Automatic Inspection Method and System for High-Rise Hyperbolic Glass Curtain Wall Based on UAV

Abstract

The present application provides an automatic inspection method and system for high-rise hyperbolic glass curtain walls based on unmanned aerial vehicles. The system includes an unmanned aerial vehicle, a data acquisition module, and a controller. The data acquisition module is carried on the unmanned aerial vehicle and includes an ultrasonic sensor array, an infrared thermal imager, and an image acquisition module. The controller is used to perform time-frequency domain analysis on the reflected wave to obtain the reflected wave signal intensity. The controller calculates the temperature gradient of the surface temperature distribution image of the structural adhesive to obtain the characteristics of the temperature anomaly area, obtains the optical image texture according to the surface optical image, and outputs the inspection result according to the reflected wave signal intensity, the characteristics of the temperature anomaly area, and the optical image texture. Generate the flight trajectory of the unmanned aerial vehicle according to the three-dimensional model and curvature parameters of the hyperbolic glass curtain wall, generate the obstacle avoidance position information according to the environmental information set by the hyperbolic glass curtain wall, and update the flight trajectory of the unmanned aerial vehicle according to the obstacle avoidance position information. Control the unmanned aerial vehicle to automatically inspect the hyperbolic glass curtain wall according to the flight trajectory of the unmanned aerial vehicle.

Description

Technical Field

[0001] This application relates to the technical field of automatic inspection, and particularly to an automatic inspection method and system for high-rise hyperbolic glass curtain walls based on unmanned aerial vehicles (UAVs). Background Art

[0002] With the acceleration of the urbanization process, high-rise buildings increasingly adopt hyperbolic glass curtain walls as part of their facade design. This design is not only beautiful and generous but also provides good natural lighting and views. However, the maintenance and inspection of hyperbolic glass curtain walls face many challenges. Traditional inspection methods usually require manual climbing or the use of equipment such as gondolas for close inspection. This method is not only inefficient but also poses significant safety hazards. In addition, due to the complex geometric shape of hyperbolic glass curtain walls, it is difficult for manual inspection to comprehensively cover all areas, and it is easy to miss subtle defects.

[0003] In recent years, the development of UAV technology has provided new solutions for the inspection of high-rise buildings. However, existing UAV inspection systems still have some deficiencies when applied to hyperbolic glass curtain walls. For example, a single image acquisition module cannot comprehensively capture various defect information of the glass curtain wall; the lack of an effective obstacle avoidance mechanism results in the UAV being prone to collisions in complex environments; moreover, the lack of a targeted flight path planning for the special geometric shape of hyperbolic glass curtain walls leads to low inspection efficiency and accuracy.

[0004] Therefore, there is an urgent need for a system and its corresponding method to solve at least one of the above problems. Summary of the Invention

[0005] This application provides an automatic inspection method and system for high-rise hyperbolic glass curtain walls based on UAVs, aiming to solve the deficiencies that still exist in existing UAV inspection systems when applied to hyperbolic glass curtain walls. For example, a single image acquisition module cannot comprehensively capture various defect information of the glass curtain wall; the lack of an effective obstacle avoidance mechanism results in the UAV being prone to collisions in complex environments; moreover, the lack of a targeted flight path planning for the special geometric shape of hyperbolic glass curtain walls leads to low inspection efficiency and accuracy.

[0006] In a first aspect, this application provides an automatic inspection system for high-rise hyperbolic glass curtain walls based on UAVs, including:

[0007] An unmanned aerial vehicle (UAV);

[0008] Data acquisition module, the data acquisition module is carried on the drone, and the data acquisition module includes an ultrasonic sensor array, an infrared thermal imager and an image acquisition module. Among them, the ultrasonic sensor array is used to emit high-frequency ultrasonic signals to the hyperbolic glass curtain wall to be inspected and receive the reflected waves. The infrared thermal imager is used to capture the surface temperature distribution image of the structural glue of the hyperbolic glass curtain wall. The image acquisition module is used to collect the surface optical image of the hyperbolic glass curtain wall;

[0009] Controller, the controller is used to perform time-frequency domain analysis on the reflected wave to obtain the reflected wave signal intensity; the controller calculates the temperature gradient of the surface temperature distribution image of the structural glue to obtain the characteristics of the temperature abnormal area; the controller obtains the optical image texture according to the surface optical image; the controller outputs the inspection result according to the reflected wave signal intensity, the characteristics of the temperature abnormal area and the optical image texture; the controller generates the drone flight trajectory according to the three-dimensional model and curvature parameters of the hyperbolic glass curtain wall, generates the obstacle avoidance position information according to the environmental information set by the hyperbolic glass curtain wall, and updates the drone flight trajectory according to the obstacle avoidance position information; the controller controls the drone to automatically inspect the hyperbolic glass curtain wall according to the drone flight trajectory.

[0010] In some embodiments, the data acquisition module includes: a polarization filter, and the polarization filter is arranged at the front end of the image acquisition module to suppress the surface reflection interference of the hyperbolic glass curtain wall.

[0011] In some embodiments, the controller generates an ultrasonic signal attenuation coefficient according to the reflected wave signal intensity and the ultrasonic signal intensity; the controller calculates the proportion of the temperature abnormal area according to the characteristics of the temperature abnormal area and the area corresponding to the hyperbolic glass curtain wall; the controller determines the crack morphology characteristics according to the optical image texture; the controller outputs the inspection result according to the ultrasonic signal attenuation coefficient, the proportion of the temperature abnormal area and the crack morphology characteristics; the inspection result includes the defect type, the defect position and the defect severity score.

[0012] Exemplarily, the controller inputs the ultrasonic signal attenuation coefficient, the proportion of the temperature abnormal area, and the crack morphology curvature corresponding to the crack morphology characteristics into a deep learning model based on the attention mechanism, and the deep learning model outputs the inspection result.

[0013] In some embodiments, the controller generates a fitness function according to the curvature change rate of the hyperbolic glass curtain wall, the real-time pose vector corresponding to the drone, the estimated total inspection duration, and the battery life threshold; the controller generates the drone flight trajectory based on the genetic algorithm according to the fitness function.

[0014] Exemplarily, the expression of the fitness function includes:

[0015] ;

[0016] where is the fitness function, is the curvature smoothing weight coefficient, used to adjust the importance of the curvature change rate in the fitness function, is the total number of flight segments corresponding to the UAV flight trajectory, is the curvature change rate of the curtain wall in the flight segment, indicating the degree of change in the curvature of the curtain wall in a certain segment of the flight path of the UAV, and is used to measure the bending change of the path, is the estimated total inspection duration, is the battery endurance threshold, indicating the maximum allowable time for a single flight of the UAV, is the time efficiency weight coefficient, adjusting the importance of the inspection time in the fitness function, is the trajectory second-order difference penalty term coefficient, used to control the smoothness of the trajectory, , and are the pose vectors of the UAV at , and times respectively, including position and attitude information, and the second-order difference term is used to measure the acceleration change of the trajectory.

