An existing building BEM modeling method, system, device and medium based on an unmanned aerial vehicle infrared thermal image

Abstract

The application provides a kind of existing building BEM modeling method, system, equipment and medium based on unmanned aerial vehicle infrared thermal image, belongs to building modeling technical field, obtains building image sequence by unmanned aerial vehicle aerial photography;Generate the three-dimensional geometric model of building main body;Through the analysis of the RGB value characteristics of the visible light point cloud, the outer window candidate area is selected, the window hole three-dimensional boundary is generated, the inner surface temperature and indoor air temperature are determined by integrating the inner surface infrared temperature and area weighting;Based on the temperature data and the construction parameters of the envelope structure, the heat transfer coefficient of the outer wall, roof and outer window is inverted, and the weighted average K value is combined to finally build the building energy consumption model.The method realizes the accurate operation from building geometric model construction to heat parameter acquisition, to building energy consumption model construction.Improves the efficiency and accuracy of existing building BEM modeling, and provides a scientific basis for building energy saving reconstruction, energy consumption evaluation, etc.

Description

Technical Field

[0001] This invention belongs to the field of architectural modeling technology, specifically relating to a method, system, equipment, and medium for BEM modeling of existing buildings based on UAV infrared thermal imaging. Background Technology

[0002] Building energy conservation has become a major focus. Buildings that have been built and put into use in cities account for the vast majority of the total number of buildings. They have a high proportion of energy consumption, but generally suffer from problems such as insufficient insulation of exterior walls and roofs, low airtightness of exterior windows, and low energy efficiency of equipment systems.

[0003] To achieve energy conservation in existing and operational buildings, a Building Energy Model (BEM) can be constructed to quantitatively analyze the actual energy consumption level of the building, assess the potential for energy-saving renovations, and provide a scientific basis for developing renovation plans and optimizing operational strategies.

[0004] Currently, existing building modeling (BEM) relies primarily on design drawings, on-site measurements, and empirical parameters. The resulting building models often fail to accurately reflect the building's true geometry, especially for structures with complex roofs and window distributions. This leads to inaccurate roof modeling; similarly, the identification and modeling of windows largely depend on on-site measurements and empirical parameters, making omissions or deviations likely.

[0005] Regarding model fusion, inconsistencies in data sources and coordinate systems between different parts of the model lead to gaps or misalignments in model splicing. Furthermore, the methods for obtaining thermal parameters of building envelopes are largely based on theoretical calculations or empirical values, resulting in significant discrepancies with the actual thermal performance of existing buildings. Consequently, they fail to accurately reflect the current actual thermal performance of existing buildings. Summary of the Invention

[0006] This invention provides a method for BEM modeling of existing buildings based on UAV infrared thermal imaging. The method automates and refines the entire process, from building the building's geometric model to acquiring thermal parameters and constructing the building's energy consumption model. This improves the efficiency and accuracy of BEM modeling of existing buildings and provides a scientific basis for building energy-saving retrofitting and energy consumption assessment.

[0007] The methods include:

[0008] S101: Obtain building image sequences through drone aerial photography and preprocess them to generate point cloud data of the building itself;

[0009] S102: Generate a three-dimensional geometric model of the building body based on the parameterized contour extraction, smoothing and solid extrusion of the building body point cloud data;

[0010] S103: Combine oblique photogrammetry data to identify point clouds in the roof area, extract roof ridge lines and elevation information, reconstruct the roof geometry, and perform boundary node fusion calibration with the main model to form a complete 3D model;

[0011] S104: By analyzing the RGB value characteristics of visible light point clouds, candidate regions for outer windows are selected, and the window opening point sets are grouped based on the spatial proximity clustering algorithm to generate the three-dimensional boundary of the window opening and extract the geometric contour of the window opening.

[0012] S105: Collect building exterior surface temperature data and calculate the average exterior surface temperature of the building envelope by combining orthorectified infrared imagery; determine the interior surface temperature and indoor air temperature by integrating interior surface infrared temperature measurement and area-weighted integration.

[0013] S106: Based on temperature data and building envelope construction parameters, the heat transfer coefficients of exterior walls, roofs and windows are inverted, and the weighted average K value of the overall building envelope is calculated by combining geometric model area data, and finally the building energy consumption model is constructed.

[0014] It should be further explained that step S102 specifically includes:

[0015] By comparing the spatial orientation deviation between the main building axis and the xyz axis of the world coordinate system, the overall rotation angle and translation parameters of the point cloud are adjusted to align the main building facade with the world coordinate axis.

[0016] Select the point cloud of the middle floor and project it onto the xy plane using the Grasshopper platform to generate a two-dimensional discrete point set; remove outliers from the projected point set and rearrange the points in clockwise or counterclockwise order to generate an initial closed polyline profile.

[0017] The initial closed polyline is smoothed by a preset number of iterations; by comparing the fit between the original point cloud and the smoothed curve, the smoothing parameters are adjusted to eliminate local jitter and jagged effects at the boundary and generate a closed boundary curve.

[0018] Based on the measured total height of the building, the smoothed closed boundary curve is extruded vertically along the positive z-axis to form a preliminary three-dimensional closed solid. The solid is then checked for holes or overlaps, and local defects are repaired to generate a three-dimensional geometric model of the building.

[0019] It should be further explained that step S103 specifically includes:

[0020] Before identifying the roof area point cloud by combining oblique photogrammetry data, the range of the roof point cloud is delineated by analyzing the elevation distribution characteristics of the overall building point cloud. Then, the slope threshold filtering algorithm is used to separate the roof tilt surface point cloud from the main building point cloud.

[0021] When extracting the roof ridge line, the roof point cloud is first meshed to generate a triangular mesh model. By identifying the abrupt change in the normal vector of adjacent triangular faces in the mesh model, the initial position of the roof ridge line is detected. Then, the ridge line breakpoints are connected and corrected by human interaction.

[0022] When obtaining roof elevation information, the three-dimensional coordinate data of the roof point cloud is used to generate an elevation contour map of the roof surface. By analyzing the distribution and direction of the contour lines, characteristic parameters such as the roof slope and drainage direction are determined.

[0023] When performing boundary node fusion calibration of the reconstructed roof geometry and the main model, Boolean operations are used to find the intersection boundary between the roof and the main model. Multiple feature nodes are selected at the intersection boundary, and the coordinate deviation of the multiple feature nodes on the roof and the main model is calculated. By adjusting the position of the boundary vertices of the roof model, the coordinates of the boundary nodes of the roof and the main model are made consistent.

[0024] It should be further explained that step S104 specifically includes:

[0025] The point cloud RGB data deconstruction module is called to extract the R, G, and B channel values ​​of each point cloud; based on the characteristic that the building's exterior windows are darker in visible light images, a color threshold is set to filter out the point sets that meet the conditions as candidate areas for exterior windows, and point clouds in non-window areas are excluded.

[0026] Spatial distribution analysis is performed on the initially screened candidate point set for the outer window. Outliers are eliminated by calculating the dispersion of the distance between points. Based on the color consistency test, color anomalies caused by reflection or shadow are filtered out, and candidate point sets with uniform color and spatial clustering are retained.

[0027] Using the RTree spatial indexing mechanism, a proximity search is performed on the preprocessed candidate point set. A spatial distance threshold is set, and adjacent points within the same facade are grouped into independent window candidate groups. Points spanning facades are assigned based on the facade normal vector.

[0028] For each candidate window opening set, the convex hull algorithm is used to generate the minimum enclosing cube to determine the three-dimensional boundary range of the window opening. Through the Boolean operation module, the cube and the point cloud of the main building facade are subjected to the difference operation to extract the accurate geometric contour of the window opening. The contour is then smoothed to generate the three-dimensional boundary of the window opening.

