Building appearance defect detection method and system based on unmanned aerial vehicle

Abstract

The invention relates to the technical field of building appearance defect detection, in particular to a building appearance defect detection method and system based on an unmanned aerial vehicle, and the method comprises the steps: obtaining the building information of a target building, and generating a hierarchical scanning path and a three-dimensional obstacle avoidance flight path, a visible light image, an infrared thermodynamic diagram and laser radar point cloud information are collected for space-time alignment processing, and an attention mechanism neural network is used for extracting multi-scale features to generate a detection report containing defect three-dimensional coordinates, damage levels and safety risk assessment. The method achieves the purpose of efficiently and accurately detecting the building appearance defects, can adapt to complex building structures and environmental conditions, supports defect trend prediction and maintenance decision, and remarkably improves the building safety management efficiency.

Description

Technical Field

[0001] The present application relates to the field of building exterior defect detection, and particularly to a method and system for building exterior defect detection based on an unmanned aerial vehicle (UAV). Background Art

[0002] In recent years, UAV technology has been widely applied in multiple fields. Especially in the construction industry, its high-definition photographing and video recording functions provide new means for building exterior detection. Traditional building detection methods mainly rely on manual climbing or the use of lifting equipment for close observation, which not only has low efficiency but also poses high safety risks. The introduction of UAVs has significantly improved the detection efficiency, reduced costs, and at the same time reduced the safety hazards brought by human factors, playing an important role in promoting the modern development of the construction industry.

[0003] In the prior art, to solve the problem of building exterior defect detection, commonly used methods include manual visual inspection, long-distance observation with the aid of a telescope, and close inspection using lifting equipment, etc. In addition, there are also some image acquisition technologies based on ground-fixed equipment, which obtain building exterior images by setting up a camera array or a mobile shooting device. Although these methods can meet the detection requirements to a certain extent, they generally have problems such as limited coverage, complex operation, or high cost.

[0004] However, the prior art still has deficiencies in terms of detection comprehensiveness and efficiency. Especially when facing high-rise buildings or complex structures, it is difficult to achieve efficient and accurate detection of the entire exterior wall area. In addition, traditional methods have limited ability to identify some hidden defects (such as cracks, peeling, etc.), which easily lead to missed detections or misjudgments, thus affecting the accuracy of building safety assessment. Summary of the Invention

[0005] In a first aspect, in order to make the detection of building exteriors more efficient and accurate, the present application provides a method for building exterior defect detection based on a UAV.

[0006] The method and system for building exterior defect detection based on a UAV provided by the present application adopt the following technical solutions: A method for building exterior defect detection based on a UAV includes: Obtaining building information of a target building, where the building information includes building height information, exterior facade geometric feature information, and high-risk area marking information; Generating a layered scanning path and a three-dimensional obstacle avoidance flight trajectory based on the building information; Obtaining acquisition information based on the layered scanning path and the three-dimensional obstacle avoidance flight trajectory; Outputting a detection report based on the acquisition information.

[0007] By adopting the above technical solution, efficient and accurate detection of building exterior defects by drones can be achieved. First, by obtaining the building information of the target building, including building height, exterior facade geometric features, and high-risk area marking information, the system can comprehensively understand the structural characteristics and potential risk areas of the building, providing data support for subsequent detection. Based on this building information, the system can automatically generate layered scanning paths and three-dimensional obstacle avoidance flight trajectories to ensure that the drone can avoid obstacles and cover all important areas of the building during flight. By collecting data according to these flight trajectories, detailed information on the building exterior facade can be obtained for defect detection and analysis. Finally, based on the collected detection information, the system can automatically generate a detection report to help relevant personnel accurately evaluate the building exterior and provide a scientific basis for subsequent maintenance and repair work. This technical solution not only improves the automation and accuracy of building exterior defect detection but also ensures the safety and comprehensiveness of the detection process, providing strong technical support for the health monitoring of buildings.

[0008] Preferably, the method for planning the three-dimensional obstacle avoidance flight trajectory includes: Dividing the vertical detection interval based on the building height information, with a spiral progressive scanning path set in each interval; Obtaining meteorological information; Adjusting the detection distance between the aircraft and the building facade based on the meteorological information; Obtaining building corners and protruding areas based on the exterior facade geometric feature information; Obtaining multi-angle image information based on the building corners and protruding areas.

[0009] By adopting the above technical solution, the accuracy and safety of building exterior defect detection can be effectively improved. Specifically, this solution combines the building height information, meteorological conditions, and building geometric features to reasonably plan the flight trajectory and detection distance, ensuring that the aircraft can efficiently and comprehensively scan the building exterior facade at a safe distance while avoiding interference from obstacles and high-risk areas. In addition, the spiral progressive scanning path and multi-angle image acquisition can ensure detailed coverage of all corners of the building, improving the comprehensiveness and accuracy of defect detection.