[0017] It should be noted that in some embodiments, the expression of the curvature smoothing weight coefficient includes:

[0018] ;

[0019] where is the curvature smoothing weight coefficient, is the average curvature change gradient, representing the average value of the absolute values of the curvature change rates of all flight segments, reflecting the bending degree of the overall path, is the curvature change rate threshold, the critical value for triggering weight adjustment. When the average curvature exceeds , increases.

[0020] In some embodiments, it further includes: a meteorological sensing unit for real-time monitoring of the wind speed information and precipitation information corresponding to the drone; an autonomous obstacle avoidance unit for detecting the convex structure of the hyperbolic glass curtain wall and temporary obstacles through the fusion of millimeter-wave radar and binocular vision, and controlling the drone to perform autonomous obstacle avoidance when the distance to the convex structure and temporary obstacles is less than a preset distance; a multi-link redundant communication unit carried on the drone to transmit data in parallel with the controller using LoRa and 4G dual channels.

[0021] In some embodiments, the controller constructs a curtain wall performance degradation prediction model based on the historical inspection data corresponding to the hyperbolic glass curtain wall. The curtain wall performance degradation prediction model analyzes the defect expansion rate and the correlation of environmental factors corresponding to the hyperbolic glass curtain wall according to the inspection results, and the controller outputs an adaptive adjustment suggestion for the inspection period according to the defect expansion rate and the correlation of environmental factors.

[0022] In a second aspect, the present application provides a method for automatically inspecting a high-rise hyperbolic glass curtain wall based on a drone, which is applied to the controller of the system provided in any embodiment of the present application. The method includes:

[0023] Performing time-frequency domain analysis on the reflected waves received by the ultrasonic sensor array to obtain the reflected wave signal intensity;

[0024] Calculating the temperature gradient of the surface temperature distribution image of the structural adhesive captured by the infrared thermal imager to obtain the characteristics of the temperature anomaly area;

[0025] Obtaining the optical image texture according to the surface optical image collected by the image acquisition module;

[0026] Generating a drone flight trajectory according to the three-dimensional model and curvature parameters of the hyperbolic glass curtain wall;

[0027] Generating obstacle avoidance position information according to the environmental information set for the hyperbolic glass curtain wall, and updating the drone flight trajectory according to the obstacle avoidance position information;

[0028] Controlling the drone to automatically inspect the hyperbolic glass curtain wall according to the drone flight trajectory.

[0029] In a third aspect, the present application provides an automatic inspection device for a high-rise hyperbolic glass curtain wall based on a drone, including:

[0030] A time-frequency analysis module for performing time-frequency domain analysis on the reflected waves received by the ultrasonic sensor array to obtain the reflected wave signal intensity;

[0031] An anomaly acquisition module for calculating the temperature gradient of the surface temperature distribution image of the structural adhesive captured by the infrared thermal imager to obtain the characteristics of the temperature anomaly area;

[0032] An image acquisition module, configured to obtain an optical image texture based on the surface optical image acquired by the image acquisition module;

[0033] A trajectory generation module, configured to generate a UAV flight trajectory according to the three-dimensional model of the hyperbolic glass curtain wall and the curvature parameter;

[0034] A trajectory update module, configured to generate obstacle avoidance position information according to the environmental information set for the hyperbolic glass curtain wall, and update the UAV flight trajectory according to the obstacle avoidance position information;

[0035] An inspection control module, configured to control the UAV to automatically inspect the hyperbolic glass curtain wall according to the UAV flight trajectory.

[0036] In a fourth aspect, the present application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, causes one or more processors to execute the method provided in any embodiment of the present application.

[0037] An automatic inspection method and system for a high-rise hyperbolic glass curtain wall based on a UAV provided by an embodiment of the present application, the system aims to overcome the deficiencies existing in the existing UAV inspection technology when applied to hyperbolic glass curtain walls. Specifically, the system integrates a variety of sensors and intelligent control algorithms to achieve a more comprehensive, efficient and safe inspection of hyperbolic glass curtain walls. The following is the detailed technical content of the system:

[0038] System composition:

[0039] UAV: As the main flight platform, carrying a data acquisition module and a controller.

[0040] Data acquisition module: Includes an ultrasonic sensor array, an infrared thermal imager, and an image acquisition module, configured to collect information of the hyperbolic glass curtain wall from different dimensions.

[0041] Ultrasonic sensor array: Emits high-frequency ultrasonic signals and receives reflected waves, and detects defects inside or on the surface of the glass curtain wall by analyzing the reflected waves.

[0042] Infrared thermal imager: Captures the surface temperature distribution image of the structural adhesive of the glass curtain wall to help identify temperature anomaly areas caused by structural problems.

[0043] Image acquisition module: Acquires the surface optical image of the glass curtain wall to facilitate the discovery of appearance damage or other visible problems.

[0044] Controller: Responsible for processing data from each sensor and generating corresponding inspection results; at the same time, planning the flight path of the UAV according to the specific geometric features of the glass curtain wall and implementing an obstacle avoidance strategy.

[0045] The corresponding working process of the system includes:

[0046] Data collection and preliminary processing: Use an ultrasonic sensor array to emit signals towards the target and analyze the echoes to evaluate the material condition. The infrared thermal imager records the temperature distribution, paying special attention to the locations that may indicate potential fault points. The image acquisition module takes high-resolution photos for subsequent visual inspection.

[0047] Data analysis: The controller comprehensively analyzes the above three types of data, including but not limited to the reflection wave signal intensity, temperature gradient changes, and texture features, etc. Based on this information, it determines whether there are any problem areas that require further investigation.

[0048] Path planning and execution: Based on the pre-constructed three-dimensional model and known curvature parameters, the controller calculates the optimal flight route. Combining real-time environmental perception (such as the location of obstacles), it dynamically adjusts the trajectory to ensure safe operation. Finally, it guides the drone to complete the entire inspection task according to the optimized path.

[0049] The provided system has at least the following beneficial effects:

[0050] Improve detection accuracy: By combining multiple sensing technologies, different types of problems can be captured more comprehensively, thus improving the ability to discover problems.

[0051] Enhance safety: Introducing an advanced obstacle avoidance mechanism can effectively avoid the risk of the drone colliding with surrounding objects, ensuring the safety of the operation process.

[0052] Improve efficiency: Automated path planning reduces the need for manual intervention, making the entire inspection process smoother and faster, saving a large amount of time costs.

[0053] Strong adaptability: The functions specially developed for the hyperboloid design enable it to better serve this special architectural form, showing good flexibility and scope of application.

[0054] Cost-effectiveness: In the long run, adopting this automated solution can significantly reduce maintenance costs while extending the service life of the building, bringing economic benefits to the owner.

[0055] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit this application. Brief Description of the Drawings

[0056] To more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can also be obtained based on these drawings.

[0057] Figure 1 is a schematic structural diagram of an automatic inspection system for high-rise hyperbolic glass curtain walls based on unmanned aerial vehicles provided by an embodiment of the present application;

[0058] Figure 2 is a schematic diagram of the inspection principle of an automatic inspection system for high-rise hyperbolic glass curtain walls based on unmanned aerial vehicles provided by an embodiment of the present application;

[0059] Figure 3 is a schematic flowchart of the steps of a method for automatically inspecting high-rise hyperbolic glass curtain walls based on unmanned aerial vehicles provided by an embodiment of the present application;

[0060] Figure 4 is a schematic block diagram of the structure of a controller provided by an embodiment of the present application.

