal response data, wind load data and temperature stress data.

[0007] Step s2: Generate the curtain wall edge line based on the building outline, dynamically adjust the segmentation parameters of the curtain wall edge line in both directions and perform adaptive piecewise spline interpolation to generate a smooth curve, and generate the basic surface of the basic curtain wall based on the smooth curve;

[0008] Step s3: Construct a basic three-dimensional geometric model based on the basic surface of the curtain wall, generate an initial lattice surface by combining the dividing points of the curtain wall edge line, construct a thermal response model by combining the basic three-dimensional geometric model, material property data and thermal response data, obtain thermal characteristic data of each area, optimize the lattice surface of the basic three-dimensional geometric model based on the thermal characteristic data, generate a three-dimensional geometric model, and perform automated grouping and panel generation operations on the three-dimensional geometric model.

[0009] Step s4: Perform multimodal feature fusion on the 3D geometric model, thermal feature data and material property data to generate a refined 3D curtain wall model, and perform multiphysics coupling verification on the refined 3D curtain wall model.

[0010] Furthermore, the process of generating curtain wall edge lines based on building outlines, dynamically adjusting the segmentation parameters of curtain wall edge lines bidirectionally, and performing adaptive piecewise spline interpolation to generate smooth curves, and generating the basic surface of the foundation curtain wall based on these smooth curves includes:

[0011] Based on the building outline, parametric design methods are used to select target modeling elements to generate the curtain wall edge line. Parametric design is a powerful design tool that allows designers to control the geometry of the model by adjusting parameters. When generating the curtain wall edge line, spline curves, geometric shapes, etc., can be used as target modeling elements. For example, when designing a curtain wall with a smooth curved shape, by setting control points and adjusting parameters, spline curves can easily follow the unique outline of the building, ensuring that the curtain wall edge line perfectly matches the building's appearance. For some regular-shaped curtain wall parts, such as rectangular and triangular geometric shapes, it is more suitable. They can quickly construct the basic framework of the curtain wall edge line through simple parameter definitions, such as side length and angle. Based on the pre-processed wind load data, temperature stress data, and user intervention commands, the segmentation parameters of the curtain wall edge line are dynamically adjusted in both directions. Based on the segmentation parameters after the two-way dynamic adjustment, several segmentation points are marked on the curtain wall edge line. Key control points are selected on the curtain wall edge line, and adaptive piecewise spline interpolation is performed on the curtain wall edge line based on the key control points to generate a smooth curve.

[0012] Lofting is performed on several smooth curves, that is, connecting these curves along a certain path to generate a continuous basic curtain wall surface.

[0013] Furthermore, the process of marking several segmentation points on the edge line of the curtain wall according to the segmentation parameters includes:

[0014] Obtain the dynamic segmentation spacing corresponding to each initial node. Assume the dynamic segmentation spacing for node Pi is di. This spacing is relative to the baseline segmentation spacing and is adjusted according to the specific conditions at each node, reflecting the actual required segmentation density at that node. Starting from the initial node, select one endpoint of the curtain wall edge line as the starting point. Find the initial node closest to this endpoint as the first node to be processed, let's say P1. Starting from P1, according to its corresponding dynamic segmentation spacing d1, measure a distance d1 along the edge line of the curtain wall to mark a new point Q1. This point is the first segmentation point determined based on the dynamic segmentation spacing of P1. Then, mark other segmentation points sequentially. For the next initial node P2, starting from the already marked point Q1, measure a distance d2 along the edge line to mark a new segmentation point Q2. Continue in this manner for each initial node Pi, marking the previously marked segmentation point... Starting from the edge line, measure a distance di along the edge line and mark the corresponding dividing point Qi. When encountering the boundary or endpoint of the curtain wall edge line during the marking process, adjust the marking method. If the remaining edge line length is less than the dynamic dividing spacing di of the current node, use the remaining length as the dividing spacing and mark the last dividing point so that it is exactly located at the endpoint of the edge line. If it is found during the marking process that the dividing points corresponding to two adjacent initial nodes may overlap or the spacing is too small (less than the preset minimum spacing threshold), the position of the dividing point is fine-tuned, and the dividing spacing of one of the nodes is appropriately increased to ensure that there is a reasonable spacing between the dividing points and that it can accurately reflect the characteristics and design requirements of the curtain wall edge line.

[0015] Furthermore, the process of dynamically adjusting the segmentation parameters of the curtain wall edge line in both directions based on the preprocessed wind load data, temperature stress data, and user intervention commands includes:

[0016] Preset the baseline segmentation spacing, structural response weights, and curvature weights, and uniformly generate an initial point set along the curtain wall edge line based on the baseline segmentation spacing. Differential operations are performed on each initial node in the initial point set to obtain the curvature value k of each initial node. Wind load data and temperature stress data are mapped to the initial point set for multi-field coupling analysis to obtain the total stress coefficient of each initial node. Based on the total stress coefficient and curvature value of each initial node, as well as the structural response weight and curvature weight, the dynamic segmentation spacing corresponding to each initial node is obtained.

[0017] The formula for calculating the dynamic segmentation distance of each initial node is as follows:

[0018] ;

[0019] ;

[0020] ;

[0021] ;

[0022] ;

[0023] in, This represents the dynamic segmentation spacing of the initial node i. Indicates the baseline segmentation spacing. Represents the structural response weights. Indicates curvature weight, This represents the total stress coefficient at the initial node i. This represents the curvature value of the initial node i. Indicates the wind load stress coefficient. Indicates the temperature stress coefficient. Indicates the wind load sensitivity coefficient. Indicates the temperature stress sensitivity coefficient. Indicates the wind vibration coefficient. Indicates the coefficient of variation of wind pressure at height. Indicates the building shape coefficient. Indicates the elastic modulus of a material. This represents the coefficient of linear expansion of the material. Indicates temperature difference. This indicates the local annual extreme temperature;

[0024] Based on the user's intervention command, obtain the fixed threshold range of segmentation spacing corresponding to each initial node, determine whether the dynamic segmentation spacing corresponding to each initial node is within the corresponding fixed threshold range of segmentation spacing, if it is, mark the dynamic segmentation spacing of the initial node as the optimal segmentation spacing, if it is not, set the optimal segmentation spacing of the initial node according to the fixed threshold range of segmentation spacing, select the maximum segmentation spacing in the fixed threshold range of segmentation spacing as the optimal segmentation spacing, and construct the segmentation parameters based on the optimal segmentation spacing corresponding to each initial node.

[0025] Furthermore, the process of selecting key control points along the curtain wall edge line and then performing adaptive piecewise spline interpolation on the curtain wall edge line based on these key control points to generate a smooth curve includes:

[0026] The curvature values ​​and total stress coefficients of each segmentation point on the edge of the curtain wall are weighted and averaged to obtain the criticality coefficients of each segmentation point. A criticality coefficient threshold is preset, and the criticality coefficients of each segmentation point are compared with the criticality coefficient threshold. Segmentation points with criticality coefficients greater than the criticality coefficient threshold are marked as critical control points. Curvature change analysis and total stress change analysis are performed on adjacent critical control points to obtain the curvature change coefficient and total stress change coefficient between adjacent critical control points.

[0027] The thresholds corresponding to the curvature variation coefficient and the total stress variation coefficient are preset. It is determined whether the curvature variation coefficient and the total stress variation coefficient between adjacent key control points are both less than the corresponding thresholds. If they are both less than the thresholds, a smooth curve is generated between adjacent key control points using cubic spline interpolation. If they are not both less than the thresholds, a smooth curve is generated between adjacent key control points using B-spline interpolation.