[0010] Preferably, in the step of outputting a detection report based on the collected information, it includes: The collected information includes visible light images, infrared thermal maps, and lidar point cloud information; Performing spatio-temporal alignment processing on the visible light images, infrared thermal maps, and lidar point cloud information to generate an enhanced defect map containing surface cracks, hollowing and peeling, and leakage traces; Use a neural network with an attention mechanism to perform multi-scale feature extraction on the enhanced defect map and output a detection report with the three-dimensional coordinates of the defect, the damage level, and the safety risk assessment.

[0011] By adopting the above technical solution, the accuracy, comprehensiveness, and intelligence level of building exterior defect detection can be significantly improved. Specifically, through spatio-temporal alignment processing and the fusion of various sensor information (visible light images, infrared thermal maps, lidar point cloud information), a more refined and comprehensive defect map can be generated to capture problems such as surface cracks, hollowing and peeling, and leakage traces. Using a neural network with an attention mechanism for multi-scale feature extraction effectively enhances the defect recognition ability, and at the same time can accurately locate the three-dimensional coordinates of the defect, and evaluate its damage level and potential safety risk. This makes the detection report more credible and provides strong data support for the maintenance, repair, and safety management of buildings.

[0012] Preferably, the method further includes: Establish a defect feature library for high-rise buildings, which contains typical damage evolution patterns in different altitude ranges; According to the three-dimensional coordinates of the defect in the detection report, match the damage pattern in the corresponding height range, generate a prediction of the defect development trend considering the influence of wind load, and determine whether a continuously distributed hollow area is detected; If a continuously distributed hollow area is detected, automatically trigger the structural bearing capacity simulation calculation and mark the danger level.

[0013] By adopting the above technical solution, the intelligent diagnosis and prediction of high-rise building defects can be realized, and the accuracy and real-time performance of building safety assessment can be significantly improved. The established defect feature library combines the damage evolution patterns in different altitude ranges, enabling the system to perform targeted damage analysis and prediction according to the specific situation of the building. By matching the three-dimensional coordinates of the defect with the corresponding damage pattern and considering external factors such as wind load, the defect development trend can be predicted more accurately, and potential safety risks can be identified in advance. At the same time, if a continuously distributed hollow area is detected, the system can automatically trigger the structural bearing capacity simulation calculation, quickly evaluate its structural safety, and mark the danger level, providing a scientific basis for building maintenance, ensuring timely adoption of effective maintenance measures, and guaranteeing the long-term safe use of the building.

[0014] Preferably, the method further includes: Obtain aircraft information; Based on the aircraft information, divide the drone swarm into a first drone swarm and a second drone swarm; Perform height determination based on the building height information; Height determination: If the building height exceeds the set height, control the first drone swarm to perform facade partition scanning to obtain the first scanning information; Obtain the location information of the defect area based on the collected information; Based on the location information of the defect area, dispatch the second drone group to detect the detected defect area to obtain the second scan information; Generate a cloud map of the health status of the building facade based on the first scan information and the second scan information.

[0015] By adopting the above technical solution, efficient and accurate health monitoring and defect detection of the building facade can be achieved. By obtaining the information of the aircraft and reasonably dispatching the drone group, the drones are divided into two groups according to the height information of the building, so that the building facades at different heights can be scanned and detected more professionally. For buildings exceeding the set height, the first drone group is specifically responsible for the facade zoning scan to obtain detailed first scan information, while the second drone group conducts more precise detection according to the location of the defect area to provide the second scan information. By integrating the two scan information, a cloud map of the health status of the building facade is generated, which can clearly display the health status of the building and potential defect problems. This method not only improves the detection efficiency, but also ensures the comprehensiveness and accuracy of the scan data, helps to discover and handle the safety hazards of the building in time, and optimizes the maintenance plan.

[0016] Preferably, the method further includes: Obtain the building BIM model based on the building information of the target building; Visually annotate the three-dimensional coordinates of the defect on the building BIM model and generate the current detection information; Obtain the historical detection information; Generate a defect evolution timeline based on the current detection information and the historical detection information; Calculate the remaining service life of the building based on the defect evolution timeline and the material aging model and propose suggestions on the maintenance priority.

[0017] By adopting the above technical solution, intelligent evaluation of the building health status and maintenance optimization can be realized. First, obtain the building BIM model based on the building information of the target building. By visually annotating the three-dimensional coordinates of the defect on the model and generating the current detection information, the current health status of the building can be intuitively presented. Combining the obtained historical detection information, a defect evolution timeline can be generated to track the development process of the building defect, and based on the defect evolution timeline and the material aging model, calculate the remaining service life of the building. According to the calculation results, the system can automatically propose suggestions on the maintenance priority to help the management personnel formulate a reasonable repair plan according to the actual situation of the building, thereby improving the maintenance efficiency and extending the service life of the building. This method not only improves the accuracy of defect detection, but also optimizes the long-term maintenance management of the building, ensuring the safety and stable operation of the building.

[0018] Preferably, the method further includes: Obtain the inspection report and determine whether there are major structural defects; If major structural defects are detected, automatically plan the verification flight path; Control the drone to take key photos from multiple angles and obtain detailed verification information of the defective part; Modify the credibility evaluation of the inspection report based on the verification information.