[0061] It should be understood that the above general description and the following detailed description are only exemplary and explanatory, and cannot limit the present application. Detailed Embodiments

[0062] The following will clearly and completely describe the technical solutions in the embodiments of the present application in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are some, but not all, of the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts fall within the scope of protection of the present application.

[0063] The flowchart shown in the accompanying drawings is only an example, and does not necessarily include all the contents and operations / steps, nor does it necessarily execute in the described order. For example, some operations / steps can be decomposed, combined, or partially merged, so the actual execution order may change according to the actual situation.

[0064] It should be understood that in order to clearly describe the technical solutions of the embodiments of the present invention, in the embodiments of the present invention, terms such as "first" and "second" are used to distinguish the same items or similar items with basically the same functions and effects. Those skilled in the art can understand that the terms "first" and "second" do not limit the quantity and execution order, and the terms "first" and "second" do not necessarily mean different.

[0065] It should be understood that the terms used in the specification of this application are merely for the purpose of describing specific embodiments and are not intended to limit this application. As used in the specification of this application and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include the plural forms.

[0066] It should also be understood that the term "and / or" used in the specification of this application and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0067] The following will, with reference to the accompanying drawings, elaborate on some embodiments of this application. Without conflict, the following embodiments and the features in the embodiments may be combined with each other.

[0068] With the acceleration of the urbanization process, hyperbolic glass curtain walls are increasingly adopted in high-rise buildings as part of their facade design. This design is not only beautiful and generous but also can provide good natural lighting and views. However, the maintenance and inspection of hyperbolic glass curtain walls face many challenges. Traditional inspection methods usually require manual climbing or the use of equipment such as gondolas for close inspection. This method is not only inefficient but also poses a relatively large safety hazard. In addition, due to the complex geometric shape of hyperbolic glass curtain walls, it is difficult for manual inspection to comprehensively cover all areas and it is easy to miss subtle defects.

[0069] In recent years, the development of drone technology has provided new solutions for the inspection of high-rise buildings. However, there are still some deficiencies in the existing drone inspection systems when applied to hyperbolic glass curtain walls. For example, a single image acquisition module cannot comprehensively capture various defect information of the glass curtain wall; the lack of an effective obstacle avoidance mechanism results in the drone being prone to collisions in complex environments; and, there is a lack of targeted flight path planning for the special geometric shape of hyperbolic glass curtain walls, resulting in low inspection efficiency and accuracy.

[0070] Therefore, there is an urgent need for a system and its corresponding method to solve at least one of the above problems.

[0071] To solve the above problems, please refer to Figure 1 . As Figure 1As shown in the figure, this application provides an automatic inspection system for high-rise hyperbolic glass curtain walls based on drones, including: drones, a data acquisition module (not shown in the figure), the data acquisition module is carried on the drone, and the data acquisition module includes an ultrasonic sensor array, an infrared thermal imager, and an image acquisition module. Among them, the ultrasonic sensor array is used to emit high-frequency ultrasonic signals to the hyperbolic glass curtain wall to be inspected and receive the reflected waves. The infrared thermal imager is used to capture the surface temperature distribution image of the structural glue of the hyperbolic glass curtain wall. The image acquisition module is used to collect the surface optical image of the hyperbolic glass curtain wall; a controller, the controller is used to perform time-frequency domain analysis on the reflected waves to obtain the reflected wave signal intensity; the controller calculates the temperature gradient of the surface temperature distribution image of the structural glue to obtain the characteristics of the temperature anomaly area; the controller obtains the optical image texture according to the surface optical image; the controller outputs the inspection result according to the reflected wave signal intensity, the characteristics of the temperature anomaly area, and the optical image texture; the controller generates a drone flight trajectory according to the three-dimensional model and curvature parameters of the hyperbolic glass curtain wall, generates obstacle avoidance position information according to the environmental information set by the hyperbolic glass curtain wall, and updates the drone flight trajectory according to the obstacle avoidance position information; the controller controls the drone to automatically inspect the hyperbolic glass curtain wall according to the drone flight trajectory.

[0072] Specifically, the purpose of this system is to provide an efficient, safe, and comprehensive automatic inspection solution for hyperbolic glass curtain walls used in high-rise buildings by combining advanced sensor technology, image processing technology, and path planning algorithms. This system mainly consists of the following key components:

[0073] Drone: As a platform for carrying various detection devices, it has good flight stability and mobility.

[0074] Data acquisition module: Ultrasonic sensor array: Used to emit high-frequency ultrasonic signals to the hyperbolic glass curtain wall to be inspected and receive the reflected waves to detect whether there are problems such as cavities or cracks in the internal structure of the curtain wall. Infrared thermal imager: Can capture the surface temperature distribution of the glass curtain wall, especially at its joints (such as parts using structural glue), so as to identify areas that may have potential hazards. Image acquisition module: Responsible for obtaining high-definition optical images of the outer surface of the curtain wall to detect external damage or other visible defects.

[0075] Controller: Integrates powerful data processing capabilities. It can not only analyze and process various information collected, but also automatically generate the optimal flight route according to preset conditions and adjust it in real time to avoid obstacles.

[0076] As Figure 2 shown, the specific implementation of this system includes the following stages:

[0077] Initialization phase: First, an accurate three-dimensional model of the target building, especially its hyperbolic glass curtain wall part (algorithm model preparation as shown), needs to be established, and this model together with relevant environmental parameters is input into the control system. Figure 2 During the preparation for inspection tours: Based on the above three-dimensional model and geometric characteristics such as curvature, the controller will pre-calculate an optimal flight trajectory (cruise route planning and design as shown) that covers the entire inspection range. At the same time, considering other buildings or objects that may exist in the actual operating environment, some obstacle avoidance points will also be set additionally.

[0078] Performing the inspection task (corresponding to the cruise test with safety measures in Figure 2 ): Start the drone to work according to the established route. During this process, each sensor (drone system composition and component preparation in

[0079] ) continuously collects relevant data and transmits it to the ground station or directly performs preliminary processing on the on-board computer. The ultrasonic sensor array regularly sends pulse signals and records the time difference and intensity change of the echo. These data help to evaluate the internal condition of the material. The infrared camera focuses on monitoring the change trend of the temperature difference on the wall surface at different time periods. Abnormally high or low temperature areas often indicate potential problems. The high-definition camera is responsible for taking clear photos or video streams for more detailed inspection of details later. Figure 2 All the collected information will ultimately be aggregated on a unified platform and, after further analysis by specialized software, form a complete diagnostic report (corresponding to the implementation effect evaluation and optimization in Figure 2 ). The report not only contains text descriptions but also intuitive charts for easy understanding of the problems and their severity by users.

[0080] Furthermore, the provided system has at least the following beneficial effects: Figure 2 Improving efficiency: Compared with the traditional manual operation mode, using drones can significantly shorten the time required to complete a comprehensive inspection.

[0081] Enhancing safety: Reducing the chance of workers being exposed to high-risk environments and lowering the possibility of accidents.