[0028] Error analysis and iterative optimization are performed on the smooth curves between adjacent key control points to extract several discrete points on the curtain wall edge between adjacent key control points. Calculate the minimum distance of the smooth curve from the key control point to the adjacent key control point. Set the maximum error threshold. If the error exceeds the maximum error threshold, a node insertion algorithm (such as the Cox-deBoor algorithm) is used to add new nodes and control points, and the interpolation is repeated until the error of all discrete points is reached. Optimization stops when the error is less than or equal to the maximum error threshold, or when the number of iterations reaches the preset upper limit.

[0029] Curvature variation analysis and total stress variation analysis are performed on adjacent critical control points to obtain the calculation formulas for the curvature variation coefficient and total stress variation coefficient between adjacent critical control points:

[0030] Set adjacent critical control points and The total stress coefficients at the locations are respectively and The curvature values ​​are respectively and ;

[0031] ;

[0032] ;

[0033] in, Represents the coefficient of curvature variation. This represents the coefficient of variation of total stress.

[0034] Furthermore, by combining the dividing points of the curtain wall edge line, an initial lattice surface is generated, and the lattice surface of the basic three-dimensional geometric model is optimized. The process of generating the three-dimensional geometric model includes:

[0035] Extract the surface parameters of the basic curtain wall surface and map them to the physical coordinate system to construct a basic three-dimensional geometric model. Use the nearest point projection algorithm to map the dividing points of the curtain wall edge line to the basic curtain wall surface to generate an initial lattice surface. Perform Delaunay triangulation on the initial lattice surface to generate several initial triangular meshes.

[0036] Thermal characteristic data includes thermal gradient, heat flux density, and temperature fluctuation frequency. Based on the thermal characteristic data of each region, evaluation indicators are set, indicator weights are set, and the membership matrix of each region to the preset thermal sensitivity level is obtained through fuzzy comprehensive evaluation.

[0037] The thermal sensitivity level of each region is obtained based on the membership matrix and index weights. A thermal sensitivity level threshold is preset, and regions with thermal sensitivity levels greater than the threshold are marked as thermal sensitive regions. The mesh density is analyzed based on the thermal characteristic data of the thermal sensitive regions to obtain the optimized mesh density of the thermal sensitive regions. Based on the optimized mesh density, several initial triangular meshes in the thermal sensitive regions of the basic 3D geometric model are re-divided to generate a 3D geometric model.

[0038] Furthermore, the process of obtaining the optimized mesh density for the heat-sensitive region by performing mesh density analysis based on the thermal characteristic data of the heat-sensitive region includes:

[0039] ;

[0040] in, To optimize the grid size, Indicates the initial grid size. Represents the thermal gradient. Indicates the maximum thermal gradient in the district. Represents heat flux density, Indicates the thermal conductivity of a material. This indicates the frequency of temperature fluctuations (unit: Hz, such as fluctuations caused by periodic solar radiation). This represents the frequency correlation coefficient (typical value 0.1~0.3, time step matching required). , and The above formulas are all numerical calculations after removing the dimensions of the weights. The formulas are obtained by software simulation based on a large amount of data and are closest to the real situation. The preset parameters and preset thresholds in the formulas are set by those skilled in the art according to the actual situation or obtained by simulation based on a large amount of data.

[0041] Furthermore, the process of constructing a thermal response model by combining the basic 3D geometric model, material property data, and thermal response data, and obtaining thermal characteristic data for each region, includes:

[0042] The material properties and thermal response data of each initial triangular mesh in the basic 3D geometric model are collected. Based on the material properties data of each initial triangular mesh, mesh optimization preprocessing is performed on each initial triangular mesh to obtain each triangular mesh after mesh optimization preprocessing. A thermal response model is constructed by combining the material properties data and thermal response data of each triangular mesh. The thermal response model is verified and optimized to obtain the verified and optimized thermal response model.

[0043] The thermal characteristic data of each region of the basic three-dimensional geometric model are output based on the thermal response model.

[0044] The process of performing mesh optimization preprocessing on each initial triangular mesh based on its material property data includes:

[0045] The mesh size of each initial triangular mesh is adjusted based on its material property data. The calculation formula is as follows:

[0046] ;

[0047] in, This indicates the adjusted grid size. This indicates the grid size before adjustment;

[0048] Furthermore, the mathematical model system required to construct the thermal response model by combining the material property data and thermal response data of each triangular mesh is as follows:

[0049] Transient heat conduction governing equations:

[0050] ;

[0051] in, Let t be the temperature field and t be the time field. For density, For specific heat capacity, Thermal conductivity, It is a heat source (such as heat generated by solar radiation).

[0052] Outer surface solar radiation boundary:

[0053] ;

[0054] in, For surface absorption rate, Solar irradiance, For emission rate, The Stefan-Boltzmann constant is... The external convective heat transfer coefficient is... The ambient temperature.

[0055] Inner surface convection boundary:

[0056] ;

[0057] in, The internal surface convective heat transfer coefficient is... Indoor air temperature.

[0058] Radiative heat transfer boundary:

[0059] ;

[0060] in, The surface radiation angle coefficient is denoted as , which takes into account radiation exchange with the surrounding environment (such as the sky and adjacent buildings).

[0061] Thermal gradient calculation:

[0062] ;

[0063] Heat flux density calculation:

[0064] ;

[0065] in, For fluid velocity, The surrounding surface temperature;

[0066] Temperature fluctuation frequency analysis:

[0067] ;

[0068] in, The characteristic thickness is used to estimate the frequency of a material's response to temperature fluctuations.

[0069] Furthermore, the thermal response model is validated and optimized. The process of obtaining a validated and optimized thermal response model includes:

[0070] ;

[0071] in, This indicates the parameter to be optimized (such as the internal surface convective heat transfer coefficient). Surface absorption rate Surface convective heat transfer coefficient By adjusting these parameters, the simulation results can be matched with the measured data. Indicates based on the current parameter The i-th simulated temperature value obtained through the thermal response model. The i-th measured temperature data point, where n represents the total number of data points used for calibration, including combinations of time series or spatial locations, is continuously optimized and adjusted using Bayesian methods. The process aims to minimize the error and stops when the error converges or reaches the preset accuracy.

[0072] Furthermore, the process of automating the grouping and panel generation of the 3D geometric model includes:

[0073] The process iterates through the lattice points of the 3D geometric model and groups them according to the smallest grid vertex. In the generated lattice surface, each lattice point is traversed one by one and grouped according to the rule of the smallest grid vertex, that is, adjacent vertices that constitute the smallest grid unit are divided into a group. This facilitates the independent processing of each small area and also conforms to the logic of curtain wall panel layout, because each group usually corresponds to the installation area of ​​a panel. Several lattice groups are obtained, the geometric features of each lattice group are extracted, and adaptive panel matching is performed based on the geometric features of each lattice group. Panels are selected from the preset panel library and embedded into the corresponding lattice group positions.

[0074] Furthermore, the process of adaptive panel matching based on the geometric features of each lattice group includes:

[0075] Obtain the geometric features of the dot matrix group to obtain the constraints of the panel. These constraints include panel size limitations, joint width requirements, and structural support positions. Panels meeting the constraints are selected from a pre-set panel library. The panels are then matched with the dot matrix group for similarity, and the panel with the highest similarity is embedded into the corresponding dot matrix group position. The similarity matching calculation formula is as follows:

[0076] ;

[0077] Where simer represents the similarity, mdr represents the overlap area between the panel and the dot matrix group, mdq represents the area of ​​the dot matrix group, and mds represents the area of ​​the panel.

[0078] Furthermore, the 3D geometric model, thermal feature data, and material property data are fused using multimodal features to generate a refined 3D curtain wall model. The process of multiphysics coupling verification of the refined 3D curtain wall model includes:

[0079] Based on the thermal response model, output the thermal feature data of each panel in the three-dimensional geometric model, extract the material property data of each panel in the three-dimensional geometric model, and map the thermal feature data and material property data of each panel to the corresponding area in the three-dimensional geometric model to generate a refined three-dimensional curtain wall model.