[0029] It should be further explained that step S105 specifically includes:

[0030] A drone equipped with a high-precision infrared camera was used to conduct aerial photography of each facade of the building along a preset route, and the POS data of each infrared image was recorded.

[0031] An orthorectification algorithm based on UAV oblique photography is used to geometrically correct the infrared image sequence from aerial photography, generating orthorectified infrared images of each facade. The orthorectified images are then segmented to extract independent regions for the exterior walls, windows, and roof. The average grayscale value of each region is calculated and converted into a temperature value to obtain the average external surface temperature of each facade component. T so ;

[0032] Infrared measuring points were set up on the inner surface of the exterior wall, the roof ceiling, and the glass area of ​​the exterior windows, and images of the measuring points were captured using an infrared thermal imager.

[0033] Calculate the average internal surface temperature of each building envelope within the selected ROI area corresponding to each measuring point. The calculation formula is as follows:

[0034] ;

[0035] Among them, T si,i Let Ai be the inner surface temperature of the i-th measurement area, and Ai be the corresponding area.

[0036] Indoor air temperature is calculated using the same area-weighted strategy as indoor surface temperature to obtain the average indoor air temperature T. in,avg The expression is as follows:

[0037] ;

[0038] Indoor air temperature.

[0039] It should be further explained that step S106 specifically includes:

[0040] The average outer surface temperature obtained in step S105 T so Average temperature of inner surface and indoor air average temperature T in,avg Data filtering was performed, and the stability of the data was verified by comparing the temperature change trends at the same measuring point at different times, thus forming a temperature dataset.

[0041] From the 3D geometric model completed in steps S102 and S103, extract the actual physical area of ​​each component, including the exterior wall, roof, and exterior windows; combine this with on-site measurements or design drawings to obtain the key structural parameters of each component, such as the thickness of the exterior wall insulation layer. Thickness of wall and roof insulation layers _Roof and exterior window glass thickness d_window;

[0042] The K value is calculated using the following formula:

[0043] ;

[0044] in, Indicates the inner surface temperature. This represents the outer surface temperature, where K is the heat transfer coefficient. T in Indicates indoor air temperature. This indicates the measured temperature of the inner surface of the outer envelope structure. h c This represents the convective heat transfer coefficient of the building's outer surface.

[0045] It should be further explained that in step S106, after extracting the internal and external surface temperatures of different components of each facade and the indoor temperature, the K-values ​​of the four facades (east, south, west, and north) are independently inverted, and an area-weighted integration method for the overall K-value of the facade is constructed based on the component area data obtained from the point cloud modeling results.

[0046] The heat transfer coefficients of the exterior walls in four directions are calculated using the K-value calculation formula. Combined with the geometric area data of each facade's exterior walls extracted from the visible light point cloud model, and assuming the areas of the east, south, west, and north facades are AE, AS, AW, and AN respectively, the average heat transfer coefficient of the building's overall exterior walls is calculated using the following weighted formula:

[0047] ;

[0048] By weighting the K values ​​obtained from the inversion of each facade with the actual area of ​​their corresponding components, the degree of heat loss contribution in each direction is reflected, providing thermal parameter input for BEM modeling.

[0049] This application also provides a BEM modeling system for existing buildings based on UAV infrared thermal imaging, the system comprising:

[0050] The aerial photography acquisition module is used to acquire building image sequences through drone aerial photography and to preprocess them to generate point cloud data of the building itself.

[0051] The parametric construction module generates a 3D geometric model of the building body based on the parameterized contour extraction, smoothing, and solid extrusion of point cloud data.

[0052] The model fusion and calibration module is used to identify the point cloud of the roof area by combining oblique photogrammetry data, extract the roof ridge line and elevation information, reconstruct the roof geometry, and perform boundary node fusion and calibration with the main model to form a complete 3D model.

[0053] The region identification and geometric contour generation module is used to filter out candidate regions for the outer window by analyzing the RGB value features of the visible light point cloud, and to group the window opening point set based on the spatial proximity clustering algorithm to generate the three-dimensional boundary of the window opening and extract the geometric contour of the window opening.

[0054] The temperature acquisition and processing module is used to collect building exterior surface temperature data and calculate the average exterior surface temperature of the building envelope by combining it with orthoradiated infrared images; the interior surface temperature and indoor air temperature are determined by integrating interior surface infrared temperature measurement and area weighting.

[0055] The building energy consumption model construction module, based on temperature data and building envelope construction parameters, inverts the heat transfer coefficients of exterior walls, roofs and windows, and calculates the weighted average K value of the overall building envelope by combining geometric model area data, and finally constructs the building energy consumption model.

[0056] According to another embodiment of this application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the existing building BEM modeling method based on UAV infrared thermal imaging.

[0057] According to another embodiment of this application, a storage medium is also provided, on which a computer program is stored, which, when executed by a processor, implements the steps of the existing building BEM modeling method based on UAV infrared thermal imaging.

[0058] As can be seen from the above technical solutions, the present invention has the following advantages:

[0059] The BEM modeling method for existing buildings based on UAV infrared thermal imaging provided by this invention utilizes UAV aerial photography, which allows for free flight to various locations within the building, rapidly acquiring omnidirectional image sequences and effectively solving the data acquisition challenge. By combining oblique photogrammetry data to identify point clouds in the roof area, the roof ridge line and elevation information are accurately extracted to reconstruct the roof geometry. Furthermore, the geometric contours of exterior windows are precisely extracted using point cloud RGB value features and spatial proximity clustering algorithms, improving the accuracy of the building model.

[0060] During the model construction process, unified coordinate correction was performed on the data of each part, and boundary node fusion calibration and other operations were used to ensure the overall closure and accuracy of the model. An infrared camera mounted on a drone was used to collect building exterior surface temperature data, and the interior surface temperature and indoor air temperature were obtained through a scientific method of interior surface temperature measurement and area-weighted integration. Based on this accurate temperature data and the structural parameters of the building envelope, the heat transfer coefficients of the exterior walls, roof, and windows obtained through inversion are more consistent with the actual situation. Consequently, the calculated weighted average K value of the overall building envelope is also more accurate, providing precise thermal parameters for the building energy consumption model. This not only improves the efficiency and accuracy of BEM modeling of existing buildings but also provides a scientific basis for building energy-saving renovation and energy consumption assessment. From the above technical solution, it can be seen that this invention has the following advantages:

[0061] The BEM modeling method for existing buildings based on UAV infrared thermal imaging provided by this invention utilizes UAV aerial photography, which allows for free flight to various locations within the building, rapidly acquiring omnidirectional image sequences and effectively solving the data acquisition challenge. By combining oblique photogrammetry data to identify point clouds in the roof area, the roof ridge line and elevation information are accurately extracted to reconstruct the roof geometry. Furthermore, the geometric contours of exterior windows are precisely extracted using point cloud RGB value features and spatial proximity clustering algorithms, improving the accuracy of the building model.

[0062] During model construction, unified coordinate correction was performed on all data components, and boundary node fusion calibration and other operations ensured the overall closedness and accuracy of the model. An infrared camera mounted on a drone was used to collect building exterior surface temperature data. Internal surface temperature and indoor air temperature were obtained through scientific methods of internal surface temperature measurement and area-weighted integration. Based on this precise temperature data and building envelope structural parameters, the heat transfer coefficients of the exterior walls, roof, and windows obtained through inversion are more consistent with actual conditions. Consequently, the calculated weighted average K-value of the overall building envelope is more accurate, providing precise thermal parameters for the building energy consumption model. This not only improves the efficiency and accuracy of existing building BEM modeling but also provides a scientific basis for building energy-saving renovation and energy consumption assessment. Attached Figure Description

[0063] To more clearly illustrate the technical solution of the present invention, the accompanying drawings used in the description will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0064] Figure 1 A flowchart of a BEM modeling method for existing buildings based on UAV infrared thermal imaging;

[0065] Figure 2 To create a model of a real-world aerial view of the building;

[0066] Figure 3 A schematic diagram of the point cloud creation and processing process;

[0067] Figure 4 To extract and flatten the outer envelope structure segments, a schematic diagram of the building's plan outline is fitted based on the picked point cloud.