[0082] Improving accuracy: The application of multi-modal sensing technology makes it difficult for even the smallest flaws to escape the system's monitoring, ensuring higher detection accuracy.

[0083] Enhancing safety: Reducing the chance of workers being exposed to high-risk environments and lowering the possibility of accidents.

[0084] Improving accuracy: The application of multi-modal sensing technology makes it difficult for even the smallest flaws to escape the system's monitoring, ensuring higher detection accuracy.

[0085] Cost - benefit: In the long run, although the initial investment is large, with the reduction of operation and maintenance costs, the overall economic benefits will gradually emerge.

[0086] High flexibility: Different sensor combinations can be flexibly configured according to specific requirements to meet the special requirements in various application scenarios.

[0087] Promote sustainable development: By detecting and fixing problems in a timely manner, the service life of the building is extended, resource waste is reduced, which is conducive to achieving the goal of green and low - carbon.

[0088] In summary, this automatic inspection system for high - rise hyperbolic glass curtain walls based on drones not only solves the current technical problems but also provides strong support for the future construction of smart cities.

[0089] In some embodiments, the data acquisition module includes: a polarization filter, which is arranged at the front end of the image acquisition module and is used to suppress the surface reflection interference of the hyperbolic glass curtain wall.

[0090] In an embodiment, the data acquisition module includes a polarization filter, which is set at the front end of the image acquisition module. The main function of the polarization filter is to suppress the reflection interference on the surface of the hyperbolic glass curtain wall, thereby improving the quality and accuracy of image acquisition.

[0091] Select a suitable polarization filter to ensure that it can effectively reduce the reflected light on the glass surface while not affecting the transmission of other light rays. The image acquisition module uses a high - resolution camera to capture the surface optical images of the hyperbolic glass curtain wall.

[0092] The drone starts the inspection task according to the preset flight trajectory. The image acquisition module takes pictures of the surface of the hyperbolic glass curtain wall through the polarization filter. The polarization filter can significantly reduce the reflected light caused by sunlight or other light sources, making the image clearer. The captured images are transmitted to the ground station or the on - board controller in real time through the wireless communication module for processing. The controller processes the collected images to identify possible defects such as cracks and scratches. According to the image analysis results, a detailed inspection report is generated, including the type, location, and severity of the defects.

[0093] The polarization filter can significantly reduce the reflection on the glass surface, making the captured images clearer and improving the detection accuracy. By reducing the reflection interference, the system can more accurately identify subtle defects such as tiny cracks or scratches. High - quality images reduce the time for post - processing, improving the overall inspection efficiency. Reducing the reflection interference helps reduce false alarms caused by changes in lighting conditions and improves the reliability of the system.

[0094] In some embodiments, the controller generates an ultrasonic signal attenuation coefficient based on the reflected wave signal intensity and the ultrasonic signal intensity; the controller calculates the proportion of the temperature anomaly area in the total area according to the characteristics of the temperature anomaly area and the area corresponding to the hyperbolic glass curtain wall; the controller determines the crack morphology characteristics based on the optical image texture; the controller outputs the inspection result according to the ultrasonic signal attenuation coefficient, the proportion of the temperature anomaly area, and the crack morphology characteristics; the inspection result includes the defect type, the defect location, and the defect severity score.

[0095] The controller conducts comprehensive analysis based on multiple sensor data to generate a detailed inspection result. Data collection: Ultrasonic sensor array: emits high-frequency ultrasonic signals and receives reflected waves. Infrared thermal imager: captures the surface temperature distribution image of the structural glue of the hyperbolic glass curtain wall. Image acquisition module: acquires the surface optical image of the hyperbolic glass curtain wall.

[0096] The controller performs time-frequency domain analysis on the reflected wave to obtain the reflected wave signal intensity and calculates the ultrasonic signal attenuation coefficient. The controller calculates the temperature gradient of the surface temperature distribution image of the structural glue to identify the temperature anomaly area and calculates the proportion of the temperature anomaly area. Optical image texture analysis: The controller determines the crack morphology characteristics based on the surface optical image. The ultrasonic signal attenuation coefficient, the proportion of the temperature anomaly area, and the crack morphology characteristics are input into a deep learning model based on the attention mechanism to output the inspection result. The inspection result includes the defect type, the defect location, and the defect severity score.

[0097] Combining multiple data sources such as ultrasonic, infrared thermal imaging, and optical images provides more comprehensive detection information. Through the deep learning model, different types of defects can be identified and classified more accurately. The system automatically analyzes the data and generates an inspection report, reducing manual intervention and improving work efficiency. Real-time processing and analysis of data can provide detailed inspection results in a short time, facilitating timely measures to be taken.

[0098] Exemplarily, the controller inputs the ultrasonic signal attenuation coefficient, the proportion of the temperature anomaly area, and the crack morphology curvature corresponding to the crack morphology characteristics into a deep learning model based on the attention mechanism, and the deep learning model outputs the inspection result.

[0099] Input data: Ultrasonic signal attenuation coefficient, proportion of temperature anomaly area, crack morphology characteristics (including crack morphology curvature). Label data: Defect type, location, and severity score in the historical inspection record.

[0100] The model architecture adopts a deep learning model based on the attention mechanism, such as the Transformer model. A large amount of historical data is used for training to optimize the model parameters so that it can accurately predict the defect type, location, and severity.

[0101] Inference process: Data input: Input the data collected during the current inspection into the trained model. Result output: The model outputs the inspection results, including the defect type, location, and severity score.

[0102] The attention mechanism can better capture key features and improve the accuracy of defect recognition. The model can be adaptively adjusted according to different data inputs and is applicable to various complex environments. The model has a fast inference speed and can generate inspection results in a short time, improving the real-time performance of the system. The attention mechanism provides interpretability, helps to understand the model decision-making process, and enhances the transparency of the system.

[0103] In some embodiments, the controller generates a fitness function based on the curvature change rate of the hyperbolic glass curtain wall, the corresponding real-time pose vector of the drone, the estimated total inspection duration, and the battery endurance threshold; the controller generates the flight trajectory of the drone based on the fitness function using a genetic algorithm.

[0104] In an embodiment, the controller generates a fitness function based on the curvature change rate of the hyperbolic glass curtain wall, the real-time pose vector of the drone, the estimated total inspection duration, and the battery endurance threshold, and generates an optimal flight trajectory based on a genetic algorithm. By considering the curvature change rate and smoothness, the generated flight trajectory is smoother, reducing the violent movement of the drone and improving flight safety. By optimizing the inspection time and path length, the inspection efficiency is improved and the battery endurance time of the drone is extended. The fitness function can dynamically adjust the weight coefficients according to actual needs, improving the flexibility and adaptability of the system. The genetic algorithm can search for the global optimal solution, avoiding the problem of local optimal solutions and improving the quality of path planning.