[0080] Historical multi-source heterogeneous data of the target building are collected. A curtain wall heat transfer prediction model is constructed based on the historical multi-source heterogeneous data. The 3D curtain wall refined model is then verified by multiphysics coupling based on the curtain wall heat transfer prediction model. The model parameters of the 3D curtain wall refined model are corrected based on the verification results. The 3D curtain wall refined model with completed multiphysics coupling verification is output. The 3D curtain wall refined model is used to generate temperature field distribution data (e.g., temperature cloud map of thermal bridge area of ​​metal frame, temperature gradient between glass layers, hourly surface temperature curve), heat flux density data (e.g., heat transfer rate per unit area, heat flux density between glass and air layers), stress and strain data (e.g., thermal expansion stress caused by temperature gradient, maximum principal stress at the edge of glass panel, mechanical load response, structural deformation under wind pressure), etc.

[0081] The process of performing multiphysics coupling verification on the refined 3D curtain wall model based on the curtain wall heat transfer prediction model includes:

[0082] The coupling physical fields are determined, including heat conduction, convection, and radiation. Thermal feature data and boundary conditions are extracted from multi-source heterogeneous data and input into the 3D curtain wall refinement model and the curtain wall heat transfer prediction model. The multi-physics coupling simulation results are obtained based on the curtain wall heat transfer prediction model. The data to be verified is output based on the 3D curtain wall refinement model. Parameters corresponding to the data to be verified, such as curtain wall surface temperature and heat flux density, are extracted from the multi-physics coupling simulation results.

[0083] The parameters in the multiphysics coupling simulation results are compared with the corresponding parameters in the data to be verified. The root mean square error of the parameters is obtained. An error threshold is preset. Parameters with root mean square errors greater than the error threshold are marked as parameters to be corrected. For example, if the surface temperature of a certain panel curtain wall output by the 3D curtain wall refinement model is generally higher than the measured value output by the curtain wall heat transfer prediction model, it may be necessary to reduce the thermal conductivity of the curtain wall material or change the mesh density of the area to which the panel belongs.

[0084] Subsequently, a sensitivity analysis was performed on the refined 3D curtain wall model to determine the weight coefficients of each model parameter for the parameter to be corrected. Common model parameters include the thermal conductivity, specific heat capacity, emissivity, convective heat transfer coefficient, and mesh density of the curtain wall material. Based on the weight coefficients of each model parameter for the parameter to be corrected, a multi-objective genetic algorithm was used to search for the optimal combination of model parameters to minimize the root mean square error corresponding to the parameter to be corrected. The model parameters in the optimal combination of model parameters were then substituted into the refined 3D curtain wall model, and multiphysics coupling verification was performed again.

[0085] Repeat the above process until the root mean square error meets the preset accuracy requirements, and output the refined three-dimensional curtain wall model that has completed multi-physics coupling verification.

[0086] Furthermore, the process of constructing a curtain wall heat transfer prediction model based on historical multi-source heterogeneous data includes:

[0087] A physical model is constructed through multiphysics coupling analysis. This physical model includes a heat transfer model (based on Fourier's law, establishing heat conduction, convection, and radiation models of the curtain wall), a fluid flow model (using computational fluid dynamics (CFD) to simulate airflow and heat exchange on the curtain wall surface (such as natural convection and wind pressure-driven airflow), and a structural stress model (analyzing material expansion or contraction stress caused by temperature changes). The received thermal characteristic data from the physical model is calculated using a coupling algorithm to generate multiphysics simulation results. The coupling algorithm includes strong coupling: iteratively solving the equations of each physical field (e.g., heat transfer affects material properties, and material deformation reacts to the heat conduction path); and weak coupling: solving each physical field sequentially (e.g., calculating heat distribution first, then importing temperature loads into structural analysis). The model is constructed based on deep learning. The wall heat transfer prediction model preprocesses and extracts features from historical multi-source heterogeneous data to obtain thermal feature data and boundary conditions (e.g., thermal boundaries: outdoor meteorological data (temperature, solar radiation), indoor air conditioning load; structural boundaries: thermal expansion coefficient of curtain wall material, mechanical stress constraints). This thermal feature data and boundary conditions are used as the training set for the curtain wall heat transfer prediction model. The training set is input into the model for training until the loss function stabilizes, and the model parameters are saved. Simultaneously, the thermal feature data and boundary conditions are input into a physical model. Based on the physical model, multiphysics simulation results are output. The curtain wall heat transfer prediction model is tested using these multiphysics simulation results until it meets preset requirements, at which point the final curtain wall heat transfer prediction model is output.

[0088] like Figure 2 As shown, a curtain wall fine model generation system includes a cloud platform, which is connected to a data acquisition module, a curtain wall generation module, an optimization processing module, and a model generation module via communication.

[0089] The data acquisition module is used to collect multi-source heterogeneous data of the target building and user intervention commands, and to perform data preprocessing on the multi-source heterogeneous data, which includes high-precision point cloud data, building outline, environmental data, material property data, thermal response data, wind load data and temperature stress data.

[0090] The curtain wall generation module is used to generate curtain wall edge lines based on the building outline, dynamically adjust the segmentation parameters of the curtain wall edge lines in both directions and perform adaptive piecewise spline interpolation to generate smooth curves, and generate the basic surface of the basic curtain wall based on the smooth curves.

[0091] The optimization processing module is used to construct a basic three-dimensional geometric model based on the basic surface of the curtain wall, generate an initial lattice surface by combining the dividing points of the curtain wall edge line, construct a thermal response model by combining the basic three-dimensional geometric model, material property data and thermal response data, obtain thermal characteristic data of each area, optimize the lattice surface of the basic three-dimensional geometric model based on the thermal characteristic data, generate a three-dimensional geometric model, and perform automated grouping and panel generation operations on the three-dimensional geometric model.

[0092] The model generation module is used to perform multimodal feature fusion of 3D geometric model, thermal feature data and material property data to generate a refined 3D curtain wall model, and to perform multiphysics coupling verification on the refined 3D curtain wall model.

[0093] Compared with the prior art, the beneficial effects of the present invention are:

[0094] 1. Advantages of Multi-Source Data Fusion: This invention extensively collects heterogeneous data from multiple sources, including high-precision point cloud data, building outlines, environmental data, material property data, thermal response data, wind load data, and temperature stress data. Compared to traditional methods that rely on only a single or limited number of data types, this comprehensive data fusion can more accurately reflect the various characteristics and behaviors of curtain walls in real-world environments. For example, by integrating thermal response data and wind load data, the impact of wind on heat transfer in curtain walls can be comprehensively analyzed, thereby more accurately assessing the performance of curtain walls under different climatic conditions and improving the model's ability to simulate complex real-world scenarios.

[0095] 2. Structural and Aesthetic Advantages of Dynamic Segmentation and Interpolation: The segmentation parameters of the curtain wall edge are dynamically adjusted bidirectionally. Segmentation points are determined based on wind load data, temperature stress data, and user intervention commands. This allows for the densification of segmentation points in areas of structural stress concentration, enhancing the structural safety of the curtain wall and effectively reducing the risk of cracking caused by thermal bridging and wind forces. Simultaneously, adaptive piecewise spline interpolation generates smooth curves. In areas with large curvature changes, it intelligently selects cubic spline or B-spline interpolation methods, ensuring excellent smoothness of the generated basic curtain wall surface. This greatly satisfies the high aesthetic requirements of architectural design for curtain wall surfaces and facilitates subsequent construction and processing, reducing material waste and installation difficulties caused by uneven surfaces.