[0068] Figure 5 This is a schematic diagram of the modeling process;

[0069] Figure 6 To create a 3D view of the battery pack and the model for a building energy consumption model driven by real data;

[0070] Figure 7This is a schematic diagram of an existing building BEM modeling system based on UAV infrared thermal imaging. Detailed Implementation

[0071] The BEM modeling method for existing buildings based on UAV infrared thermal imaging provided in this application constructs a parametric 3D geometric model using visible light point clouds from UAVs, inverts the heat transfer mode of the building envelope based on infrared images, and achieves rapid generation of building energy consumption models by combining the Grasshopper platform. The method has small geometric modeling errors, and the inverted values ​​of heat transfer coefficients for exterior walls, roofs, and windows are 2.20, 2.29, and 3.95 W / (m²·K), respectively, improving modeling efficiency and accuracy.

[0072] The following describes in detail the BEM modeling method for existing buildings based on UAV infrared thermal imaging, as described in this application. Specific details, such as particular system structures and technologies, are presented for illustrative purposes and not for limitation, to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application can also be implemented in other embodiments without these specific details.

[0073] It should be understood that, when used in this specification, the term "comprising" indicates the presence of the described feature, integral, step, operation, element, and / or component, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or collections thereof. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0074] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0075] Please see Figure 1 The diagram shows a flowchart of a BEM modeling method for existing buildings based on UAV infrared thermal imaging in a specific embodiment. The method includes:

[0076] Step S101: Obtain building image sequences through drone aerial photography and preprocess them to generate point cloud data of the building body.

[0077] In one exemplary embodiment, the flight path of the drone is planned according to the scale, shape, and structural characteristics of the building. Appropriate flight altitude, speed, shooting angle, and shooting interval are set to ensure that a sequence of images of the building can be acquired from all directions and multiple angles, with a certain degree of overlap between the images. The acquired image sequence is imported into DJI Terra, a professional camera, and its 3D modeling function based on the structure-of-motion (SOMO) algorithm is used to perform multi-view photogrammetric reconstruction, generating an original point cloud model.

[0078] The original point cloud model contains a large number of neighboring buildings, trees, roads, etc. around the building. Using LeicaCyclone3DR point cloud processing software, environmental point clouds are removed through manual selection and automatic classification, while the point cloud data of the building itself is retained. The point cloud data is then filtered and denoised to improve the purity and spatial accuracy of the point cloud.

[0079] In this way, drone aerial photography can efficiently and conveniently acquire all-around image data of buildings. Compared with traditional measurement methods, it is not limited by terrain and building height, and can quickly cover large areas. Through 3D reconstruction and point cloud preprocessing, accurate point cloud data of the building itself can be generated.

[0080] Step S102: Generate a three-dimensional geometric model of the building body by parametrically extracting the contour, smoothing the surface, and extruding the solid based on the point cloud data of the building body.

[0081] In this embodiment, the preprocessed building point cloud data is imported into the Rhino or Grasshopper 3D modeling platform. In Rhino, the coordinate system is rotated and corrected so that the building outline is roughly aligned with the xyz axis of the world coordinate system.

[0082] As one embodiment of this application, step S102 specifically includes:

[0083] Step S1021: By comparing the spatial orientation deviation between the main building axis and the xyz axis of the world coordinate system, adjust the overall rotation angle and translation parameters of the point cloud to align the main building facade with the world coordinate axis.

[0084] Step S1022: Select the point cloud of the middle floor and project it onto the xy plane using the Grasshopper platform to generate a two-dimensional discrete point set; remove outliers from the projected point set and rearrange the points in clockwise or counterclockwise order to generate an initial closed polyline profile.

[0085] Step S1023: Perform a preset number of iterations to smooth the initial closed polyline; by comparing the fit between the original point cloud and the smoothed curve, adjust the smoothing parameters to eliminate local jitter and jagged effects at the boundary, and generate a closed boundary curve.

[0086] Step S1024: Based on the measured total height of the building, the smoothed closed boundary curve is extruded vertically along the positive z-axis to form a preliminary three-dimensional closed solid; check whether there are holes or overlaps inside the solid, repair local defects, and generate a three-dimensional geometric model of the building body.

[0087] As can be seen, the point cloud of the middle floors, which are less affected by the environment and have a relatively regular structure, is selected and flattened onto the xy plane using Grasshopper, forming a series of two-dimensional discrete point sets. These two-dimensional point sets are initially screened to remove outliers or distorted points, and then the points are rearranged in a clockwise or counterclockwise order to generate closed polylines. Multiple segment smoothing processes are then performed on the polylines, typically 50-200 times, to eliminate local jitter and jagged effects on the boundary lines, resulting in accurate and smooth closed boundary curves. Finally, combined with the total building height determined by UAV oblique photography, the closed boundary curves are extruded along the z-axis to construct a three-dimensional closed entity that realistically reflects the shape of the main building. Through the screening and smoothing of the contour lines, the errors and noise in the point cloud data can be effectively eliminated, making the generated model more closely match the actual shape of the building, ensuring the geometric accuracy of the model, and providing an accurate geometric basis for subsequent energy consumption simulation and analysis.

[0088] Step S103: Combine oblique photogrammetry data to identify the point cloud of the roof area, extract the roof ridge line and elevation information, reconstruct the roof geometry, and perform boundary node fusion calibration with the main model to form a complete three-dimensional model.

[0089] In step S103 of this embodiment, before identifying the roof area point cloud by combining oblique photogrammetry data, the elevation distribution characteristics of the overall building point cloud are analyzed to define the range of the roof point cloud. Then, the slope threshold filtering algorithm is used to separate the roof tilt surface point cloud from the main building point cloud.

[0090] When extracting the roof ridge line, the roof point cloud is first meshed to generate a triangular mesh model. By identifying the abrupt change in the normal vector of adjacent triangular faces in the mesh model, the initial position of the roof ridge line is detected. Then, the ridge line breakpoints are connected and corrected by human interaction.

[0091] When acquiring roof elevation information, the three-dimensional coordinate data of the roof point cloud are used to generate an elevation contour map of the roof surface. By analyzing the distribution and direction of the contour lines, characteristic parameters such as the roof slope and drainage direction are determined.

[0092] When performing boundary node fusion calibration of the reconstructed roof geometry and the main model, Boolean operations are used to find the intersection boundary between the roof and the main model. Multiple feature nodes are selected at the intersection boundary, and the coordinate deviation of the multiple feature nodes on the roof and the main model is calculated. By adjusting the position of the boundary vertices of the roof model, the coordinates of the boundary nodes of the roof and the main model are made consistent.

[0093] In this way, by combining oblique photogrammetry data to accurately identify the point cloud of the roof area and reconstruct the roof geometry, complex roof structures can be realistically reproduced, improving the integrity and realism of the building model.

[0094] Step S104: By analyzing the RGB value characteristics of the visible light point cloud, candidate regions for the outer window are selected, and the window opening point set is grouped based on the spatial proximity clustering algorithm to generate the three-dimensional boundary of the window opening and extract the geometric contour of the window opening.

[0095] In step S104 of this embodiment, the point cloud RGB data deconstruction module is called to extract the R, G, and B channel values ​​of each point cloud; based on the characteristic that the building's exterior windows are darker in visible light images, a color threshold is set to filter out the point sets that meet the conditions as candidate areas for exterior windows, and point clouds in non-window areas are excluded.