[0105] Exemplarily, the expression of the fitness function includes:

[0106] ;

[0107] where is the fitness function, is the curvature smoothness weight coefficient, used to adjust the importance of the curvature change rate in the fitness function, is the total number of flight segments corresponding to the flight trajectory of the drone, is the curvature change rate of the curtain wall in the is the estimated total inspection duration, is the battery life threshold, representing the maximum allowable time for a single flight of the drone, is the time efficiency weight coefficient, which adjusts the importance of the inspection time in the fitness function, is the second-order difference penalty term coefficient of the trajectory, which is used to control the smoothness of the trajectory, 、 and are respectively 、 and the pose vectors of the drone at moments, including position and attitude information. The second-order difference term is used to measure the acceleration change of the trajectory.

[0108] Dynamically adjusting the weight coefficient according to the actual curvature change improves the flexibility and adaptability of path planning. By dynamically adjusting the weight coefficient, it is ensured that the path is smoother in areas with large curvature changes, improving the safety and stability of flight. Dynamically adjusting the weight coefficient helps to balance path smoothness and inspection efficiency, improving the performance of the overall system.

[0109] It should be noted that in some embodiments, the expression of the curvature smoothing weight coefficient includes:

[0110] ;

[0111] wherein, is the curvature smoothing weight coefficient, is the average curvature change gradient, representing the average value of the absolute values of the curvature change rates of all flight segments, reflecting the bending degree of the overall path, is the curvature change rate threshold, which is the critical value for triggering weight adjustment. When the average curvature exceeds , increases.

[0112] In some embodiments, it further includes: a meteorological perception unit for real-time monitoring of the wind speed information and precipitation information corresponding to the drone; an autonomous obstacle avoidance unit for detecting the convex structure of the hyperbolic glass curtain wall and temporary obstacles through the fusion of millimeter-wave radar and binocular vision, and controlling the drone to perform autonomous obstacle avoidance when the distance to the convex structure and temporary obstacles is less than a preset distance; a multi-link redundant communication unit carried on the drone to transmit data in parallel with the controller using LoRa and 4G dual channels.

[0113] In the embodiment, the system adds a meteorological perception unit, an autonomous obstacle avoidance unit and a multi-link redundant communication unit to improve the robustness and safety of the system.

[0114] The wind speed sensor real-time monitors the wind speed information at the location of the drone. The precipitation sensor real-time monitors the precipitation information.

[0115] Autonomous obstacle avoidance unit: Millimeter-wave radar: Used for detecting obstacles at a long distance. Binocular vision system: Used for detecting obstacles at a short distance and three-dimensional reconstruction. Obstacle avoidance algorithm: When the distance to a convex structure or a temporary obstacle is less than a preset distance, the UAV is controlled to perform autonomous obstacle avoidance.

[0116] Multi-link redundant communication unit: LoRa communication module: Used for long-distance and low-power communication. 4G communication module: Used for high-speed data transmission. Dual-channel parallel transmission: Ensures the reliability and real-time performance of data transmission.

[0117] By monitoring meteorological conditions in real time, inspections are avoided under adverse weather conditions, reducing risks. Combining the millimeter-wave radar and the binocular vision system enables efficient obstacle detection and avoidance, improving flight safety. Multi-link redundant communication ensures the stability and reliability of data transmission and can still work properly even when a single communication link fails. Real-time monitoring of meteorological conditions and obstacle information helps to adjust the flight plan in a timely manner, improving the response speed and flexibility of the system.

[0118] In some embodiments, the controller constructs a curtain wall performance degradation prediction model based on historical inspection data corresponding to the hyperbolic glass curtain wall. The curtain wall performance degradation prediction model analyzes the defect expansion rate and the correlation of environmental factors corresponding to the hyperbolic glass curtain wall according to the inspection results. The controller outputs an adaptive adjustment suggestion for the inspection cycle according to the defect expansion rate and the correlation of environmental factors.

[0119] The controller constructs a curtain wall performance degradation prediction model based on the historical inspection data of the hyperbolic glass curtain wall, analyzes the defect expansion rate and the correlation of environmental factors according to the inspection results, and outputs an adaptive adjustment suggestion for the inspection cycle.

[0120] Data collection: Historical inspection data: Includes the results, defect types, locations, and severity scores of each inspection. Environmental data: Includes environmental factors such as temperature, humidity, and wind speed.

[0121] Model construction: Based on historical data, construct a curtain wall performance degradation prediction model and analyze the correlation between the defect expansion rate and environmental factors. Use methods such as regression analysis and time series analysis to establish the prediction model.

[0122] Inspection cycle adjustment: According to the inspection results and historical data, analyze the defect expansion rate. Analyze the influence of environmental factors on defect expansion. Output an adaptive adjustment suggestion for the inspection cycle according to the defect expansion rate and the correlation of environmental factors.

[0123] Dynamically adjust the inspection cycle according to the actual defect expansion rate and environmental factors, improving the effectiveness and economy of inspections. By predicting the defect expansion trend in advance, maintenance measures can be taken in a timely manner to extend the service life of the curtain wall. Reasonably arrange the inspection cycle to avoid over-inspection or under-inspection, optimizing resource allocation. Based on the comprehensive analysis of historical data and environmental factors, the scientificity and reliability of the inspection cycle adjustment are improved.

[0124] A method and system for automatic inspection of high-rise hyperbolic glass curtain walls based on drones provided by an embodiment of the present application, which aims to overcome the deficiencies existing in the application of existing drone inspection technologies to hyperbolic glass curtain walls. Specifically, the system realizes a more comprehensive, efficient and safe inspection of hyperbolic glass curtain walls by integrating a variety of sensors and intelligent control algorithms. The following is the detailed technical content of the system:

[0125] System composition:

[0126] Drone: As the main flight platform, it carries a data acquisition module and a controller.

[0127] Data acquisition module: It includes an ultrasonic sensor array, an infrared thermal imager, and an image acquisition module, which are used to collect information about the hyperbolic glass curtain wall from different dimensions.

[0128] Ultrasonic sensor array: It emits high-frequency ultrasonic signals and receives reflected waves, and detects defects inside or on the surface of the glass curtain wall by analyzing the reflected waves.

[0129] Infrared thermal imager: It captures the surface temperature distribution image of the structural sealant of the glass curtain wall to help identify temperature anomaly areas caused by structural problems.

[0130] Image acquisition module: It acquires the surface optical images of the glass curtain wall to facilitate the discovery of appearance damage or other visible problems.

[0131] Controller: It is responsible for processing data from each sensor and generating corresponding inspection results; at the same time, it plans the flight path of the drone according to the specific geometric characteristics of the glass curtain wall and implements an obstacle avoidance strategy.

[0132] The corresponding working process of the system includes:

[0133] Data collection and preliminary processing: Use the ultrasonic sensor array to emit signals to the target and analyze the echo to evaluate the material condition. The infrared thermal imager records the temperature distribution, especially paying attention to those positions that may indicate potential failure points. The image acquisition module takes high-resolution photos for subsequent visual inspection.

[0134] Data analysis: The controller comprehensively analyzes the above three types of data, including but not limited to the intensity of reflected wave signals, temperature gradient changes, and texture features, etc. Based on this information, it determines whether there are any problem areas that require further investigation.

[0135] Path planning and execution: Based on the pre-constructed three-dimensional model and known curvature parameters, the controller calculates the optimal flight route. Combining real-time environmental perception (such as obstacle positions), it dynamically adjusts the trajectory to ensure safe operation. Finally, it guides the drone to complete the entire inspection task according to the optimized path.