[0096] 3. Advantages of Thermal Feature-Driven Optimization: A thermal response model is constructed by combining a basic 3D geometric model, material property data, and thermal response data. The lattice surfaces of the basic 3D geometric model are then optimized based on thermal feature data. In thermally sensitive areas such as high thermal gradients and high radiation, targeted increases in mesh density through thermal feature-based mesh density analysis can reduce the heat flux density calculation error from the conventional 8% to approximately 2%, significantly improving the accuracy of heat transfer simulation. This helps to accurately identify thermally weak points in the curtain wall, providing strong support for energy-saving design. For example, appropriately adjusting material configuration in highly heat-sensitive areas can reduce summer air conditioning load by 15%-20%, effectively reducing building energy consumption.

[0097] 4. Efficiency and cost advantages of automated grouping and panel generation: Automated grouping and panel generation of 3D geometric models are performed. The system traverses lattice points and faces, grouping them according to the smallest grid vertex. Panels are adaptively matched from a pre-set panel library based on the geometric characteristics of the lattice groups. Compared to traditional manual design and panel selection methods, this significantly improves design efficiency. Simultaneously, the pre-set panel library reduces the need for customized processing, increases material utilization, and effectively saves costs.

[0098] 5. Advantages of Multiphysics Coupling Verification and Model Correction: Multiphysics coupling verification is performed on the refined 3D curtain wall model. A curtain wall heat transfer prediction model is constructed based on historical multi-source heterogeneous data to verify and correct the refined model. Compared to traditional single-physics analysis or models lacking effective verification and correction mechanisms, this method can more comprehensively consider the interactions between multiple physical fields such as heat, structure, and fluid, and identify potential problems that traditional methods cannot detect, such as sealant failure caused by temperature stress. This greatly improves the reliability and applicability of the model, providing a more accurate basis for the maintenance and management of the curtain wall throughout its entire life cycle.

[0099] 6. Predictive Advantages of Hybrid Model Construction: A physical model is constructed through multiphysics coupling analysis, and a curtain wall heat transfer prediction model is built by combining deep learning. This hybrid model construction method combines the interpretability of the physical model with the powerful learning ability and rapid prediction capability of the deep learning model. Attached Figure Description

[0100] Figure 1 This is a schematic diagram of a method for generating a detailed curtain wall model according to an embodiment of this application.

[0101] Figure 2 This is a schematic diagram of a curtain wall fine model generation system according to an embodiment of this application. Detailed Implementation

[0102] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this application. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0103] like Figure 1 As shown, a method for generating a detailed curtain wall model includes the following steps:

[0104] Step s1: Collect multi-source heterogeneous data of the target building and user intervention commands, and perform data preprocessing on the multi-source heterogeneous data, which includes high-precision point cloud data, building outline, environmental data, material property data, thermal response data, wind load data and temperature stress data.

[0105] Step s2: Generate the curtain wall edge line based on the building outline, dynamically adjust the segmentation parameters of the curtain wall edge line in both directions and perform adaptive piecewise spline interpolation to generate a smooth curve, and generate the basic surface of the basic curtain wall based on the smooth curve;

[0106] Step s3: Construct a basic three-dimensional geometric model based on the basic surface of the curtain wall, generate an initial lattice surface by combining the dividing points of the curtain wall edge line, construct a thermal response model by combining the basic three-dimensional geometric model, material property data and thermal response data, obtain thermal characteristic data of each area, optimize the lattice surface of the basic three-dimensional geometric model based on the thermal characteristic data, generate a three-dimensional geometric model, and perform automated grouping and panel generation operations on the three-dimensional geometric model.

[0107] Step s4: Perform multimodal feature fusion on the 3D geometric model, thermal feature data and material property data to generate a refined 3D curtain wall model, and perform multiphysics coupling verification on the refined 3D curtain wall model.

[0108] The process of collecting multi-source heterogeneous data of the target building includes: obtaining the building outline, wind load sensitivity coefficient, temperature stress sensitivity coefficient, wind vibration coefficient, wind pressure height variation coefficient, building shape coefficient, material elastic modulus, and material linear expansion coefficient based on the architectural design drawings; acquiring high-precision point cloud data of the target building using a laser scanner or drone; simultaneously acquiring material reflection, refraction characteristics, and thermal response data of the target building using multispectral and thermal imaging technologies; and performing data preprocessing on the collected multi-source heterogeneous data, including noise reduction to remove outliers caused by environmental interference and other factors, followed by calibration to ensure the consistency of data collected by different devices in spatial coordinates.

[0109] It should be further explained that, in the specific implementation process, the curtain wall edge line is generated based on the building outline, and the segmentation parameters of the curtain wall edge line are dynamically adjusted in both directions and adaptive piecewise spline interpolation is performed to generate a smooth curve. The process of generating the basic surface of the curtain wall based on the smooth curve includes:

[0110] Based on the building outline, parametric design methods are used to select target modeling elements to generate the curtain wall edge line. Parametric design is a powerful design tool that allows designers to control the geometry of the model by adjusting parameters. When generating the curtain wall edge line, spline curves, geometric shapes, etc., can be used as target modeling elements. For example, when designing a curtain wall with a smooth curved shape, by setting control points and adjusting parameters, spline curves can easily follow the unique outline of the building, ensuring that the curtain wall edge line perfectly matches the building's appearance. For some regular-shaped curtain wall parts, such as rectangular and triangular geometric shapes, it is more suitable. They can quickly construct the basic framework of the curtain wall edge line through simple parameter definitions, such as side length and angle. Based on the pre-processed wind load data, temperature stress data, and user intervention commands, the segmentation parameters of the curtain wall edge line are dynamically adjusted in both directions. Based on the segmentation parameters after the two-way dynamic adjustment, several segmentation points are marked on the curtain wall edge line. Key control points are selected on the curtain wall edge line, and adaptive piecewise spline interpolation is performed on the curtain wall edge line based on the key control points to generate a smooth curve.

[0111] Lofting is performed on several smooth curves, that is, connecting these curves along a certain path to generate a continuous basic curtain wall surface.

[0112] It should be further explained that, in the specific implementation process, the process of marking several dividing points on the edge line of the curtain wall according to the dividing parameters includes:

[0113] Obtain the dynamic segmentation spacing corresponding to each initial node. Assume the dynamic segmentation spacing for node Pi is di. This spacing is adjusted relative to the baseline segmentation spacing based on the specific conditions at each node, reflecting the actual required segmentation density at that node. Starting from the initial node, select one endpoint of the curtain wall edge line as the starting point. Find the initial node closest to this endpoint as the first node to be processed, let's say P1. Starting from P1, measure a distance d1 along the edge line based on its corresponding dynamic segmentation spacing d1, marking a new point Q1. This point is the first segmentation point determined based on the dynamic segmentation spacing of P1. Then, mark other segmentation points sequentially. For the next initial node P2, starting from the already marked point Q1, measure a distance d2 along the edge line to mark a new segmentation point Q2. Continue this process for each initial node Pi, starting from the previously marked segmentation point Qi−1, measuring a distance di along the edge line to mark the corresponding segmentation point Qi. When encountering the boundary or endpoint of the curtain wall edge line during the marking process, the marking method is adjusted. If the remaining edge line length is less than the dynamic segmentation spacing di of the current node, the remaining length is used as the segmentation spacing to mark the last segmentation point, ensuring it is located exactly at the endpoint of the edge line. If, during the marking process, it is found that the segmentation points corresponding to two adjacent initial nodes may overlap or the spacing is too small (less than the preset minimum spacing threshold), the position of the segmentation point is fine-tuned, and the segmentation spacing of one of the nodes is appropriately increased to ensure a reasonable spacing between the segmentation points and to accurately reflect the characteristics and design requirements of the curtain wall edge line.