[0019] By adopting the above technical solutions, the accuracy and credibility of building structure defect detection can be effectively improved. First of all, by obtaining the inspection report and determining whether there are major structural defects, key problems that may affect building safety can be identified in a timely manner. Once major structural defects are detected, the system will automatically plan the verification flight path and control the drone to take key photos from multiple angles to obtain detailed verification information of the defective part. Through these detailed verification information, the credibility evaluation of the inspection report can be further modified to ensure the accuracy and reliability of the report. This method can not only improve the ability to identify major structural defects, but also enhance the accuracy of the inspection report, providing a more scientific basis for subsequent maintenance and repair decisions, and ensuring the safety and stability of the building.

[0020] In the second aspect, in order to make the inspection of the building appearance more efficient and accurate, the present application provides a drone-based building appearance defect detection system.

[0021] The drone-based building appearance defect detection system provided by the present application adopts the following technical solutions: A drone-based building appearance defect detection system includes, A building information acquisition module for acquiring building information of the target building, where the building information includes building height information, external facade geometric feature information, and high-risk area marking information; A path planning module for generating a layered scanning path and a three-dimensional obstacle avoidance flight trajectory based on the building information; An information acquisition module for acquiring acquisition information based on the layered scanning path and the three-dimensional obstacle avoidance flight trajectory; A detection report generation module for outputting a detection report based on the acquisition information.

[0022] By adopting the above technical solutions, efficient and accurate detection of building appearance defects can be achieved. Through the building information acquisition module, the system can comprehensively obtain the height of the target building, the geometric features of the facade, and the marking information of high-risk areas, providing accurate data support for subsequent detection work. The path planning module automatically generates a stratified scanning path and a three-dimensional obstacle avoidance flight trajectory according to the building information, ensuring that the drone can safely and efficiently cover all areas of the building during flight, avoiding collisions and ensuring the comprehensiveness of detection. The information acquisition module collects building appearance data in real time by executing the predetermined flight trajectory, providing first-hand information for defect detection. Finally, the detection report generation module outputs a detection report based on the collected information, helping relevant personnel accurately evaluate the health status of the building facade and formulate corresponding maintenance and repair measures according to the report content. This method not only improves the automation degree of building defect detection, but also enhances the accuracy and safety of detection, contributing to scientific decision-making in building management.

[0023] In summary, the present application includes at least one of the following beneficial technical effects: 1. It not only improves the automation and accuracy of building appearance defect detection, but also ensures the safety and comprehensiveness of the detection process, providing strong technical support for the health monitoring of buildings; 2. The spiral progressive scanning path and multi-angle image acquisition can ensure detailed coverage of all corners of the building, improving the comprehensiveness and accuracy of defect detection; 3. The system can automatically propose maintenance priority suggestions, helping managers formulate reasonable maintenance plans according to the actual situation of the building, thereby improving maintenance efficiency and extending the service life of the building. This method not only improves the accuracy of defect detection, but also optimizes the long-term maintenance management of the building, ensuring the safe and stable operation of the building. Description of the Drawings

[0024] Figure 1 is a method flow block diagram of a method for detecting building appearance defects based on a drone according to an embodiment of the present application, mainly showing steps S100 - S400.

[0025] Figure 2 is a method flow block diagram of a method for detecting building appearance defects based on a drone according to an embodiment of the present application, mainly showing steps SA1 - SA5.

[0026] Figure 3 is a method flow block diagram of a method for detecting building appearance defects based on a drone according to an embodiment of the present application, mainly showing steps SB1 - SB3.

[0027] Figure 4 is a method flow block diagram of a method for detecting building appearance defects based on a drone according to an embodiment of the present application, mainly showing steps SD1 - SD7.

[0028] Figure 5 It is a flowchart of a method for detecting building appearance defects based on an unmanned aerial vehicle (UAV) in an embodiment of the present application, mainly showing steps SE1-SE5. Detailed implementation manners

[0029] The following further elaborates on the present application in conjunction with all the accompanying drawings.

[0030] An embodiment of the present application discloses a method for detecting building appearance defects based on an unmanned aerial vehicle. Referring to Figures 1 - 5 , a method for detecting building appearance defects based on an unmanned aerial vehicle includes: Step S100: Obtain building information of a target building, where the building information includes building height information, geometric features information of the outer facade, and high-risk area marking information.

[0031] Specifically, the building planning drawing, BIM (Building Information Model), or building design document of the target building can be obtained in advance by input or through networking, and the total height and height data of each floor of the building can be obtained by using them. These information helps to plan the flight height range and the hierarchical scanning path; the geometric features of the building outer facade (such as the shape of the wall surface, protruding parts, positions and sizes of windows and doors, etc.) can be obtained through building design drawings, BIM models, or by using laser scanning, 3D modeling technologies (such as structured light scanning or photogrammetry); by checking the design of the building, historical inspection reports, or based on expert experience, mark the areas in the building that may have potential safety hazards, such as old parts, cracks, corrosion areas, or high-risk areas (such as parts near the skyline, windows of super high-rise buildings, etc.).