[0096] Spatial distribution analysis was performed on the initially screened candidate point set for exterior windows. Outliers were eliminated by calculating the dispersion of distances between points. Based on color consistency testing, color anomalies caused by reflections or shadows were filtered out, retaining candidate point sets with uniform color and spatial clustering. The RTree spatial indexing mechanism was used to perform proximity search on the preprocessed candidate point set. A spatial distance threshold was set to group adjacent points within the same facade into independent candidate window opening groups. Points spanning facades were assigned based on facade normal vectors to ensure that points for the same window opening were concentrated within the same facade.

[0097] For each candidate window opening set, the convex hull algorithm is used to generate the minimum enclosing cube to determine the three-dimensional boundary range of the window opening. Through the Boolean operation module, the cube and the point cloud of the main building facade are subjected to the difference operation to extract the accurate geometric contour of the window opening. The contour is then smoothed to generate the three-dimensional boundary of the window opening.

[0098] As can be seen, the exterior window areas in the building facade of this embodiment typically appear as dark areas in visible light images, meaning they have low RGB values. Utilizing this visual characteristic of exterior windows in the point cloud data, the RGB channel values ​​of each point cloud data point are deconstructed using the Grasshopper 3D modeling platform, through the GhPython programming module and Volvox plugin. Color thresholds are set, such as R, G, and B all being less than 90, to initially filter out point sets that may be window openings, obtaining preliminary color recognition results. Since the initially filtered window point clouds may contain misidentified points in non-window areas, such as facade shadows, surface contamination, and abnormal points caused by glass reflections, the RTree spatial indexing mechanism built into the 3D modeling platform, combined with breadth-first search logic, automatically clusters dense point sets within a certain spatial range to form independent point cloud groups. By setting a minimum point count threshold, isolated or excessively small stray point clouds are automatically filtered out, retaining only window opening point sets of reasonable size and high concentration. By utilizing the Boolean operation function of the 3D modeling platform, a 2D window opening boundary plane is automatically generated where the cube intersects with the main building facade, resulting in a precise window opening geometric outline. This effectively eliminates interference points, accurately extracts the window opening geometric outline, improves the integrity and accuracy of the exterior window details in the building model, and thus enhances the accuracy of building energy consumption simulation.

[0099] Step S105: Collect building exterior surface temperature data and calculate the average exterior surface temperature of the building envelope by combining it with orthophoto infrared images; determine the interior surface temperature and indoor air temperature by integrating interior surface infrared temperature measurement and area weighting.

[0100] This embodiment uses a drone equipped with an infrared camera to collect a sequence of high-precision infrared thermal images of various facades of a building with a certain degree of overlap, following a planned flight path. Based on the orthorectified infrared images, orthorectified infrared images of exterior windows and exterior walls are obtained through image segmentation. The relationship between grayscale values ​​and temperature values ​​is fitted, outliers are removed by image masking, and the average infrared temperature of each facade is calculated to obtain the exterior surface temperature.

[0101] For internal surface temperature measurement, typical functional spaces such as living rooms and bedrooms were selected. Infrared thermography was conducted on the top, middle, and bottom floors of the exterior wall interior surfaces, as well as inside the corner and middle units. Roof measurement points were concentrated on the ceilings of the bedrooms, kitchens, and bathrooms of the top-floor units. For exterior windows, the main window areas of the aforementioned rooms were selected. Using a FLUKET i32 handheld infrared thermal imager, images were taken from a vertical perspective at a distance of 1.5-2.0 meters from the measurement surface. A high emissivity calibration point was attached to the center of the window, and the camera emissivity was uniformly set to 0.95. The image temperature scale was calibrated using the calibration point. Multiple thermal images were taken of the exterior walls, roof, and windows for each room. Using FLUKE's official SmartView software, ROI region extraction rules were set for different component types to extract the average temperature value. Image stability and thermal gradient uniformity were assessed through image grayscale distribution, and abnormal image samples were removed.

[0102] Based on the actual physical area of ​​each room at each measuring point (extracted from the established building point cloud geometric model), temperature data were integrated using area as the weight to obtain the average temperature of the inner surface. Indoor air temperature was measured using a HOBOU23-001 temperature and humidity recorder, installed in the center of each room at a height of approximately 1.2-1.5 meters, away from obvious heat sources. For areas with external windows, measuring points were set on the indoor side 10-20 cm from the glass and 0.3-0.5 m above the windowsill. The same area-weighted strategy was used to integrate and obtain the average indoor air temperature.

[0103] In this way, the internal surface temperature and indoor air temperature are integrated using a scientific method of measuring point arrangement and area weighting, which fully considers the actual heat conduction path and area ratio of the building space, effectively improving the accuracy and representativeness of the temperature data.

[0104] Step S106: Based on temperature data and building envelope construction parameters, invert the heat transfer coefficients of exterior walls, roof and windows, and calculate the weighted average K value of the overall building envelope by combining geometric model area data, and finally construct the building energy consumption model.

[0105] This embodiment is based on the steady-state heat conduction theory and uses measured data of external surface temperature, internal surface temperature and indoor air temperature to calculate the heat transfer coefficient K values ​​of external walls, roof and external windows respectively.

[0106] For the exterior walls, the K values ​​for the east, south, west, and north facades are calculated separately. Then, combined with the geometric area data of each facade's exterior walls extracted from the visible light point cloud model, the average heat transfer coefficient of the building's overall exterior walls is calculated using a weighted formula. The overall K value for the roof and exterior windows is calculated using the same method.

[0107] Based on the measured K-value, the insulation layer thickness was inferred using typical construction materials from different eras to determine the thermal performance parameters of the building envelope. The window construction type was identified, and the solar heat gain coefficient and visible light transmittance were referenced. Typical climate year data for Jinan from the EnergyPlus standard database were directly used as meteorological data, and indoor occupancy, lighting, and electrical load settings were based on typical recommended values ​​in the standards. The geometric model was imported into the energy consumption simulation platform, and modeling was performed by dividing the building into unified thermal zones according to the overall floor plan, thus completing the construction of the building energy consumption model.

[0108] In this way, the heat transfer coefficient, derived from measured temperature data, can accurately reflect the actual thermal performance of existing building envelopes, and is more consistent with reality than theoretical calculations or empirical values. The overall weighted average K value is calculated by combining geometric model area data, taking into account the contribution of different parts of the envelope to the overall building heat transfer. The constructed building energy consumption model comprehensively considers multiple factors, accurately simulating building energy consumption and reducing building operating energy consumption.

[0109] Based on the above embodiments, in order to further improve the reliability of the existing building BEM modeling method based on UAV infrared thermal imaging provided in the above embodiments, and to accurately simulate building energy consumption and reduce building operating energy consumption, the following is a more specific implementation method. This embodiment will take a residential building in a certain city as an example to explain in detail the BEM establishment method based on UAV infrared thermal imaging technology. The method includes data acquisition, three-dimensional geometric model establishment, thermal parameter inversion of the building envelope, and extraction and setting of model input parameters.

[0110] The selected building for modeling is an old residential building constructed in 2000. It has seven floors, with an attic on the top floor. The exterior walls and roof lack insulation, and the heating system uses convection radiators. This building is representative of the current state of old residential buildings in heating areas. The building envelope's insulation performance was inadequate from the initial design and construction, and its performance has further deteriorated over time. Data was collected on February 1, 2024, using a DJIM 300RTK drone equipped with a compatible dual-vision camera system platform, the ZENMUSE H20T, capable of simultaneously acquiring visible light and infrared images.