[0114] It should be further explained that, in the specific implementation process, the process of dynamically adjusting the segmentation parameters of the curtain wall edge line in both directions based on the pre-processed wind load data, temperature stress data, and user intervention commands includes:

[0115] Preset the baseline segmentation spacing, structural response weights, and curvature weights, and uniformly generate an initial point set along the curtain wall edge line based on the baseline segmentation spacing. Differential operations are performed on each initial node in the initial point set to obtain the curvature value k of each initial node. Wind load data and temperature stress data are mapped to the initial point set for multi-field coupling analysis to obtain the total stress coefficient of each initial node. Based on the total stress coefficient and curvature value of each initial node, as well as the structural response weight and curvature weight, the dynamic segmentation spacing corresponding to each initial node is obtained.

[0116] The formula for calculating the dynamic segmentation distance of each initial node is as follows:

[0117] ;

[0118] ;

[0119] ;

[0120] ;

[0121] ;

[0122] in, This represents the dynamic segmentation spacing of the initial node i. Indicates the baseline segmentation spacing. Represents the structural response weights. Indicates curvature weight, This represents the total stress coefficient at the initial node i. This represents the curvature value of the initial node i. Indicates the wind load stress coefficient. Indicates the temperature stress coefficient. Indicates the wind load sensitivity coefficient. Indicates the temperature stress sensitivity coefficient. Indicates the wind vibration coefficient. Indicates the coefficient of variation of wind pressure at height. Indicates the building shape coefficient. Indicates the elastic modulus of a material. This represents the coefficient of linear expansion of the material. Indicates temperature difference. This indicates the local annual extreme temperature;

[0123] Based on user intervention commands, the fixed threshold range of segmentation spacing corresponding to each initial node is obtained. It is then determined whether the dynamic segmentation spacing corresponding to each initial node is within the corresponding fixed threshold range. If it is, the dynamic segmentation spacing of the initial node is marked as the optimal segmentation spacing. If not, the optimal segmentation spacing of the initial node is set according to the fixed threshold range. The maximum segmentation spacing in the fixed threshold range is selected as the optimal segmentation spacing. Segmentation parameters are constructed based on the optimal segmentation spacing corresponding to each initial node. The density of segmentation points is dynamically adjusted based on curvature analysis to ensure the accuracy of surface fitting. Subsequently, structural factors, such as the impact of wind load and temperature changes on the curtain wall, are further considered to optimize the position of segmentation points, reduce stress concentration, improve the stability of the curtain wall's refined model structure, and enhance the accuracy of subsequent analysis.

[0124] It should be further explained that, in the specific implementation process, the process of selecting key control points on the curtain wall edge line and performing adaptive piecewise spline interpolation on the curtain wall edge line based on the key control points to generate a smooth curve includes:

[0125] The curvature values ​​and total stress coefficients of each segmentation point on the edge of the curtain wall are weighted and averaged to obtain the criticality coefficients of each segmentation point. A criticality coefficient threshold is preset, and the criticality coefficients of each segmentation point are compared with the criticality coefficient threshold. Segmentation points with criticality coefficients greater than the criticality coefficient threshold are marked as critical control points. Curvature change analysis and total stress change analysis are performed on adjacent critical control points to obtain the curvature change coefficient and total stress change coefficient between adjacent critical control points.

[0126] The system presets thresholds for the curvature variation coefficient and the total stress variation coefficient. It then determines whether the curvature variation coefficient and the total stress variation coefficient between adjacent key control points are both less than the corresponding thresholds. If both are less than the thresholds, a smooth curve is generated between adjacent key control points using cubic spline interpolation. Cubic spline interpolation is a commonly used spline interpolation method that uses a cubic polynomial to approximate the curve in each small interval and ensures the continuity of the first and second derivatives of the curve at the control points, thus generating a smooth curve. If neither is less than the thresholds, a smooth curve is generated between adjacent key control points using B-spline interpolation. B-spline interpolation controls the local shape of the curve through node vectors, supports piecewise polynomials, and is suitable for complex geometry.

[0127] Error analysis and iterative optimization are performed on the smooth curves between adjacent key control points to extract several discrete points on the curtain wall edge between adjacent key control points. Calculate the minimum distance of the smooth curve from the key control point to the adjacent key control point. Set the maximum error threshold. If the error exceeds the maximum error threshold, a node insertion algorithm (such as the Cox-deBoor algorithm) is used to add new nodes and control points, and the interpolation is repeated until the error of all discrete points is reached. Optimization stops when the error is less than or equal to the maximum error threshold, or when the number of iterations reaches the preset upper limit.

[0128] In regions with small changes in curvature and total stress, the relatively simple cubic spline interpolation method is used, avoiding the additional computational burden of the more complex B-spline interpolation method. Cubic spline interpolation is faster and requires fewer computational resources, improving computational efficiency and reducing computation time and cost while maintaining curve quality. In regions with large changes in curvature and total stress, B-spline interpolation, although more complex in its calculation process, effectively handles complex curve shapes, avoiding repeated adjustments and corrections caused by the inability of simpler methods to accurately fit the curve. Overall, this can save time and resources, improving design and production efficiency.

[0129] Curvature variation analysis and total stress variation analysis are performed on adjacent critical control points to obtain the calculation formulas for the curvature variation coefficient and total stress variation coefficient between adjacent critical control points:

[0130] Set adjacent critical control points and The total stress coefficients at the locations are respectively and The curvature values ​​are respectively and ;

[0131] ;

[0132] ;

[0133] in, Represents the coefficient of curvature variation. This represents the coefficient of variation of total stress.

[0134] It should be further explained that, in the specific implementation process, based on the dividing points of the curtain wall edge line, an initial lattice surface is generated, and the lattice surface of the basic three-dimensional geometric model is optimized. The process of generating the three-dimensional geometric model includes:

[0135] Extract the surface parameters of the basic curtain wall surface and map them to the physical coordinate system to construct a basic three-dimensional geometric model. Use the nearest point projection algorithm to map the dividing points of the curtain wall edge line to the basic curtain wall surface to generate an initial lattice surface. Perform Delaunay triangulation on the initial lattice surface to generate several initial triangular meshes.

[0136] Thermal characteristic data includes thermal gradient, heat flux density, and temperature fluctuation frequency. Based on the thermal characteristic data of each region, evaluation indicators are set, indicator weights are set, and the membership matrix of each region to the preset thermal sensitivity level is obtained through fuzzy comprehensive evaluation.

[0137] The thermal sensitivity level of each region is obtained based on the membership matrix and index weights. A thermal sensitivity level threshold is preset, and regions with thermal sensitivity levels greater than the threshold are marked as thermal sensitive regions. The mesh density is analyzed based on the thermal characteristic data of the thermal sensitive regions to obtain the optimized mesh density of the thermal sensitive regions. Based on the optimized mesh density, several initial triangular meshes in the thermal sensitive regions of the basic 3D geometric model are re-divided to generate a 3D geometric model.

[0138] It should be further explained that, in the specific implementation process, the process of obtaining the heat sensitivity level of each region based on the membership matrix and index weights includes:

[0139] The evaluation index weights and membership matrix of the evaluation index are fused by formula to obtain the fuzzy comprehensive evaluation matrix of the evaluation index. The membership degree of each region to different heat sensitivity levels is obtained according to the fuzzy comprehensive evaluation matrix. The heat sensitivity level with the highest membership degree for each region is selected and the heat sensitivity level with the highest membership degree for each region is taken as the heat sensitivity level of each region.

[0140] The formula is as follows:

[0141] ;

[0142] in, The fuzzy comprehensive evaluation matrix of the evaluation indicators is... The weights of the evaluation indicators are: For the membership matrix, " "" indicates that the elements at corresponding positions in the weight matrix and membership matrix of the evaluation index are multiplied together. The weighting parameter is used to control the balance between the weight matrix and the membership matrix in the fuzzy comprehensive evaluation matrix of the evaluation index.