[0032] Step S200: Generate a hierarchical scanning path and a three-dimensional obstacle avoidance flight trajectory based on the building information; Specifically, hierarchical scanning path: Based on the height information and geometric features of the outer facade of the building, the system can generate a series of hierarchical scanning paths. These paths cover each floor of the building and adjust the scanning frequency and detail according to the complexity of the building. For example, for the top of a high-rise building or a structurally complex area, the interval of the paths may be closer to ensure that each area can be comprehensively covered; Three-dimensional obstacle avoidance flight trajectory: According to the geometric features of the building and the high-risk area marking information, the system will generate a flight trajectory for the UAV to avoid the UAV from colliding with protruding objects of the building (such as balconies, windows, etc.). The planning of the flight trajectory will comprehensively consider external obstacles of the building, wind speed, flight height limits, etc. to ensure the safe flight of the UAV. The three-dimensional trajectory is generated by a modeling algorithm and can dynamically adjust the path to cope with environmental changes. The planning method of the three-dimensional obstacle avoidance flight trajectory includes: Step SA1: Divide the vertical detection intervals based on the building height information, and set a spiral progressive scanning path for each interval; Step SA2: Obtain meteorological information; Step SA3: Adjust the detection distance between the aircraft and the building facade based on the meteorological information; Step SA4: Obtain the building corners and protruding areas based on the geometric feature information of the facade; Step SA5: Obtain multi-angle image information based on the building corners and protruding areas.

[0033] Step S300: Obtain the acquisition information based on the hierarchical scanning path and the three-dimensional obstacle avoidance flight trajectory; Specifically, the drone executes the scanning task according to the planned flight trajectory, and uses sensors such as high-precision cameras, lidar (LiDAR), and infrared imaging devices to collect images or point cloud data of the building surface. These data include images of the building facade, depth information, and possible temperature differences (captured by infrared cameras), which help to detect defects such as cracks, structural damage, aging, or water vapor penetration; the collected data will be transmitted to the ground control system in real time for preliminary image processing and defect identification. If a problem area is found, the control system can adjust the drone flight trajectory for secondary shooting to ensure that the problem area is fully verified.

[0034] Step S400: Output a detection report based on the acquisition information.

[0035] The specific methods for the above steps include: Step SB1: The acquisition information includes visible light images, infrared thermal maps, and lidar point cloud information; Step SB2: Perform spatio-temporal alignment processing on the visible light images, infrared thermal maps, and lidar point cloud information to generate an enhanced defect map containing surface cracks, hollowing and peeling, and leakage traces; Step SB3: Use an attention mechanism neural network to perform multi-scale feature extraction on the enhanced defect map, and output a detection report with defect three-dimensional coordinates, damage levels, and safety risk assessments.

[0036] Specifically, all three types of acquisition information can be collected by the equipment carried by the drone. Specifically, a high-definition camera is used to take images of the building surface to capture visible defects such as cracks and hollowing and peeling. An infrared thermal imaging instrument is used to capture the temperature distribution map of the building surface, which can help detect defects caused by factors such as water leakage, thermal expansion and contraction, such as leakage traces and internal cavities in the wall. The building surface is scanned by lidar (LiDAR) to obtain three-dimensional point cloud data for analyzing the geometric shape and precise spatial information of the building surface to help identify structural defects.

[0037] The system needs to perform effective fusion of data from different sensors (visible light images, infrared thermal maps, and LiDAR point clouds). First, spatio-temporal alignment processing is required to ensure that the data is aligned in the same spatial and temporal reference frames. Spatial alignment: Use the point cloud data generated by LiDAR (Light Detection and Ranging) to establish a three-dimensional coordinate system of the building surface.

[0038] Project the visible light image and the infrared thermal map onto the same three-dimensional coordinate system through the internal and external parameters of the camera and the coordinate transformation model (such as camera calibration, distortion correction, etc.). This usually requires the combination of visual odometry or SLAM algorithms to ensure the precise alignment of the image and the point cloud data in space.

[0039] Use the geometric information of the point cloud data as a reference to correct the spatial position information of the infrared thermal map to ensure that the thermal map is geometrically consistent with the building surface.

[0040] Temporal alignment: Since the image, infrared thermal map, and point cloud data are usually collected at different time points, time synchronization is necessary. Ensure the temporal alignment of the data through timestamps, flight control systems, or sensor synchronization systems.

[0041] Based on the spatio-temporally aligned data, combine the visible light image, infrared thermal map, and LiDAR point cloud information to generate an enhanced defect map: Crack detection: Use image processing algorithms (such as edge detection, morphological processing, etc.) in the visible light image to identify cracks. Combine the point cloud data to extract the three-dimensional spatial positions of the cracks.

[0042] Hollow and peeling detection: Combine the infrared thermal map and the visible light image to detect the temperature change areas (which may indicate thermal expansion and contraction problems) and visible defect areas, and further combine the point cloud data to determine whether there are problems such as hollowing and peeling.

[0043] Leakage trace detection: Discover leakage areas (such as thermal bridges or damp areas) through the infrared thermal map, and combine the spatial information of the visible light image and the point cloud to enhance the detection and positioning of leakage traces.