[0111] To accurately construct the thermal boundary of the building envelope (BEM), aerial visible light images were first acquired using drones. The acquisition area should cover the four facades and roof of the main building. Its geometric accuracy directly determines the accuracy of the simulation. Aerial photography paths with sufficient coverage and a certain degree of redundancy should be designed, taking into account the building's shape characteristics and on-site occlusion conditions.

[0112] The image sequence obtained after aerial photography is imported into 3D reconstruction software. In this embodiment, DJI Terra is used, based on the SFM algorithm, to perform multi-view photogrammetric reconstruction through its "3D modeling" function, generating a high-precision point cloud model. The original point cloud obtained by aerial surveying contains environmental information around the building, such as neighboring buildings, trees, and roads. The point cloud is preprocessed using LeicaCyclone3DR software.

[0113] After preprocessing, the purity and spatial accuracy of the point cloud are significantly improved, providing a reliable data foundation for geometric modeling. The processing flow is as follows: Figure 3 As shown.

[0114] To achieve accurate geometric modeling of thermal zone boundaries in building energy consumption simulation, it is necessary to transform the processed unstructured point cloud data into a parametric 3D geometric model. The transformation process is based on the Rhino and Grasshopper platforms, and completes the entire process from point cloud discretization to geometry generation through parametric operations.

[0115] This embodiment imports the preprocessed point cloud into Rhino and rotates and corrects the coordinate system so that the building outline is roughly aligned with the xyz axes of the world coordinate system. Then, a point cloud from a layer less affected by the environment (such as an intermediate layer) is selected and flattened onto the xy plane using Grasshopper to form a series of two-dimensional discrete point sets. These two-dimensional point sets reflect the projection of the building facade onto the horizontal plane, serving as the basis for outline extraction.

[0116] In the Grasshopper platform, the flattened point set undergoes initial screening to remove outliers or distorted points. The points are then rearranged in a clockwise or counter-clockwise order to generate closed polylines. To eliminate local jitter and jagged edges in the boundary lines, approximately 50-200 smoothing operations are required. Experiments show that about 100 smoothing operations achieve optimal results, ultimately generating accurate and smooth closed boundary curves that precisely approximate the actual building plan shape. The specific process is as follows... Figure 4 As shown. This fitted boundary curve serves as an important input parameter for generating the basic building mass.

[0117] Based on the fitted closed boundary curve and the total building height determined by UAV oblique photography, the planar curve is extruded along the z-axis to construct a three-dimensional closed entity that truly reflects the shape of the main building.

[0118] To accurately reconstruct the roof structure, the point cloud of the roof area needs to be identified separately. By extracting the roof ridge line and elevation information, Grasshopper is used to call the automatic fitting function of the Ironbug plugin to accurately identify the roof feature points and reconstruct the roof geometry. To ensure a tight connection between the roof and the main model, fusion calibration is performed at the boundary nodes of the roof and the main model to ensure the closure and integrity of the overall model.

[0119] After the geometric model of the main building is constructed, in order to further improve the integrity of the model details and the accuracy of energy consumption simulation, the external window components are accurately identified and embedded. This embodiment utilizes the obvious visual characteristics of the external windows in the visible light image from the point cloud data, such as darker colors and lower RGB values, to design an automated identification and parametric modeling process based on the Grasshopper platform, GhPython module, and Volvox plugin, realizing rapid geometric reconstruction at the window component level.

[0120] Exterior window areas on building facades typically appear as dark areas in images, indicating low RGB values. Based on this characteristic, the RGB channel values ​​of each point cloud data are deconstructed using the GhPython module and Volvox plugin on the Grasshopper platform. Color thresholds (such as R, G, and B all being less than 90) are set to quickly and initially filter out point sets that may be window openings, generating preliminary color recognition results.

[0121] The initially selected window point cloud may contain misidentified points in non-window areas, such as facade shadows, surface contamination, and abnormal points caused by glass reflections, which seriously interfere with the accurate extraction of window opening areas. Therefore, this embodiment proposes a scattered point grouping algorithm based on spatial proximity clustering to further improve the accuracy of window opening area identification. The algorithm is based on Grasshopper's built-in RTree spatial indexing mechanism and combines breadth-first search (BFS) logic to automatically cluster dense point sets within a certain spatial range, forming independent point cloud groups. By setting a minimum point count threshold, isolated or excessively small scattered point clouds can be automatically filtered out, thus retaining only window opening point sets of reasonable size and high concentration. The window opening point cloud after this step has good spatial aggregation and clear contour boundaries. Using a standard floor with minimal impact from structural changes and occlusion as a benchmark, the accurate three-dimensional boundaries of each window opening are extracted from the spatially clustered window opening point sets, and the minimum bounding cube algorithm is used to construct the window opening blocks. Subsequently, using the Boolean operation function of the Grasshopper platform, the two-dimensional window opening boundary plane formed by the intersection of the cube and the main facade of the building is automatically generated, forming a precise window opening geometric outline.

[0122] Considering the repetitive and consistent characteristics among floors in multi-story residential buildings, the window opening boundaries generated by fitting standard floors can be quickly and batch-copied to corresponding positions on other floors. This enables rapid parametric modeling of the overall exterior window components, significantly improving the efficiency and consistency of overall geometric modeling. Automatic identification is challenging for complex components with significant 3D structural variations, such as dormer windows and bay windows in the roof area. Therefore, manual modeling is necessary to supplement complex window components and ensure the integrity of the overall model. After the exterior windows are fitted and embedded into the main building model, a detailed check is performed on the entire geometric model to ensure the validity and geometric closure of the energy consumption model input data. The optimized parametric geometric model possesses precise window component-level details and high structural integrity, effectively meeting the requirements for building energy consumption models. It provides a reliable data foundation for accurate setting of building envelope material parameters, thermal zone division, and energy consumption simulation analysis. The modeling results are as follows: Figure 5 As shown.

[0123] To accurately assess the thermal performance of existing building envelopes, this embodiment establishes a method for inverting the heat transfer coefficient (K-value) based on measured temperature data from infrared images. This method, based on steady-state heat conduction theory, achieves precise inversion of the K-values ​​of exterior walls, roofs, and windows through component thermal balance equations, and is particularly suitable for existing residential buildings with significant thermal performance degradation.

[0124] This embodiment is based on the steady-state K-value inversion formula of infrared thermometry:

[0125] ;

[0126] in, This indicates the temperature of the inner surface, in Kelvin (K). The external surface temperature is represented by K; K is the heat transfer coefficient, with units of W / (m²·K); ε represents the surface emissivity, which is typically between 0.95 and 1.00 for exterior wall tiles or paint surfaces, and is set to 1 here; σ represents the Stefan-Boltzmann constant.

[0127] It is 5.67 × 10 -8 W·m -2 ·K -4 ; T refl This indicates the reflection temperature, expressed in Kelvin (K). T in Indicates indoor air temperature, in Kelvin (K). This indicates the measured temperature of the inner surface of the external envelope structure, in Kelvin (K). h c The convective heat transfer coefficient of the building's exterior surface is 8.7 W / (m²). 2 ·k).

[0128] Under conditions of high emissivity structures such as plastering and brickwork, when the environment is stable and there is no direct sunlight, the reflection temperature is approximately equal to the inner surface temperature T. refl ≈T si The assumption can effectively simplify the calculation of the radiation term. At this time, the error of the K value is generally controlled within ±10%, which is suitable for rapid on-site assessment. The K value calculation formula can be further simplified to:

[0129] ;

[0130] The simplified model is uniformly applicable to the inversion of heat transfer performance of exterior walls, roofs and most uncoated exterior windows, avoiding the introduction of uncertainty in radiation terms. While ensuring the reliability of temperature measurement, it improves the stability of calculation and operational efficiency. In this embodiment, the calculation of K value will adopt Equation (2).