[0143] The process of obtaining the optimal mesh density for heat-sensitive areas by performing mesh density analysis based on the thermal characteristic data of the heat-sensitive areas includes:

[0144] ;

[0145] in, To optimize the grid size, Indicates the initial grid size. Represents the thermal gradient. Indicates the maximum thermal gradient in the district. Represents heat flux density, Indicates the thermal conductivity of a material. This indicates the frequency of temperature fluctuations (unit: Hz, such as fluctuations caused by periodic solar radiation). This represents the frequency correlation coefficient (typical value 0.1~0.3, time step matching required). , and The above formulas are all numerical calculations after removing the dimensions of the weights. The formulas are obtained by software simulation based on a large amount of data and are closest to the real situation. The preset parameters and preset thresholds in the formulas are set by those skilled in the art according to the actual situation or obtained by simulation based on a large amount of data.

[0146] For steady-state problems, this formula is based on... and For primary problems, and for transient problems, consider both... It is compatible with both steady-state and transient thermal analysis requirements, achieving precise matching between grid density and thermal characteristics, and significantly improving the reliability of curtain wall heat transfer prediction.

[0147] It should be further explained that, in the specific implementation process, the process of constructing a thermal response model by combining the basic 3D geometric model, material property data, and thermal response data, and obtaining thermal characteristic data for each region includes:

[0148] The material properties and thermal response data of each initial triangular mesh in the basic 3D geometric model are collected. Based on the material properties data of each initial triangular mesh, mesh optimization preprocessing is performed on each initial triangular mesh to obtain each triangular mesh after mesh optimization preprocessing. A thermal response model is constructed by combining the material properties data and thermal response data of each triangular mesh. The thermal response model is verified and optimized to obtain the verified and optimized thermal response model.

[0149] The thermal characteristic data of each region of the basic three-dimensional geometric model are output based on the thermal response model.

[0150] The process of performing mesh optimization preprocessing on each initial triangular mesh based on its material property data includes:

[0151] The mesh size of each initial triangular mesh is adjusted based on its material property data. The calculation formula is as follows:

[0152] ;

[0153] in, This indicates the adjusted grid size. This indicates the grid size before adjustment;

[0154] It should be further explained that the mathematical model system required to construct the thermal response model by combining the material property data and thermal response data of each triangular mesh is as follows:

[0155] Transient heat conduction governing equations:

[0156] ;

[0157] in, Let t be the temperature field and t be the time field. For density, For specific heat capacity, Thermal conductivity, It is a heat source (such as heat generated by solar radiation).

[0158] Outer surface solar radiation boundary:

[0159] ;

[0160] in, For surface absorption rate, Solar irradiance, For emission rate, The Stefan-Boltzmann constant is... The external convective heat transfer coefficient is... The ambient temperature.

[0161] Inner surface convection boundary:

[0162] ;

[0163] in, The internal surface convective heat transfer coefficient is... Indoor air temperature.

[0164] Radiative heat transfer boundary:

[0165] ;

[0166] in, The surface radiation angle coefficient is denoted as , which takes into account radiation exchange with the surrounding environment (such as the sky and adjacent buildings).

[0167] Thermal gradient calculation:

[0168] ;

[0169] Heat flux density calculation:

[0170] ;

[0171] in, For fluid velocity, The surrounding surface temperature;

[0172] Temperature fluctuation frequency analysis:

[0173] ;

[0174] in, The characteristic thickness is used to estimate the frequency of a material's response to temperature fluctuations.

[0175] It should be further explained that the process of validating and optimizing the thermal response model to obtain a validated and optimized thermal response model includes:

[0176] ;

[0177] in, This indicates the parameter to be optimized (such as the internal surface convective heat transfer coefficient). Surface absorption rate Surface convective heat transfer coefficient By adjusting these parameters, the simulation results can be matched with the measured data. Indicates based on the current parameter The i-th simulated temperature value obtained through the thermal response model. The i-th measured temperature data point, where n represents the total number of data points used for calibration, including combinations of time series or spatial locations, is continuously optimized and adjusted using Bayesian methods. The process aims to minimize the error and stops when the error converges or reaches the preset accuracy.

[0178] It should be further explained that, in the specific implementation process, the automated grouping and panel generation of the 3D geometric model includes:

[0179] The process iterates through the lattice points of the 3D geometric model and groups them according to the smallest grid vertex. In the generated lattice surface, each lattice point is traversed one by one and grouped according to the rule of the smallest grid vertex, that is, adjacent vertices that constitute the smallest grid unit are divided into a group. This facilitates the independent processing of each small area and also conforms to the logic of curtain wall panel layout, because each group usually corresponds to the installation area of ​​a panel. Several lattice groups are obtained, the geometric features of each lattice group are extracted, and adaptive panel matching is performed based on the geometric features of each lattice group. Panels are selected from the preset panel library and embedded into the corresponding lattice group positions.

[0180] It should be further explained that, in the specific implementation process, the adaptive panel matching process based on the geometric features of each dot matrix group includes:

[0181] Obtain the geometric features of the dot matrix group to obtain the constraints of the panel. These constraints include panel size limitations, joint width requirements, and structural support positions. Panels meeting the constraints are selected from a pre-set panel library. The panels are then matched with the dot matrix group for similarity, and the panel with the highest similarity is embedded into the corresponding dot matrix group position. The similarity matching calculation formula is as follows:

[0182] ;

[0183] Where simer represents the similarity, mdr represents the overlap area between the panel and the dot matrix group, mdq represents the area of ​​the dot matrix group, and mds represents the area of ​​the panel.

[0184] It should be further explained that, in the specific implementation process, the three-dimensional geometric model, thermal feature data, and material property data are fused using multimodal features to generate a refined three-dimensional curtain wall model. The process of multiphysics coupling verification of the refined three-dimensional curtain wall model includes:

[0185] Based on the thermal response model, output the thermal feature data of each panel in the three-dimensional geometric model, extract the material property data of each panel in the three-dimensional geometric model, and map the thermal feature data and material property data of each panel to the corresponding area in the three-dimensional geometric model to generate a refined three-dimensional curtain wall model.

[0186] Historical multi-source heterogeneous data of the target building are collected. A curtain wall heat transfer prediction model is constructed based on the historical multi-source heterogeneous data. The 3D curtain wall refined model is then verified by multiphysics coupling based on the curtain wall heat transfer prediction model. The model parameters of the 3D curtain wall refined model are corrected based on the verification results. The 3D curtain wall refined model with completed multiphysics coupling verification is output. The 3D curtain wall refined model is used to generate temperature field distribution data (e.g., temperature cloud map of thermal bridge area of ​​metal frame, temperature gradient between glass layers, hourly surface temperature curve), heat flux density data (e.g., heat transfer rate per unit area, heat flux density between glass and air layers), stress and strain data (e.g., thermal expansion stress caused by temperature gradient, maximum principal stress at the edge of glass panel, mechanical load response, structural deformation under wind pressure), etc.

[0187] The process of performing multiphysics coupling verification on the refined 3D curtain wall model based on the curtain wall heat transfer prediction model includes:

[0188] The coupling physical fields are determined, including heat conduction, convection, and radiation. Thermal feature data and boundary conditions are extracted from multi-source heterogeneous data and input into the 3D curtain wall refinement model and the curtain wall heat transfer prediction model. The multi-physics coupling simulation results are obtained based on the curtain wall heat transfer prediction model. The data to be verified is output based on the 3D curtain wall refinement model. Parameters corresponding to the data to be verified, such as curtain wall surface temperature and heat flux density, are extracted from the multi-physics coupling simulation results.