[0044] Through the above processing, generate an enhanced map containing defects such as cracks, hollow and peeling, and leakage traces, where the position, type, size, etc. of each defect can be calibrated and visualized.

[0045] Use an attention mechanism neural network based on convolutional neural network (CNN) (such as self-attention mechanism Transformer, SENet, etc.) to process the enhanced defect map. The attention mechanism can help the model focus on the most important defect areas on the building surface.

[0046] Multi-scale convolution: Extract features of the image at different scales through convolutional kernels of different sizes to help capture detailed information on the building surface at different resolutions.

[0047] Feature fusion: Merge the features extracted from visible light images, infrared thermal maps, and point cloud data through feature fusion techniques to further enhance the complementarity of different sensor data.

[0048] Spatial attention: In the extracted multi-scale features, apply the spatial attention mechanism to focus on areas where defects are concentrated in the image, such as crack or peeling areas.

[0049] Defect classification and localization: Through a trained neural network model, classify the defects in the image (such as cracks, voids, leaks) and predict the three-dimensional coordinates of the defects to form a complete three-dimensional spatial position of the defects.

[0050] The network outputs information such as the type, location (coordinates), and severity level (damage level) of the defects.

[0051] Finally, output a detection report with the three-dimensional coordinates of the defects, damage level, and safety risk assessment. Three-dimensional coordinates of defects in the detection report: Extract the position coordinates of each defect from the output of the neural network, and combine with the three-dimensional spatial coordinate system of the point cloud data to output the three-dimensional coordinates of each defect; Damage level assessment: Based on the output of the neural network, evaluate the severity of the defects based on features such as the size, depth, and temperature change of the defects. Common criteria include crack width, depth, temperature change range, etc.; Safety risk assessment: Conduct a safety risk assessment by comprehensively analyzing the quantity, location, type, and severity level of the defects. For example, large-scale peeling at the top may require high attention, while some small cracks can be ignored.

[0052] Utilize existing building structure safety standards (such as "Building Structure Safety Assessment Standard") to assign a risk level (such as low risk, medium risk, high risk) to each defect and give corresponding repair suggestions.

[0053] Finally, the detection report can be displayed on human-computer interaction interfaces such as computers and mobile phones, and use data visualization tools (such as Tableau, Power BI, or a custom 3D visualization platform) to present the defect distribution map, three-dimensional model display, heat map overlay, etc., to help building managers more intuitively understand the building's health status.

[0054] In addition, there are other methods, specifically including: Step SC1: Establish a defect feature library for high-rise buildings, which includes typical damage evolution patterns in different altitude intervals; Step SC2: Match the three-dimensional coordinates of the defects in the inspection report with the damage patterns in the corresponding altitude intervals, generate a prediction of the defect development trend considering the influence of wind load, and determine whether a continuously distributed hollow area is detected; Step SC3: If a continuously distributed hollow area is detected, automatically trigger the structural bearing capacity simulation calculation and mark the danger level.

[0055] Specifically, a typical defect data set including different altitude intervals needs to be established. The data sources can include: Historical inspection data: Collect historical defect data of buildings in different altitude intervals (such as low-rise, mid-rise, high-rise), including cracks, hollowing, leakage, etc. Simulation data: According to the design and materials of the building, conduct simulation to generate defect evolution patterns in different altitude intervals. On-site monitoring data: Obtain on-site inspection data through sensors, drones or other detection devices, and establish defect characteristics related to altitude.

[0056] Classify and label the defect data in different altitude intervals, and establish a feature library. The feature library for each altitude interval should include: Defect type: such as cracks, hollowing, leakage, etc. Defect evolution pattern: including the early manifestations of the defect, evolution speed and possible influence range. Related environmental factors: such as wind speed, humidity, temperature, etc., which will affect the evolution of the defect.

[0057] Extract the three-dimensional coordinates of the defects (including information such as location and type) from the inspection report, and based on the actual structure and height of the building, match these coordinates with the altitude intervals in the defect feature library. Through coordinate data matching, associate the defect information of each area of the building with the typical damage evolution pattern in the corresponding altitude interval. For example, some defects may be more likely to appear in the upper part of high-rise buildings (affected by wind load), while the lower floors may be affected by different environmental and structural factors. Through the structural mechanics model, consider the influence of wind load on different altitude intervals of the building. The influence of wind load changes with altitude. Based on the existing wind mechanics model, combined with the structural data of the building, predict the defect evolution trend of the building under the action of wind load.

[0058] Specific prediction methods include: Wind load model: Establish a wind load model, considering factors such as building height, wind speed and wind direction.

[0059] Defect evolution model: Combine the material properties of the building and the influence of wind load to establish a physical model of defect evolution. Defect development prediction can be carried out through physical-based models (such as finite element analysis, mechanical models, etc.).

[0060] Machine learning and data fitting: Using machine learning models (such as regression analysis, time series analysis, etc.), predict the trend of defect evolution in different height intervals, and identify which areas will be at greater risk in the future and which defects may expand in the future.