[0136] The provided system has at least the following beneficial effects:

[0137] Improve detection accuracy: By combining multiple sensing technologies, it can capture different types of problems more comprehensively, thus improving the ability to detect problems.

[0138] Enhance safety: Introducing an advanced obstacle avoidance mechanism can effectively avoid the risk of the drone colliding with surrounding objects, ensuring the safety of the operation process.

[0139] Improve efficiency: Automated path planning reduces the need for manual intervention, making the entire inspection process smoother and faster, saving a large amount of time costs.

[0140] Strong adaptability: The functions specifically developed for the hyperboloid design enable it to better serve this special building form, showing good flexibility and scope of application.

[0141] Cost-effectiveness: In the long run, adopting this automated solution can significantly reduce maintenance costs and extend the service life of the building, bringing economic benefits to the owner.

[0142] Please refer to Figure 3 , such as Figure 3 shown, the provided method for automatically inspecting high-rise hyperbolic glass curtain walls based on drones is applied to the controller of the system provided in any embodiment of the present application. The method includes steps S101 to step S106. Details are as follows:

[0143] Step S101. Perform time-frequency domain analysis on the reflected waves received by the ultrasonic sensor array to obtain the intensity of the reflected wave signals;

[0144] Step S102. Calculate the temperature gradient of the temperature distribution image of the structural glue surface captured by the infrared thermal imager to obtain the characteristics of the temperature anomaly area;

[0145] Step S103. Obtain the optical image texture according to the surface optical images collected by the image acquisition module;

[0146] Step S104. Generate a drone flight trajectory based on the three-dimensional model of the hyperbolic glass curtain wall and the curvature parameters;

[0147] Step S105. Generate obstacle avoidance position information according to the environmental information set for the hyperbolic glass curtain wall, and update the drone flight trajectory according to the obstacle avoidance position information;

[0148] Step S106. Control the drone to automatically inspect the hyperbolic glass curtain wall according to the drone flight trajectory.

[0149] The specific implementation manners corresponding to Steps S101 - S106 are as follows:

[0150] Step S101. Perform time-frequency domain analysis on the reflected waves received by the ultrasonic sensor array to obtain the reflected wave signal intensity:

[0151] Data acquisition: The ultrasonic sensor array emits high-frequency ultrasonic signals towards the hyperbolic glass curtain wall and receives the reflected waves. Time-frequency domain analysis: Use time-frequency domain analysis techniques (such as Fourier transform) to process the reflected wave signals and extract the frequency and time characteristics of the signals.

[0152] Signal intensity calculation: Calculate the intensity of the reflected wave signals, which reflects the integrity of the internal structure of the glass curtain wall. If there are voids or cracks, the intensity of the reflected wave signals will change.

[0153] Step S102. Calculate the temperature gradient of the surface temperature distribution image captured by the infrared thermal imager to obtain the characteristics of the temperature anomaly area: Data acquisition: The infrared thermal imager captures the surface temperature distribution image of the hyperbolic glass curtain wall. Temperature gradient calculation: Calculate the temperature gradient of the temperature distribution image to identify the temperature anomaly areas. These areas may be caused by structural adhesive aging, cracks or other defects. Feature extraction: Extract the characteristics of the temperature anomaly areas, such as temperature gradient, temperature distribution, etc.

[0154] Step S103. Obtain the optical image texture based on the surface optical image collected by the image acquisition module: Data acquisition: The image acquisition module takes the surface optical image of the hyperbolic glass curtain wall. Image processing: Use image processing techniques (such as edge detection, texture analysis, etc.) to process the image and extract the surface texture characteristics. Feature extraction: Identify and extract the defect characteristics such as cracks and scratches on the surface.

[0155] Step S104. Generate a drone flight trajectory based on the three-dimensional model of the hyperbolic glass curtain wall and the curvature parameters: Three-dimensional model establishment: Establish an accurate three-dimensional model of the hyperbolic glass curtain wall, including its geometric shape and curvature parameters. Path planning: Generate an optimal flight trajectory covering the entire curtain wall according to the three-dimensional model and the curvature parameters. Considering the complex geometric shape of the curtain wall, ensure that the drone can comprehensively cover all areas.

[0156] Step S105. Generate obstacle avoidance position information based on the environmental information set for the hyperbolic glass curtain wall, and update the flight trajectory of the drone according to the obstacle avoidance position information: Environmental information collection: Collect the environmental information around the hyperbolic glass curtain wall, including other buildings, temporary obstacles, etc. Obstacle avoidance position generation: Generate obstacle avoidance position information and mark the areas that need to be avoided. Path update: According to the obstacle avoidance position information, update the flight trajectory of the drone in real time to ensure that the drone can safely avoid obstacles during the inspection process.

[0157] Step S106. Control the drone to automatically inspect the hyperbolic glass curtain wall according to the flight trajectory of the drone: Flight control: The controller sends instructions to the drone according to the generated flight trajectory to control it to fly along the predetermined path. Data collection and transmission: During the flight, the drone continuously collects ultrasonic, infrared thermal imaging, and optical image data and transmits it to the ground station or on-board controller in real time through the wireless communication module. Inspection report generation: After the data is processed, a detailed inspection report is generated, including the defect type, location, and severity score.

[0158] The provided method has the following beneficial effects: Improve detection accuracy: Through the comprehensive analysis of multi-modal data (ultrasonic, infrared thermal imaging, optical images), the accuracy and comprehensiveness of defect detection are improved.

[0159] Enhance safety: By generating obstacle avoidance position information and updating the flight trajectory in real time, the risk of the drone colliding in a complex environment is effectively avoided.

[0160] Improve work efficiency: Automated flight trajectory planning and data processing reduce manual intervention and improve the inspection efficiency.

[0161] Optimize path planning: The flight trajectory generated based on the 3D model and curvature parameters ensures that the drone can cover all areas comprehensively and improves the inspection coverage rate.

[0162] Real-time monitoring and feedback: Real-time data collection and transmission enable problems during the inspection process to be discovered and processed in a timely manner, enhancing the response speed and flexibility of the system.

[0163] Adaptive adjustment: Dynamically adjust the flight trajectory according to the actual environmental information, improving the adaptability and robustness of the system.

[0164] Reduce resource waste: Through reasonable path planning and efficient inspection methods, unnecessary flight time and energy consumption are reduced, and the service life of the drone is extended.

[0165] In summary, the method for automatically inspecting high-rise hyperbolic glass curtain walls based on drones not only solves many problems of traditional manual inspections, but also improves the inspection efficiency and accuracy through advanced technologies and algorithms, providing strong support for the safety maintenance of high-rise buildings.

[0166] In some embodiments, the data acquisition module includes: a polarization filter, which is disposed at the front end of the image acquisition module and is used to suppress the surface reflection interference of the hyperbolic glass curtain wall.