[0189] The parameters in the multiphysics coupling simulation results are compared with the corresponding parameters in the data to be verified. The root mean square error of the parameters is obtained. An error threshold is preset. Parameters with root mean square errors greater than the error threshold are marked as parameters to be corrected. For example, if the surface temperature of a certain panel curtain wall output by the 3D curtain wall refinement model is generally higher than the measured value output by the curtain wall heat transfer prediction model, it may be necessary to reduce the thermal conductivity of the curtain wall material or change the mesh density of the area to which the panel belongs.

[0190] Subsequently, a sensitivity analysis was performed on the refined 3D curtain wall model to determine the weight coefficients of each model parameter for the parameter to be corrected. Common model parameters include the thermal conductivity, specific heat capacity, emissivity, convective heat transfer coefficient, and mesh density of the curtain wall material. Based on the weight coefficients of each model parameter for the parameter to be corrected, a multi-objective genetic algorithm was used to search for the optimal combination of model parameters to minimize the root mean square error corresponding to the parameter to be corrected. The model parameters in the optimal combination of model parameters were then substituted into the refined 3D curtain wall model, and multiphysics coupling verification was performed again.

[0191] Repeat the above process until the root mean square error meets the preset accuracy requirements, and output the refined three-dimensional curtain wall model that has completed multi-physics coupling verification.

[0192] The process of constructing a curtain wall heat transfer prediction model based on historical multi-source heterogeneous data includes:

[0193] A physical model is constructed through multiphysics coupling analysis. This physical model includes a heat transfer model (based on Fourier's law, establishing heat conduction, convection, and radiation models of the curtain wall), a fluid flow model (using computational fluid dynamics (CFD) to simulate airflow and heat exchange on the curtain wall surface (such as natural convection and wind pressure-driven airflow), and a structural stress model (analyzing material expansion or contraction stress caused by temperature changes). The received thermal characteristic data from the physical model is calculated using a coupling algorithm to generate multiphysics simulation results. The coupling algorithm includes strong coupling: iteratively solving the equations of each physical field (e.g., heat transfer affects material properties, and material deformation reacts to the heat conduction path); and weak coupling: solving each physical field sequentially (e.g., calculating heat distribution first, then importing temperature loads into structural analysis). The model is constructed based on deep learning. The wall heat transfer prediction model preprocesses and extracts features from historical multi-source heterogeneous data to obtain thermal feature data and boundary conditions (e.g., thermal boundaries: outdoor meteorological data (temperature, solar radiation), indoor air conditioning load; structural boundaries: thermal expansion coefficient of curtain wall material, mechanical stress constraints). This thermal feature data and boundary conditions are used as the training set for the curtain wall heat transfer prediction model. The training set is input into the model for training until the loss function stabilizes, and the model parameters are saved. Simultaneously, the thermal feature data and boundary conditions are input into a physical model. Based on the physical model, multiphysics simulation results are output. The curtain wall heat transfer prediction model is tested using these multiphysics simulation results until it meets preset requirements, at which point the final curtain wall heat transfer prediction model is output.

[0194] Developing a curtain wall heat transfer prediction model is a complex process involving multiple steps such as model selection, training, validation, and testing. The following is a detailed supplementary explanation of this process:

[0195] A Convolutional Neural Network (CNN) suitable for time series analysis was chosen as the deep learning architecture, and the cross-entropy loss function was selected as the optimization objective. The prepared training set was then input into the chosen deep learning model to begin training. During training, the weights were continuously updated using the backpropagation algorithm, causing the loss function to gradually decrease until a stable state was reached. During this period, techniques such as early stopping were used to avoid overfitting. In addition to the basic training process, grid search was used to fine-tune various parameters of the model, including the learning rate, batch size, and regularization coefficient.

[0196] Once the model training is complete and the parameters have been tuned, a final evaluation is performed using a test set to obtain the model's evaluation results. These results include classification metrics such as accuracy, recall, and F1 score. Based on the evaluation results on the test set, it is determined whether the model has met the expected standards. If the requirements are met, the model parameters are saved and deployment is prepared; otherwise, it is necessary to return to a previous stage to re-examine issues such as data quality, model structure, or training strategy.

[0197] like Figure 2 As shown, a curtain wall fine model generation system includes a cloud platform, which is connected to a data acquisition module, a curtain wall generation module, an optimization processing module, and a model generation module via communication.

[0198] The data acquisition module is used to collect multi-source heterogeneous data of the target building and user intervention commands, and to perform data preprocessing on the multi-source heterogeneous data, which includes high-precision point cloud data, building outline, environmental data, material property data, thermal response data, wind load data and temperature stress data.

[0199] The curtain wall generation module is used to generate curtain wall edge lines based on the building outline, dynamically adjust the segmentation parameters of the curtain wall edge lines in both directions and perform adaptive piecewise spline interpolation to generate smooth curves, and generate the basic surface of the basic curtain wall based on the smooth curves.

[0200] The optimization processing module is used to construct a basic three-dimensional geometric model based on the basic surface of the curtain wall, generate an initial lattice surface by combining the dividing points of the curtain wall edge line, construct a thermal response model by combining the basic three-dimensional geometric model, material property data and thermal response data, obtain thermal characteristic data of each area, optimize the lattice surface of the basic three-dimensional geometric model based on the thermal characteristic data, generate a three-dimensional geometric model, and perform automated grouping and panel generation operations on the three-dimensional geometric model.

[0201] The model generation module is used to perform multimodal feature fusion of 3D geometric model, thermal feature data and material property data to generate a refined 3D curtain wall model, and to perform multiphysics coupling verification on the refined 3D curtain wall model.

[0202] The above embodiments are only used to illustrate the technical methods of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical methods of the present invention without departing from the spirit and scope of the technical methods of the present invention.

Claims

1. A method for generating a detailed curtain wall model, characterized in that, Includes the following steps: Step s1: Collect multi-source heterogeneous data of the target building and user intervention commands, and perform data preprocessing on the multi-source heterogeneous data, which includes high-precision point cloud data, building outline, environmental data, material property data, thermal response data, wind load data and temperature stress data. Step s2: Generate the curtain wall edge line based on the building outline, dynamically adjust the segmentation parameters of the curtain wall edge line in both directions and perform adaptive piecewise spline interpolation to generate a smooth curve, and generate the basic surface of the basic curtain wall based on the smooth curve; Step s3: Construct a basic three-dimensional geometric model based on the basic surface of the curtain wall, map the dividing points of the curtain wall edge line onto the basic surface of the curtain wall to generate an initial lattice surface, triangulate the initial lattice surface to generate several initial triangular meshes, and construct a thermal response model by combining the basic three-dimensional geometric model, material property data and thermal response data to obtain thermal characteristic data of each region. Based on the thermal characteristic data of each region as the evaluation index, the index weights of the evaluation index are set, and the membership matrix of each region to the preset thermal sensitivity level is obtained through fuzzy comprehensive evaluation. The thermal sensitivity level of each region is obtained based on the membership matrix and index weights. A thermal sensitivity level threshold is preset. Regions with thermal sensitivity levels greater than the thermal sensitivity level threshold are marked as thermal sensitive regions. A grid density analysis based on thermal characteristics is performed on the thermal sensitive regions to obtain the optimized grid density of the thermal sensitive regions. Based on the optimized grid density, several initial triangular grids in the thermal sensitive regions of the basic 3D geometric model are re-divided to generate a 3D geometric model. Automated matrix point grouping and panel generation operations are performed on the 3D geometric model. The process of performing grid density analysis based on thermal characteristics in heat-sensitive areas to obtain the optimized grid density for these areas includes: ； in, To optimize the grid size, Indicates the initial grid size. Represents the thermal gradient. Indicates the maximum thermal gradient in the district. Represents heat flux density, Indicates the thermal conductivity of a material. Indicates the frequency of temperature fluctuations. Represents the frequency correlation coefficient. , and The weighting factor is represented by the formula, which is obtained by removing the dimension and taking its numerical value through software simulation based on a large amount of data. Step s4: Perform multimodal feature fusion on the 3D geometric model, thermal feature data and material property data to generate a refined 3D curtain wall model, and perform multiphysics coupling verification on the refined 3D curtain wall model.