[0131] external surface temperature of building envelope T so This is a crucial boundary condition for accurately retrieving the K-value. In this embodiment, a UAV equipped with an infrared camera acquires a sequence of high-precision infrared thermal images of each facade with a certain degree of overlap. Then, using SFM technology, orthorectified infrared images of each facade are generated. Finally, based on the orthorectified infrared images, the average infrared temperature of each facade is calculated as the outer surface temperature. T so .

[0132] Determining the outer surface temperature based on orthorectified infrared images T so The process mainly includes image segmentation to obtain orthoradio images of the outer window and outer wall respectively, fitting the relationship between gray values ​​and temperature values, and image masking to remove outliers.

[0133] Finally, the average external surface temperature values ​​of the walls, windows, and roof components on the four facades (east, south, west, and north) were extracted and calculated. This provided boundary conditions for subsequent internal and external temperature difference analysis and K-value calculation. The relevant numerical results are detailed in Table 1.

[0134] ;

[0135] To address the practical difficulties in obtaining internal surface temperature (such as entry restrictions, furniture obstruction, and localized thermal disturbances), this embodiment designs a measurement scheme centered on the layout of measurement points in typical functional spaces and area-weighted integration. Infrared thermography of the internal surface primarily selects the living room and bedroom as typical functional spaces, taking into account their high area proportion, strong heat load sensitivity, and wide exposure area, thus providing a good representation of the overall thermal performance of the residential space. For the layout of measurement points on the internal surface of the exterior walls, infrared thermography is conducted inside the corner and middle units on the top, middle, and bottom floors. Roof measurement points are concentrated in the top-floor units, including the ceilings of bedrooms, kitchens, and bathrooms. For exterior windows, the main window areas in the aforementioned rooms are selected, emphasizing the high proportion of heat flux through windows and the significant thermal disturbance at the glass-wall interface to ensure the most representative heat transfer path is reflected. The measurement point layout excludes enclosed balconies, bay window sills, recesses, and irregular corners to avoid systematic errors caused by spatial irregularities and localized heat source interference.

[0136] The internal surface temperature was measured using a FLUKETi32 handheld infrared thermal imager, with the instrument positioned 1.5-2.0 meters from the measurement surface and captured from a vertical perspective. To improve the accuracy of glass surface measurements, a high emissivity calibration point was affixed to the center of the window, and the camera emissivity was uniformly set to 0.95. The image temperature scale was calibrated using the calibration point. Multiple thermal images were taken of the exterior walls, roof, and windows of each room to avoid potential environmental interference or data anomalies from single measurements. To ensure the accurate correspondence between internal and external surface measurement data, the measurements strictly adhered to the principle of "consistent spatial location and synchronized time." The temperature measurement locations on the internal and external surfaces should correspond spatially, and the acquisition time interval should be controlled within 15 minutes to ensure minimal changes in the building's thermal inertia.

[0137] The internal surface temperature data processing employed FLUKE's official SmartView software. Region of Interest (ROI) rules were set for different component types. For walls and roofs, large, unobstructed areas with uniform texture were prioritized, avoiding interference from doorways, corners, beam-column junctions, and furniture. For exterior windows, the ROI was limited to the center of the glass's field of view, excluding frames, window sills, and curtain areas. High-reflectivity hotspot pixels were thresholded and filtered out, retaining only the main glass area with stable emissivity response for temperature enhancement. At least two ROIs were set for each room, and their average temperature values ​​were extracted for each. Image stability and thermal gradient uniformity were assessed using image grayscale distribution. Image samples with overexposed edges, strong interference sources, or abrupt temperature changes were removed, retaining only reliable thermal image data for subsequent analysis.

[0138] After the above processing, the average inner surface temperature of each enclosure structure within the selected ROI area can be obtained at each measuring point. Simultaneously, the actual physical area Ai of the corresponding region is recorded, which can be extracted from the established building point cloud geometric model. Finally, the temperature data of each room at each measuring point are integrated with area as the weight to obtain the weighted average internal surface temperature of the exterior walls, roof, and windows in each direction. The calculation formula is as follows:

[0139] ;

[0140] Where Tsi,i represents the inner surface temperature of the i-th measurement area, and Ai represents the corresponding area. This method takes into account both the room's spatial distribution and the actual area ratio of heat conduction paths, avoiding structural biases caused by single-point sampling, and effectively improving... Physical representativeness.

[0141] Indoor air temperature (Tin) was measured using a HOBOU23-001 temperature and humidity recorder, installed in the center of each measuring point in the room at a height of approximately 1.2-1.5 meters, away from obvious heat sources such as air conditioners, radiators, and doors and windows, to accurately reflect the overall thermal environment of the room. The room temperature in the area with external windows was measured separately, with measuring points set up on the indoor side 10-20 cm from the glass and 0.3-0.5 m above the windowsill to reduce interference from the air temperature gradient near the window. Considering the differences in thermal environment between rooms, the indoor air temperature was calculated using the same area-weighted strategy as the inner surface temperature, resulting in the average indoor air temperature (Tin,avg), expressed as follows:

[0142] ;

[0143] This homogenization process, under the premise of relatively mild thermal field fluctuations, can effectively reduce the interference of local deviations on the stability of K-value inversion. Finally, the average inner surface temperatures of the exterior walls, windows, and roof components in each direction were obtained. With indoor air temperature T in,avg .

[0144] Based on the extraction of the internal and external surface temperatures of different components on each facade and the indoor temperature, the K-values ​​of the four facades (east, south, west, and north) need to be independently inverted. An area-weighted integration method for the overall K-value of the facades is then constructed based on the component area data obtained from the point cloud modeling results. Formula 2 is used to calculate the heat transfer coefficient of the facades in each of the four directions. Based on this, and combining the geometric area data of the facades extracted from the visible light point cloud model, assuming the areas of the east, south, west, and north facades are AE, AS, AW, and AN respectively, the average heat transfer coefficient of the overall building facades can be calculated using the following weighted formula:

[0145] ;

[0146] This weighted integration method fully considers the area proportion of different facades in the overall heat transfer of the building envelope, effectively avoiding the risk of overall K-value deviation due to insufficient representativeness of local measurement areas. By weighting the K-values ​​obtained from the inversion of each facade with the actual area of ​​their corresponding components, the contribution of heat loss in each direction can be systematically reflected, providing more representative and physically meaningful thermal parameter inputs for BEM. By dividing the weighted sum by the total area, the overall K-value of the building exterior walls is found to be 2.20 W / m²·K, the K-value of the roof is 2.29 W / (m²·K), and the average K-value of the exterior windows is 3.95 W / (m²·K).

[0147] After completing the building geometry modeling and K-value inversion of the building envelope, other key input parameters need to be determined to ensure the accuracy and real-world fit of the BEM. This embodiment, based on Grasshopper and Ladybug Tools, uniformly sets boundary conditions such as building envelope construction, window performance, indoor load, and meteorological data to ensure the integrity and adaptability of the model.

[0148] The thermal performance parameters of the building envelope were converted into structural information that could be directly used in the Building Image Processing (BEM). Based on the measured K-value, the insulation layer thickness was inferred using typical structural materials from different eras. The exterior walls used EPS foam board insulation, with an inferred insulation layer thickness of approximately 12.5 mm. The roof used XPS extruded polystyrene board, with a thickness of approximately 8.6 mm. This parameterized structural thickness input not only reflects the current insulation status of the building but also provides a quantitative basis for subsequent renovation strategy analysis (such as assessing the energy-saving potential of insulation layer thickening schemes). Regarding window performance, in addition to the heat transfer coefficient, the solar heat gain coefficient (SHGC) and visible light transmittance (VT) also significantly affect energy consumption results. For existing buildings, considering window aging, pollution, and shading conditions, structural type identification was performed based on on-site photographs, and parameters were estimated using appropriate correction coefficients, referencing existing research literature.