[0167] In some embodiments, the controller generates an ultrasonic signal attenuation coefficient based on the reflected wave signal intensity and the ultrasonic signal intensity; the controller calculates the proportion of the temperature anomaly area based on the temperature anomaly area characteristics and the area corresponding to the hyperbolic glass curtain wall; the controller determines the crack morphology characteristics based on the optical image texture; the controller outputs the inspection result based on the ultrasonic signal attenuation coefficient, the proportion of the temperature anomaly area, and the crack morphology characteristics; the inspection result includes the defect type, the defect location, and the defect severity score.

[0168] Exemplarily, the controller inputs the ultrasonic signal attenuation coefficient, the proportion of the temperature anomaly area, and the crack morphology curvature corresponding to the crack morphology characteristics into a deep learning model based on the attention mechanism, and the deep learning model outputs the inspection result.

[0169] In some embodiments, the controller generates a fitness function based on the curvature change rate of the hyperbolic glass curtain wall, the real-time pose vector corresponding to the drone, the estimated total inspection duration, and the battery endurance threshold; the controller generates the flight trajectory of the drone based on the fitness function using a genetic algorithm.

[0170] Exemplarily, the expression of the fitness function includes:

[0171] ;

[0172] where is the fitness function, is the curvature smoothing weight coefficient, which is used to adjust the importance of the curvature change rate in the fitness function, is the total number of flight segments corresponding to the flight trajectory of the drone, is the curvature change rate of the curtain wall in the flight segment, which represents the change degree of the curvature of a certain section of the curtain wall in the flight path of the drone and is used to measure the bending change of the path, is the estimated total inspection duration, is the time efficiency weight coefficient, which adjusts the importance of the inspection time in the fitness function. is the coefficient of the second-order difference penalty term of the trajectory, which is used to control the smoothness of the trajectory. , and are respectively , and the UAV pose vectors at moments, including position and attitude information. The second-order difference term is used to measure the acceleration change of the trajectory. is the calculation formula of the two-norm.

[0173] It should be noted that in some embodiments, the expression of the curvature smoothing weight coefficient includes:

[0174] ;

[0175] where is the curvature smoothing weight coefficient, is the average curvature change gradient, which represents the average value of the absolute values of the curvature change rates of all flight segments and reflects the bending degree of the overall path. is the curvature change rate threshold, which is the critical value for triggering weight adjustment. When the average curvature exceeds , increases.

[0176] In some embodiments, it further includes: a meteorological sensing unit for real-time monitoring of the wind speed information and precipitation information corresponding to the UAV; an autonomous obstacle avoidance unit for detecting the convex structure of the hyperbolic glass curtain wall and temporary obstacles through the fusion of millimeter-wave radar and binocular vision, and controlling the UAV to perform autonomous obstacle avoidance when the distance to the convex structure and temporary obstacles is less than a preset distance; a multi-link redundant communication unit carried on the UAV to transmit data in parallel with the controller using LoRa and 4G dual channels.

[0177] In some embodiments, the controller constructs a curtain wall performance degradation prediction model based on the historical inspection data corresponding to the hyperbolic glass curtain wall. The curtain wall performance degradation prediction model analyzes the defect expansion rate and environmental factor correlation corresponding to the hyperbolic glass curtain wall according to the inspection results, and the controller outputs an adaptive adjustment suggestion for the inspection period according to the defect expansion rate and environmental factor correlation.

[0178] It should be noted that those skilled in the art can clearly understand that for the convenience and simplicity of description, the above-described method for automatically inspecting high-rise hyperbolic glass curtain walls based on UAVs and the specific working processes of each step can refer to the corresponding processes in the embodiments of the method for automatically inspecting high-rise hyperbolic glass curtain walls based on UAVs described above, and will not be elaborated here.

[0179] The embodiment of the present application provides an automatic inspection device for high-rise hyperbolic glass curtain walls based on drones. The automatic inspection device for high-rise hyperbolic glass curtain walls based on drones is used to execute the steps of the automatic inspection method for high-rise hyperbolic glass curtain walls based on drones shown in the above embodiments. The automatic inspection device for high-rise hyperbolic glass curtain walls based on drones can be a single server or a server cluster, or the automatic inspection device for high-rise hyperbolic glass curtain walls based on drones can be a terminal, and the terminal can be a handheld terminal, a laptop computer, a wearable device or a robot, etc.

[0180] The automatic inspection device for high-rise hyperbolic glass curtain walls based on drones includes:

[0181] A time-frequency analysis module, configured to perform time-frequency domain analysis on the reflected waves received by the ultrasonic sensor array to obtain the reflected wave signal intensity;

[0182] An anomaly acquisition module, configured to calculate the temperature gradient of the surface temperature distribution image captured by the infrared thermal imager to obtain the characteristics of the temperature anomaly area;

[0183] An image acquisition module, configured to obtain the optical image texture according to the surface optical images acquired by the image acquisition module;

[0184] A trajectory generation module, configured to generate a drone flight trajectory according to the three-dimensional model and curvature parameters of the hyperbolic glass curtain wall;

[0185] A trajectory update module, configured to generate obstacle avoidance position information according to the environmental information set for the hyperbolic glass curtain wall, and update the drone flight trajectory according to the obstacle avoidance position information;

[0186] An inspection control module, configured to control the drone to automatically inspect the hyperbolic glass curtain wall according to the drone flight trajectory.

[0187] In some embodiments, the data acquisition module includes: a polarization filter, and the polarization filter is disposed at the front end of the image acquisition module for suppressing the surface reflection interference of the hyperbolic glass curtain wall.

[0188] In some embodiments, the controller generates an ultrasonic signal attenuation coefficient according to the reflected wave signal intensity and the ultrasonic signal intensity; the controller calculates the proportion of the temperature anomaly area according to the characteristics of the temperature anomaly area and the area corresponding to the hyperbolic glass curtain wall; the controller determines the crack morphology characteristics according to the optical image texture; the controller outputs the inspection result according to the ultrasonic signal attenuation coefficient, the proportion of the temperature anomaly area and the crack morphology characteristics; the inspection result includes the defect type, the defect position and the defect severity score.

[0189] Exemplarily, the controller inputs the ultrasonic signal attenuation coefficient, the proportion of the temperature anomaly area, and the crack morphology curvature corresponding to the crack morphology characteristics into a deep learning model based on the attention mechanism, and the deep learning model outputs the inspection result.

[0190] In some embodiments, the controller generates a fitness function according to the curvature change rate of the hyperbolic glass curtain wall, the real-time pose vector corresponding to the drone, the estimated total inspection duration, and the battery endurance threshold; the controller generates the flight trajectory of the drone based on the fitness function and the genetic algorithm.

[0191] Exemplarily, the expression of the fitness function includes:

[0192] ;

[0193] where, is the fitness function, is the curvature smoothing weight coefficient, which is used to adjust the importance of the curvature change rate in the fitness function, is the total number of flight segments corresponding to the flight trajectory of the drone, is the curvature change rate of the curtain wall in the th flight segment, which represents the change degree of the curvature of a certain section of the curtain wall in the flight path of the drone and is used to measure the bending change of the path, is the estimated total inspection duration, is the battery endurance threshold, which represents the maximum allowable time for a single flight of the drone, is the time efficiency weight coefficient, which adjusts the importance of the inspection time in the fitness function, is the trajectory second-order difference penalty term coefficient, which is used to control the smoothness of the trajectory, , and are respectively , and the pose vectors of the drone at, and moments, including position and attitude information, and the second-order difference term is used to measure the acceleration change of the trajectory.