2. The method for generating a refined curtain wall model according to claim 1, characterized in that, The process of generating curtain wall edge lines, dynamically adjusting the segmentation parameters of the curtain wall edge lines in both directions, and performing adaptive piecewise spline interpolation to generate smooth curves, and then generating the basic surface of the basic curtain wall based on the smooth curves, includes: Based on the building outline, a parametric design method is used to select target modeling elements to generate the curtain wall edge line. Based on the pre-processed wind load data, temperature stress data, and user intervention commands, the segmentation parameters of the curtain wall edge line are dynamically adjusted in both directions. Based on the segmentation parameters, several segmentation points are marked on the curtain wall edge line. Key control points are selected from the segmentation points. Based on the key control points, adaptive piecewise spline interpolation is performed on the curtain wall edge line to generate a smooth curve. Lofting operations are performed on several smooth curves to generate the basic surface of the foundation curtain wall.

3. The method for generating a refined curtain wall model according to claim 2, characterized in that, The process of dynamically adjusting the segmentation parameters of the curtain wall edge line based on preprocessed wind load data, temperature stress data, and user intervention commands includes: The baseline segmentation spacing, structural response weight, and curvature weight are preset. An initial point set is generated uniformly along the edge of the curtain wall based on the baseline segmentation spacing. The curvature value of each initial node is obtained. Wind load data and temperature stress data are mapped to the initial point set for multi-field coupling analysis. The total stress coefficient of each initial node is obtained. Based on the total stress coefficient and curvature value of each initial node, as well as the structural response weight and curvature weight, the dynamic segmentation spacing corresponding to each initial node is obtained. Based on the user's intervention instructions, obtain the fixed threshold range of segmentation spacing corresponding to each initial node, determine whether the dynamic segmentation spacing corresponding to each initial node is within the corresponding fixed threshold range of segmentation spacing, if it is, mark the dynamic segmentation spacing of the initial node as the optimal segmentation spacing, if it is not, set the optimal segmentation spacing of the initial node according to the fixed threshold range of segmentation spacing, and construct the segmentation parameters according to the optimal segmentation spacing corresponding to each initial node.

4. The method for generating a refined curtain wall model according to claim 3, characterized in that, The process of selecting key control points and generating a smooth curve by adaptive piecewise spline interpolation of the curtain wall edge line based on these key control points includes: The curvature values ​​and total stress coefficients of each segmentation point on the edge of the curtain wall are weighted and averaged to obtain the criticality coefficients of each segmentation point. A criticality coefficient threshold is preset, and segmentation points with criticality coefficients greater than the criticality coefficient threshold are marked as critical control points. Curvature change analysis and total stress change analysis are performed on adjacent critical control points to obtain the curvature change coefficient and total stress change coefficient between adjacent critical control points. The thresholds corresponding to the curvature variation coefficient and the total stress variation coefficient are preset. It is determined whether the curvature variation coefficient and the total stress variation coefficient between adjacent key control points are both less than the corresponding thresholds. If they are both less than the thresholds, a smooth curve is generated between adjacent key control points using cubic spline interpolation. If they are not both less than the thresholds, a smooth curve is generated between adjacent key control points using B-spline interpolation. Error analysis and iterative optimization are performed on the smooth curves between adjacent key control points.

5. The method for generating a refined curtain wall model according to claim 4, characterized in that, The process of constructing a thermal response model by combining a basic 3D geometric model, material property data, and thermal response data, and obtaining thermal characteristic data for each region, includes: The material properties and thermal response data of each initial triangular mesh in the basic 3D geometric model are collected. Based on the material properties data of each initial triangular mesh, mesh optimization preprocessing is performed on each initial triangular mesh to obtain each triangular mesh after mesh optimization preprocessing. A thermal response model is constructed by combining the material properties data and thermal response data of each triangular mesh. The thermal response model is verified and optimized to obtain the verified and optimized thermal response model. The thermal characteristic data of each region of the basic three-dimensional geometric model are output based on the thermal response model.

6. The method for generating a refined curtain wall model according to claim 5, characterized in that, The process of automating raster grouping and panel generation for 3D geometric models includes: Traverse the lattice points of the 3D geometric model, group them according to the smallest mesh vertex, obtain several lattice groups, extract the geometric features of each lattice group, perform adaptive panel matching based on the geometric features of each lattice group, select panels from the preset panel library and embed them into the corresponding lattice group positions.

7. The method for generating a refined curtain wall model according to claim 6, characterized in that, The process of generating a detailed 3D curtain wall model and performing multiphysics coupling verification on the detailed 3D curtain wall model includes: Based on the thermal response model, output the thermal feature data of each panel in the three-dimensional geometric model, extract the material property data of each panel in the three-dimensional geometric model, and map the thermal feature data and material property data of each panel to the corresponding area in the three-dimensional geometric model to generate a refined three-dimensional curtain wall model. Historical multi-source heterogeneous data of the target building are collected. A curtain wall heat transfer prediction model is constructed based on the historical multi-source heterogeneous data. The 3D curtain wall refined model is verified by multi-physics coupling based on the curtain wall heat transfer prediction model. The model parameters of the 3D curtain wall refined model are corrected based on the verification results.

8. The method for generating a refined curtain wall model according to claim 7, characterized in that, The process of constructing a curtain wall heat transfer prediction model based on historical multi-source heterogeneous data includes: A physical model is constructed through multiphysics coupling analysis, and a curtain wall heat transfer prediction model is built based on deep learning. Historical multi-source heterogeneous data is preprocessed and features are extracted to obtain thermal feature data and boundary conditions. The thermal feature data and boundary conditions are used as the training set for the curtain wall heat transfer prediction model. The training set is input into the curtain wall heat transfer prediction model for training until the loss function is stable and the model parameters are saved. At the same time, the thermal feature data and boundary conditions are input into the physical model, and multiphysics simulation results are output based on the physical model. The curtain wall heat transfer prediction model is tested through multiphysics simulation results until it meets the preset requirements, and then the curtain wall heat transfer prediction model is output.

9. A system for generating a detailed curtain wall model, specifically applied to the method for generating a detailed curtain wall model as described in any one of claims 1 to 8, characterized in that, This includes a cloud platform, which has communication connections to a data acquisition module, a curtain wall generation module, an optimization processing module, and a model generation module. The data acquisition module is used to collect multi-source heterogeneous data of the target building and user intervention commands, and to perform data preprocessing on the multi-source heterogeneous data, which includes high-precision point cloud data, building outline, environmental data, material property data, thermal response data, wind load data and temperature stress data. The curtain wall generation module is used to generate curtain wall edge lines based on the building outline, dynamically adjust the segmentation parameters of the curtain wall edge lines in both directions and perform adaptive piecewise spline interpolation to generate smooth curves, and generate the basic surface of the basic curtain wall based on the smooth curves. The optimization processing module is used to construct a basic three-dimensional geometric model based on the basic surface of the curtain wall, generate an initial lattice surface by combining the dividing points of the curtain wall edge line, construct a thermal response model by combining the basic three-dimensional geometric model, material property data and thermal response data, obtain thermal characteristic data of each area, optimize the lattice surface of the basic three-dimensional geometric model based on the thermal characteristic data, generate a three-dimensional geometric model, and perform automated lattice point grouping and panel generation operations on the three-dimensional geometric model. The model generation module is used to perform multimodal feature fusion of 3D geometric model, thermal feature data and material property data to generate a refined 3D curtain wall model, and to perform multiphysics coupling verification on the refined 3D curtain wall model.

Images