[0061] Finally, use the combined results of lidar point cloud, infrared thermal map and visible light image to determine whether there are continuously distributed hollow areas in the building. By detecting continuous defects (such as the connectivity of hollow cracks) in the image or point cloud data, analyze whether these hollow defects continue to expand in a specific altitude interval, and set the judgment criteria for detecting hollow areas, such as the connectivity of defects (distance and size between adjacent defects), and the relevance to other defects (such as cracks and hollows occurring simultaneously). When the system detects a continuously distributed hollow area through step SC2, immediately trigger the structural simulation calculation program. Specific trigger conditions need to be set, such as the continuity of the hollow area, the size, density, and possible influence range of the defects.

[0062] Structural bearing capacity simulation calculation: Finite element analysis: According to the location, size, depth, etc. of the hollow area, calculate the structural stress and bearing capacity of this area through finite element analysis (FEA). Through simulation calculation, evaluate whether this area can maintain structural stability under the current load. Load and safety factor: According to the detected hollow defects and possible external loads (such as wind load, live load, etc.), evaluate the safety factor of this area. Structural mechanics formulas and software (such as ANSYS, ABAQUS) can be used for stress analysis and bearing capacity assessment. Local and overall bearing capacity: The influence of the hollow area on local and overall bearing capacity needs to be considered. Even if there are hollows in the local area, it is also necessary to conduct simulation evaluation on whether the overall structure can still ensure safety.

[0063] Finally, according to the structural simulation results (such as stress, deformation and other parameters) and design standards (such as "Building Structure Safety Assessment Standard"), give the risk level of the hollow area. Classify it as "low risk", "medium risk", "high risk" according to the severity of the defect, the bearing capacity calculation result and the safety factor. Generate a detection report with the risk level in the system, mark the risk level of the specific hollow area, and give repair or reinforcement suggestions according to the evaluation results.

[0064] It is also possible to establish a cloud map of the health status of the building facade. The specific methods include: Step SD1: Obtain aircraft information; Specifically, each drone obtains flight information through integrated GPS, IMU (inertial measurement unit), altitude sensor, etc., and transmits information to the main control system through a wireless communication system; use an aircraft management system or a drone control platform to obtain and manage the flight status and position of the drone swarm in real time.

[0065] Step SD2: Divide the UAV swarm into a first UAV swarm and a second UAV swarm based on the aircraft information; Specifically, according to information such as the capabilities and positions of the aircraft, divide the UAVs into a first UAV swarm (performing facade zoning scans) and a second UAV swarm (performing defect area detection). Dynamic allocation can be based on positions, flight task priorities, etc.

[0066] Step SD3: Perform height determination based on the building height information; Specifically, obtain the building height information through the building's CAD drawings, sensors (such as lidar, laser scanners), or GPS systems.

[0067] Step SD4: Height determination: If the building height exceeds the set height, control the first UAV swarm to perform facade zoning scans to obtain first scan information; Specifically, compare the building height with a preset threshold. If the height exceeds the set value, control the first UAV swarm to perform facade zoning scans. Based on the facade information of the building, divide the building into multiple scan areas, and the first UAV swarm scans by area to obtain images, laser point cloud data, or infrared scan data, etc. The UAVs automatically fly along the building facade to scan and obtain relevant images and data.

[0068] Step SD5: Obtain the position information of the defect area based on the collected information; Specifically, use image recognition technologies (such as computer vision, machine learning, etc.) to analyze the first scan information to identify defects (such as cracks, delamination, spalling, etc.). At the same time, auxiliary identification can be performed by analyzing sensor data (such as infrared thermal imagers, ultrasonic sensors, etc.). Based on the analysis results, mark the specific positions with defects and convert them into coordinate positions.

[0069] Step SD6: Based on the position information of the defect area, dispatch the second UAV swarm to detect the detected defect area to obtain second scan information; Specifically, according to the position information of the defect area obtained by the first UAV scan, automatically dispatch the second UAV swarm to detect the specific defect area. The second UAV swarm flies to the designated position to obtain more detailed scan information. The second UAV swarm can use devices such as high-resolution cameras and lidar for more accurate detection to obtain more detailed defect information.

[0070] Step SD7: Generate a cloud map of the health status of the building facade based on the first scan information and the second scan information.

[0071] Specifically, all the scanning information (images, point cloud data, etc.) collected by the first and second drone swarms is fused. Data processing and visualization tools (such as OpenCV, Matplotlib in Python, or 3D modeling tools) are used to generate a health status cloud map from the collected defect area data. The cloud map can present the health status of different areas, showing defect distribution, severity, etc. Based on the results generated from the cloud map, further analysis can be carried out to generate a building health status report and provide repair suggestions.

[0072] In addition, maintenance suggestions can also be put forward based on the remaining service life of the building. The specific methods include: Step SE1: Obtain the building BIM model based on the building information of the target building; Specifically, the three-dimensional geometry and structural information of the building are obtained through the building's BIM (Building Information Modeling) system. BIM software (such as Revit, ArchiCAD, Navisworks, etc.) can be used to export the BIM model, or interact with BIM data through an API interface (such as the API of Revit).