[0149] Regarding other environmental and equipment operating parameters, meteorological data were directly adopted from the EnergyPlus standard database for typical climatic years in Jinan. Indoor occupancy, lighting, and electrical load settings were based on typical recommended values ​​in the specifications. Specifically, the settings were: lighting power density 5.0 W / m², electrical equipment power density 3.8 W / m², per capita building area 25 m², and fresh air exchange rate 0.5 times / h. For air conditioning system parameters, based on the actual heating and cooling conditions in Jinan, the winter heating temperature was set at 18℃, and the summer cooling temperature was set at 26℃. The operating periods for the heating and cooling seasons were set according to local usage conditions, and the system energy efficiency parameters were also set according to the recommended values ​​in GB55015-2021.

[0150] After the geometric model is imported into the energy consumption simulation platform, it is modeled according to the overall division of hot zones across the floors to reduce model complexity and improve computational efficiency, making it suitable for overall energy-saving potential analysis. Through the complete parameter settings and model building process described above, an accurate, stable, and realistic building energy consumption model is established, such as... Figure 6 As shown, this provides a reliable data foundation and methodological support for the subsequent evaluation and comparative analysis of energy-saving renovation schemes.

[0151] The UAV point cloud-driven BEM reverse modeling framework constructed in this embodiment effectively solves the problems of missing geometric information and inaccurate thermal parameters in existing buildings through multi-source data fusion. The constructed model can accurately reflect the actual energy consumption characteristics of buildings, providing a data foundation for the quantitative evaluation of building envelope renovation strategies. This method can be extended to the field of urban building stock renewal, providing technical support for shortening the energy-saving diagnosis cycle and optimizing carbon neutrality pathways.

[0152] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0153] The following are embodiments of the existing building BEM modeling system based on UAV infrared thermal imaging provided in this disclosure. This system and the existing building BEM modeling method based on UAV infrared thermal imaging in the above embodiments belong to the same inventive concept. For details not described in detail in the embodiments of the existing building BEM modeling system based on UAV infrared thermal imaging, please refer to the embodiments of the existing building BEM modeling method based on UAV infrared thermal imaging described above.

[0154] like Figure 7 As shown, the system includes:

[0155] The aerial photography acquisition module 301 is used to acquire building image sequences through drone aerial photography and to preprocess them to generate point cloud data of the building itself.

[0156] The parametric construction module 302 generates a three-dimensional geometric model of the building body based on the parameterized contour extraction, smoothing and solid extrusion of point cloud data.

[0157] The model fusion calibration module 303 is used to identify the point cloud of the roof area by combining oblique photogrammetry data, extract the roof ridge line and elevation information, reconstruct the roof geometry, and perform boundary node fusion calibration with the main model to form a complete three-dimensional model.

[0158] The region identification and geometric contour generation module 304 is used to filter out candidate regions of the outer window by analyzing the RGB value features of the visible light point cloud, and to group the window opening point set based on the spatial proximity clustering algorithm to generate the three-dimensional boundary of the window opening and extract the geometric contour of the window opening.

[0159] The temperature acquisition and processing module 305 is used to acquire building exterior surface temperature data, calculate the average exterior surface temperature of the building envelope by combining it with orthoradiated infrared images, and determine the interior surface temperature and indoor air temperature by integrating interior surface infrared temperature measurement and area weighting.

[0160] The building energy consumption model construction module 306, based on temperature data and building envelope construction parameters, inverts the heat transfer coefficients of exterior walls, roofs and windows, and calculates the weighted average K value of the overall building envelope by combining geometric model area data, and finally constructs the building energy consumption model.

[0161] Based on the above method, this application also provides an electronic device for implementing the steps of the existing building BEM modeling method based on UAV infrared thermal imaging. The electronic device can be implemented in various forms. For example, the terminal described in the embodiments of this invention may include electronic devices such as mobile phones, smartphones, laptops, digital broadcast receivers, personal digital assistants, tablet computers (PADs), portable multimedia players, navigation devices, etc., as well as fixed terminals such as digital TVs, desktop computers, etc.

[0162] Electronic devices may include wireless communication units, audio / video input units, user input units, sensing units, output units, memory, interface units, controllers, and power supply units, etc. However, it should be understood that it is not required to implement all the components shown. More or fewer components may be implemented alternatively. The elements of the electronic device will be described in detail below.

[0163] The input device can receive input numerical or character information, as well as generate key signal inputs related to user settings and function control of the electronic device used to implement the vehicle positioning method in this embodiment. Examples of input devices include touchscreens, keypads, mice, trackpads, touch panels, joysticks, one or more mouse buttons, trackballs, and joysticks. The output device may include display devices, auxiliary lighting devices, and haptic feedback devices.

[0164] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user, such as a cathode ray tube or LCD monitor; and a keyboard and pointing device, such as a mouse or trackball, through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback, such as visual feedback, auditory feedback, or tactile feedback; and input from the user can be received in any form, including sound input, voice input, or tactile input.

[0165] This application also provides a storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of a method for BEM modeling of existing buildings based on UAV infrared thermal imaging.

[0166] The storage medium may be any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example,, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of readable storage media include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0167] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined in these embodiments may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for BEM modeling of existing buildings based on UAV infrared thermal imaging, characterized in that, The methods include: S101: Obtain building image sequences through drone aerial photography and preprocess them to generate point cloud data of the building itself; S102: Generate a three-dimensional geometric model of the building body based on the parameterized contour extraction, smoothing and solid extrusion of the building body point cloud data; S103: Combine oblique photogrammetry data to identify point clouds in the roof area, extract roof ridge lines and elevation information, reconstruct the roof geometry, and perform boundary node fusion calibration with the main model to form a complete 3D model; S104: By analyzing the RGB value characteristics of visible light point clouds, candidate regions for outer windows are selected, and the window opening point sets are grouped based on the spatial proximity clustering algorithm to generate the three-dimensional boundary of the window opening and extract the geometric contour of the window opening. S105: Collect building exterior surface temperature data and calculate the average exterior surface temperature of the building envelope by combining orthorectified infrared imagery; determine the interior surface temperature and indoor air temperature by integrating interior surface infrared temperature measurement and area-weighted integration. S106: Based on temperature data and building envelope construction parameters, the heat transfer coefficients of exterior walls, roofs and windows are inverted, and the weighted average K value of the overall building envelope is calculated by combining geometric model area data, and finally a building energy consumption model is constructed. In step S106, after extracting the internal and external surface temperatures and indoor temperatures of the different components of each facade, the K-values ​​of the four facades (east, south, west, and north) are independently inverted, and an area-weighted integration method for the overall K-value of the facade is constructed based on the component area data obtained from the point cloud modeling results. The heat transfer coefficient of the exterior walls in four directions was calculated using the K-value calculation formula. Combined with the geometric area data of the exterior walls on each facade extracted from the visible light point cloud model, the areas of the east, south, west, and north facades were set as follows: , , , The average heat transfer coefficient of the building's overall exterior walls is calculated using the following weighted formula: ； in: The heat transfer coefficient of the east facade exterior wall obtained through independent inversion; The heat transfer coefficient of the south facade exterior wall obtained through independent inversion; The heat transfer coefficient of the west facade exterior wall is obtained through independent inversion. The heat transfer coefficient of the north facade exterior wall obtained through independent inversion; In each symbol, K refers to the heat transfer coefficient of the building envelope, and the subscripts "E, S, W, N" correspond to the four facades of east, south, west, and north, respectively, which is consistent with the meaning of the subscripts of the facade area symbols in the formula. By weighting the K values ​​obtained from the inversion of each facade with the actual area of ​​their corresponding components, the degree of heat loss contribution in each direction is reflected, providing thermal parameter input for BEM modeling.

2. The method for BEM modeling of existing buildings based on UAV infrared thermal imaging according to claim 1, characterized in that, Step S102 specifically includes: By comparing the spatial orientation deviation between the main building axis and the xyz axis of the world coordinate system, the overall rotation angle and translation parameters of the point cloud are adjusted to align the main building facade with the world coordinate axis. Select the point cloud of the middle floor and project it onto the xy plane using the Grasshopper platform to generate a two-dimensional discrete point set; remove outliers from the projected point set and rearrange the points in clockwise or counterclockwise order to generate an initial closed polyline profile. The initial closed polyline is smoothed by a preset number of iterations; by comparing the fit between the original point cloud and the smoothed curve, the smoothing parameters are adjusted to eliminate local jitter and jagged effects at the boundary and generate a closed boundary curve. Based on the measured total height of the building, the smoothed closed boundary curve is extruded vertically along the positive z-axis to form a preliminary three-dimensional closed solid. The solid is then checked for holes or overlaps, and local defects are repaired to generate a three-dimensional geometric model of the building.

3. The method for BEM modeling of existing buildings based on UAV infrared thermal imaging according to claim 1, characterized in that, Step S103 specifically includes: Before identifying the roof area point cloud by combining oblique photogrammetry data, the range of the roof point cloud is delineated by analyzing the elevation distribution characteristics of the overall building point cloud. Then, the slope threshold filtering algorithm is used to separate the roof tilt surface point cloud from the main building point cloud. When extracting the roof ridge line, the roof point cloud is first meshed to generate a triangular mesh model. By identifying the abrupt change in the normal vector of adjacent triangular faces in the mesh model, the initial position of the roof ridge line is detected. Then, the ridge line breakpoints are connected and corrected by human interaction. When obtaining roof elevation information, the three-dimensional coordinate data of the roof point cloud are used to generate an elevation contour map of the roof surface. By analyzing the distribution and direction of the contour lines, the characteristic parameters of the roof slope and drainage direction are determined. When performing boundary node fusion calibration of the reconstructed roof geometry and the main model, Boolean operations are used to find the intersection boundary between the roof and the main model. Multiple feature nodes are selected at the intersection boundary, and the coordinate deviation of the multiple feature nodes on the roof and the main model is calculated. By adjusting the position of the boundary vertices of the roof model, the coordinates of the boundary nodes of the roof and the main model are made consistent.

4. The method for BEM modeling of existing buildings based on UAV infrared thermal imaging according to claim 1, characterized in that, Step S104 specifically includes: The point cloud RGB data deconstruction module is called to extract the R, G, and B channel values ​​of each point cloud; based on the characteristic that the building's exterior windows are darker in visible light images, a color threshold is set to filter out the point sets that meet the conditions as candidate areas for exterior windows, and point clouds in non-window areas are excluded. Spatial distribution analysis is performed on the initially screened candidate point set for the outer window. Outliers are eliminated by calculating the dispersion of the distance between points. Based on the color consistency test, color anomalies caused by reflection or shadow are filtered out, and candidate point sets with uniform color and spatial clustering are retained. Using the RTree spatial indexing mechanism, a proximity search is performed on the preprocessed candidate point set. A spatial distance threshold is set, and adjacent points within the same facade are grouped into independent window candidate groups. Points spanning facades are assigned based on the facade normal vector. For each candidate window opening set, the convex hull algorithm is used to generate the minimum enclosing cube to determine the three-dimensional boundary range of the window opening. Through the Boolean operation module, the cube and the point cloud of the main building facade are subjected to the difference operation to extract the accurate geometric contour of the window opening. The contour is then smoothed to generate the three-dimensional boundary of the window opening.

5. The method for BEM modeling of existing buildings based on UAV infrared thermal imaging according to claim 1, characterized in that, Step S105 specifically includes: A drone equipped with a high-precision infrared camera was used to conduct aerial photography of each facade of the building along a preset route, and the POS data of each infrared image was recorded. An orthorectification algorithm based on UAV oblique photography is used to geometrically correct the infrared image sequence from aerial photography, generating orthorectified infrared images of each facade. The orthorectified images are then segmented to extract independent regions for the exterior walls, windows, and roof. The average grayscale value of each region is calculated and converted into a temperature value to obtain the average external surface temperature of each facade component. T so ; Infrared measuring points were set up on the inner surface of the exterior wall, the roof ceiling, and the glass area of ​​the exterior windows, and images of the measuring points were captured using an infrared thermal imager. Calculate the average internal surface temperature of each building envelope within the selected ROI area corresponding to each measuring point. The calculation formula is as follows: ； in, Let be the inner surface temperature of the i-th measurement area. For the corresponding area; Indoor air temperature is calculated using the same area-weighted strategy as indoor surface temperature to obtain the average indoor air temperature T. in,avg The expression is as follows: ; in Let be the indoor air temperature of the i-th measurement area.

6. The method for BEM modeling of existing buildings based on UAV infrared thermal imaging according to claim 1, characterized in that, Step S106 specifically includes: The average outer surface temperature obtained in step S105 T so Average internal surface temperature and average indoor air temperature T in,avg Data filtering was performed, and the stability of the data was verified by comparing the temperature change trends at the same measuring point at different times, thus forming a temperature dataset. From the three-dimensional geometric model completed in steps S102 and S103, extract the actual physical area of ​​each component of the exterior wall, roof, and exterior window; combine with on-site measurements or design drawings to obtain the key structural parameters of each component: exterior wall insulation layer thickness δ_wall, roof insulation layer thickness δ_roof, and exterior window glass thickness d_window; The K value is calculated based on the following formula: ； in, Indicates the inner surface temperature; This represents the outer surface temperature; K is the heat transfer coefficient. T in Indicates indoor air temperature; indicates the measured temperature of the inner surface of the external envelope structure; h c This represents the convective heat transfer coefficient of the building's outer surface.

7. A BEM modeling system for existing buildings based on UAV infrared thermal imaging, characterized in that, The system is used to implement the existing building BEM modeling method based on UAV infrared thermal imaging as described in any one of claims 1 to 6; The system includes: The aerial photography acquisition module is used to acquire building image sequences through drone aerial photography and to preprocess them to generate point cloud data of the building itself. The parametric construction module generates a 3D geometric model of the building body based on the parameterized contour extraction, smoothing, and solid extrusion of point cloud data. The model fusion and calibration module is used to identify the point cloud of the roof area by combining oblique photogrammetry data, extract the roof ridge line and elevation information, reconstruct the roof geometry, and perform boundary node fusion and calibration with the main model to form a complete 3D model. The region identification and geometric contour generation module is used to filter out candidate regions for the outer window by analyzing the RGB value features of the visible light point cloud, and to group the window opening point set based on the spatial proximity clustering algorithm to generate the three-dimensional boundary of the window opening and extract the geometric contour of the window opening. The temperature acquisition and processing module is used to collect building exterior surface temperature data and calculate the average exterior surface temperature of the building envelope by combining it with orthoradiated infrared images; the interior surface temperature and indoor air temperature are determined by integrating interior surface infrared temperature measurement and area weighting. The building energy consumption model construction module, based on temperature data and building envelope construction parameters, inverts the heat transfer coefficients of exterior walls, roofs and windows, and calculates the weighted average K value of the overall building envelope by combining geometric model area data, and finally constructs the building energy consumption model.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the existing building BEM modeling method based on UAV infrared thermal imaging as described in any one of claims 1 to 6.

9. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the existing building BEM modeling method based on UAV infrared thermal imaging as described in any one of claims 1 to 6.

Images