[0073] Step SE2: Visually annotate the three-dimensional coordinates of the defects on the building BIM model and generate the current detection information; Specifically, building defect detection is carried out through drones or other sensors (such as infrared scanning, laser scanning, ultrasonic, etc.) to obtain the three-dimensional coordinates of the defects (such as cracks, spalling, hollowing, etc.). The currently detected defect information is combined with the BIM model, and BIM visualization software or tools (such as Unity, Unreal Engine, etc.) are used to annotate the defect information at the corresponding positions on the BIM model.

[0074] Step SE3: Obtain historical detection information; Specifically, historical detection records are obtained from existing building maintenance management systems or detection systems. These records usually include past defect detection results, defect locations, types, severities, etc. The historical detection information is integrated with the current BIM model and defect data. A database management system (such as SQL) can be used to manage this historical information and combined with the BIM model through an API.

[0075] Step SE4: Generate a defect evolution timeline based on the current detection information and historical detection information; Specifically, the current detection data and historical detection data are combined, and the detection records of each defect are sorted in chronological order. For example, the occurrence time of the defect, severity changes, repair status, etc.

[0076] Step SE5: Calculate the remaining service life of the building based on the defect evolution timeline and the material aging model and put forward suggestions on maintenance priorities.

[0077] Specifically, based on the aging models of building materials (such as the deterioration of concrete, the corrosion of steel structures, etc.), predict the aging process of materials according to the evolution of defects. Common aging models include: Exponential decay model: Used to describe the trend of materials gradually decaying over time.

[0078] Linear / nonlinear regression model: Based on the evolution data of defects (such as crack propagation speed, corrosion rate, etc.) to predict future deterioration.

[0079] Calculate the remaining life: Based on the material aging model and defect evolution information, calculate the remaining service life of the building. Prediction can be achieved through simulation software (such as MATLAB, scikit-learn in Python) or professional building maintenance systems.

[0080] Evaluate the impact of each defect on the safety and function of the building according to the type, location, and size of the defect. Assign priorities to each defect based on the defect evolution timeline, material aging degree, and severity. A weighted scoring system (for example, combining factors such as the impact of the defect, occurrence frequency, location, etc.) can be used to determine which defects need to be repaired first. Generate maintenance priority suggestions for the building, and based on current and historical detection information, propose the parts that most need to be repaired, the urgency of repair, and the specific repair plan.

[0081] Finally, it is also possible to evaluate the credibility of the inspection report. The specific steps include: Step SF1: Obtain the inspection report and determine whether there are major structural defects; Step SF2: If major structural defects are detected, automatically plan a verification flight path; Step SF3: Control the drone to take key photos from multiple angles and obtain detailed verification information of the defect location; Step SF4: Correct the credibility assessment of the inspection report based on the verification information. Specifically, import the inspection report of the building into the system. The report usually includes the structural health monitoring data of the building, inspection results, defect types and locations. According to the defect information listed in the report (such as cracks, corrosion, displacement, etc.), determine whether it belongs to "major structural defects" according to preset standards or thresholds (such as crack width, location, impact, etc.). According to the location and nature of the defects (for example, wall cracks, roof leaks, etc.), plan the flight path of the drone; use a drone control system (such as DJISDK, Pixhawk, etc.) to provide flight path planning functions; automatically generate a flight path map to ensure that the drone can cover the defect area, and set reasonable flight heights, angles and speeds; the flight path can be automatically generated by software (such as Mission Planner, QGroundControl), and combined with the building's 3D model to determine the optimal perspective.

[0082] Execute this path through the drone control system, and control the drone to automatically adjust the shooting angle, focal length, etc. during the flight. Set multiple predetermined shooting points to shoot the defect area from different angles to ensure sufficient details are obtained. Use the high-resolution camera or other sensors (such as infrared cameras, laser scanners, etc.) of the drone to take detailed photos or videos of the defects.

[0083] Collect high-quality images or videos taken by the drone, perform image processing (such as image enhancement, feature extraction, etc.), and further verify the nature, size and severity of the defects. Compare the verified images of the drone with the defect information described in the original inspection report to determine whether there are errors.

[0084] If the verified information supports the defects in the original report, increase the credibility of the report. If the verification result deviates from the original report (such as misjudgment), then adjust the credibility score of the report. Use a credibility assessment algorithm (such as weighted average method, machine learning scoring model) to correct the original report according to the verified information.

[0085] The embodiment of the present application discloses a drone-based building exterior defect detection system.

[0086] A drone-based building exterior defect detection system includes A building information acquisition module for acquiring the building information of the target building, where the building information includes building height information, exterior facade geometric feature information and high-risk area marking information; A path planning module for generating a layered scanning path and a three-dimensional obstacle avoidance flight trajectory based on the building information; An information acquisition module for acquiring acquisition information based on the layered scanning path and the three-dimensional obstacle avoidance flight trajectory; A detection report generation module for outputting a detection report based on the acquisition information.

[0087] The implementation principle of an unmanned aerial vehicle (UAV)-based building exterior defect detection system in an embodiment of this application is as follows: it can achieve efficient and accurate detection of building exterior defects by UAVs. First, by obtaining the building information of the target building, including building height, geometric features of the facade, and marked information of high-risk areas, the system can comprehensively understand the structural characteristics and potential risk areas of the building, providing data support for subsequent detection. Based on this building information, the system can automatically generate layered scanning paths and three-dimensional obstacle avoidance flight trajectories to ensure that the UAV can avoid obstacles and cover all important areas of the building during flight. By collecting data according to these flight trajectories, detailed information about the building facade can be obtained for defect detection and analysis. Finally, based on the collected detection information, the system can automatically generate a detection report to help relevant personnel accurately evaluate the building exterior and provide a scientific basis for subsequent maintenance and repair work. This technical solution not only improves the automation and accuracy of building exterior defect detection but also ensures the safety and comprehensiveness of the detection process, providing strong technical support for the health monitoring of buildings.

[0088] The above are all preferred embodiments of this application. Without limiting the protection scope of this application accordingly, therefore, any equivalent changes made according to the structure, shape, and principle of this application shall be covered within the protection scope of this application.

Claims

1. A method for detecting building appearance defects based on drones, characterized in that: Including: Obtain the building information of the target building, where the building information includes building height information, external facade geometric feature information, and high-risk area marking information; Generate a layered scanning path and a three-dimensional obstacle avoidance flight trajectory based on the building information; Obtain the acquisition information based on the layered scanning path and the three-dimensional obstacle avoidance flight trajectory; Output a detection report based on the acquisition information.

2. The method for detecting building appearance defects based on an unmanned aerial vehicle according to claim 1, wherein: The method for planning the three-dimensional obstacle avoidance flight trajectory includes: Divide the vertical detection intervals based on the building height information, where a spiral progressive scanning path is set for each interval; Obtain meteorological information; Adjust the detection distance between the aircraft and the building facade based on the meteorological information; Obtain the building corners and convex areas based on the external facade geometric feature information; Obtain multi-angle image information based on the building corners and convex areas.

3. The method for detecting building appearance defects based on an unmanned aerial vehicle according to claim 1, wherein: In the step of outputting the detection report based on the acquisition information, it includes: The acquisition information includes visible light images, infrared thermal maps, and lidar point cloud information; Perform spatio-temporal alignment processing on the visible light images, infrared thermal maps, and lidar point cloud information to generate an enhanced defect map containing surface cracks, hollowing and peeling, and leakage traces; Use an attention mechanism neural network to perform multi-scale feature extraction on the enhanced defect map and output a detection report with defect three-dimensional coordinates, damage levels, and safety risk assessments.

4. The method for detecting building appearance defects based on an unmanned aerial vehicle according to claim 1, characterized in that: The method further includes: Establish a high-rise building defect feature library, which contains typical damage evolution patterns in different altitude intervals; Match the damage patterns in the corresponding height intervals according to the defect three-dimensional coordinates in the detection report, generate a prediction of the defect development trend considering the influence of wind load, and determine whether a continuously distributed hollow area is detected; If a continuously distributed hollow area is detected, automatically trigger the structural bearing capacity simulation calculation and mark the danger level.

5. The method for detecting building appearance defects based on an unmanned aerial vehicle according to claim 1, wherein: The method further includes: Obtain the aircraft information; Divide the drone swarm into a first drone swarm and a second drone swarm based on the aircraft information; Perform height determination based on the building height information; Height determination: If the building height exceeds the set height, control the first drone swarm to perform facade partition scanning to obtain the first scanning information; Obtain the position information of the defect area based on the acquisition information; Based on the position information of the defect area, dispatch the second drone swarm to detect the detected defect area to obtain the second scanning information; Generate a building external facade health status cloud map based on the first scanning information and the second scanning information.

6. The method for detecting building appearance defects based on an unmanned aerial vehicle according to claim 1, wherein: The method further includes: Obtain the building BIM model based on the building information of the target building; Visually mark the defect three-dimensional coordinates on the building BIM model and generate the current detection information; Obtain the historical detection information; Generate a defect evolution timeline based on the current detection information and the historical detection information; Calculate the remaining service life of the building based on the defect evolution timeline and the material aging model and propose maintenance priority suggestions.

7. A method for detecting building appearance defects based on an unmanned aerial vehicle according to claim 1, characterized in that: The method further includes: Obtain the detection report and determine whether there are major structural defects; If major structural defects are detected, automatically plan a verification flight path; Control the drone to take key shots from multiple angles and obtain detailed verification information of the defect part; Correct the credibility assessment of the detection report based on the verification information.

8. An unmanned aerial vehicle-based building exterior defect detection system, characterized in that: Including, A building information acquisition module, which is used to acquire the building information of a target building, and the building information includes building height information, facade geometric feature information, and high-risk area marking information; A path planning module, which is used to generate a layered scanning path and a three-dimensional obstacle avoidance flight trajectory based on the building information; An information acquisition module, which is used to acquire acquisition information based on the layered scanning path and the three-dimensional obstacle avoidance flight trajectory; A detection report generation module, which is used to output a detection report based on the acquisition information.

Images