[0194] It should be noted that in some embodiments, the expression of the curvature smoothing weight coefficient includes:

[0195] ;

[0196] where, is the curvature smoothing weight coefficient, is the average curvature change gradient, which represents the average value of the absolute values of the curvature change rates of all flight segments and reflects the bending degree of the overall path, is the curvature change rate threshold, which is the critical value for triggering weight adjustment. When the average curvature exceeds it increases.

[0197] In some embodiments, it further includes: a meteorological sensing unit for real-time monitoring of the wind speed information and precipitation information corresponding to the drone; an autonomous obstacle avoidance unit for detecting the convex structure of the hyperbolic glass curtain wall and temporary obstacles through the fusion of millimeter-wave radar and binocular vision, and controlling the drone to perform autonomous obstacle avoidance when the distance to the convex structure and temporary obstacles is less than a preset distance; a multi-link redundant communication unit carried on the drone to transmit data in parallel with the controller using LoRa and 4G dual channels.

[0198] In some embodiments, the controller constructs a curtain wall performance degradation prediction model based on the historical inspection data corresponding to the hyperbolic glass curtain wall. The curtain wall performance degradation prediction model analyzes the defect expansion rate and the correlation of environmental factors corresponding to the hyperbolic glass curtain wall according to the inspection results, and the controller outputs an adaptive adjustment suggestion for the inspection period according to the defect expansion rate and the correlation of environmental factors.

[0199] It should be noted that those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working processes of the above-described high-rise hyperbolic glass curtain wall automatic inspection device based on drones and each module can refer to the corresponding processes in the embodiments of the high-rise hyperbolic glass curtain wall automatic inspection method based on drones described above, and will not be elaborated here.

[0200] The above high-rise hyperbolic glass curtain wall automatic inspection method based on drones can be implemented in the form of a computer program, and the computer program can run on the provided device.

[0201] Please refer to Figure 4 , Figure 4 which is a schematic block diagram of the structure of the controller provided by the embodiment of the present application. The controller includes a processor, a memory, and a network interface connected through a device bus. Among them, the memory can include a storage medium and an internal memory.

[0202] The storage medium can store an operating device and a computer program. The computer program includes program instructions, and when the program instructions are executed, the processor can execute any high-rise hyperbolic glass curtain wall automatic inspection method based on drones.

[0203] The processor is used to provide computing and control capabilities to support the operation of the entire controller.

[0204] The internal memory provides an environment for the operation of a computer program in a non-volatile storage medium. When the computer program is executed by a processor, the processor can be made to execute any one of the high-level automatic inspection methods for hyperbolic glass curtain walls based on unmanned aerial vehicles.

[0205] This network interface is used for network communication, such as sending assigned tasks, etc. Those skilled in the art can understand that Figure 4 The structure shown in is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the terminal to which the solution of this application is applied. The specific controller may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0206] It should be understood that the processor may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Among them, the general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0207] Among them, in one embodiment, the processor is used to run a computer program stored in the memory to implement the following steps:

[0208] Perform time-frequency domain analysis on the reflected waves received by the ultrasonic sensor array to obtain the reflected wave signal intensity;

[0209] Calculate the temperature gradient of the surface temperature distribution image captured by the infrared thermal imager to obtain the characteristics of the temperature anomaly area;

[0210] Obtain the optical image texture according to the surface optical image collected by the image acquisition module;

[0211] Generate a UAV flight trajectory according to the three-dimensional model and curvature parameters of the hyperbolic glass curtain wall;

[0212] Generate obstacle avoidance position information according to the environmental information set by the hyperbolic glass curtain wall, and update the UAV flight trajectory according to the obstacle avoidance position information;

[0213] Control the UAV to automatically inspect the hyperbolic glass curtain wall according to the UAV flight trajectory.

[0214] In some embodiments, before inputting the visible light image and the thermal imaging image into the damage feature recognition model based on a convolutional neural network, the following steps are further included: preprocessing the information collected by the sensor, the visible light image, and the thermal imaging image; the preprocessing includes at least filtering, denoising, enhancement, and cropping preprocessing; annotating damage information in the preprocessed information collected by the sensor, the visible light image, and the thermal imaging image; the damage information includes at least the damage type and the timing anomaly situation.

[0215] In some embodiments, the data acquisition module includes: a polarization filter, which is arranged at the front end of the image acquisition module and is used to suppress the surface reflection interference of the hyperbolic glass curtain wall.

[0216] In some embodiments, the controller generates an ultrasonic signal attenuation coefficient according to the reflected wave signal intensity and the ultrasonic signal intensity; the controller calculates the proportion of the temperature anomaly area according to the temperature anomaly area feature and the area corresponding to the hyperbolic glass curtain wall; the controller determines the crack morphology feature according to the optical image texture; the controller outputs the inspection result according to the ultrasonic signal attenuation coefficient, the proportion of the temperature anomaly area, and the crack morphology feature; the inspection result includes the defect type, the defect location, and the defect severity score.

[0217] Exemplarily, the controller inputs the ultrasonic signal attenuation coefficient, the proportion of the temperature anomaly area, and the crack morphology curvature corresponding to the crack morphology feature into a deep learning model based on an attention mechanism, and the deep learning model outputs the inspection result.

[0218] In some embodiments, the controller generates a fitness function according to the curvature change rate of the hyperbolic glass curtain wall, the real-time pose vector corresponding to the drone, the estimated total inspection duration, and the battery endurance threshold; the controller generates the flight trajectory of the drone based on the genetic algorithm according to the fitness function.

[0219] Exemplarily, the expression of the fitness function includes:

[0220] ;

[0221] where is the fitness function, is the curvature smoothing weight coefficient, which is used to adjust the importance of the curvature change rate in the fitness function, is the total number of flight segments corresponding to the flight trajectory of the drone, is the curvature change rate of the curtain wall in the th flight segment, which represents the change degree of the curvature of a certain section of the curtain wall in the flight path of the drone and is used to measure the bending change of the path, is the estimated total inspection duration, is the battery life threshold, representing the maximum allowable time for a single flight of the drone, is the time efficiency weight coefficient, which adjusts the importance of the inspection time in the fitness function, is the second-order difference penalty term coefficient of the trajectory, which is used to control the smoothness of the trajectory, 、 and are respectively 、 and the pose vectors of the drone at moments, including position and attitude information, and the second-order difference term is used to measure the acceleration change of the trajectory.

[0222] It should be noted that, in some embodiments, the expression of the curvature smoothing weight coefficient includes:

[0223] ;

[0224] wherein, is the curvature smoothing weight coefficient, is the average curvature change gradient, representing the average value of the absolute values of the curvature change rates of all flight segments, reflecting the bending degree of the overall path, is the curvature change rate threshold, which is the critical value for triggering weight adjustment. When the average curvature exceeds , increases.

[0225] In some embodiments, it further includes: a meteorological perception unit for real-time monitoring of the wind speed information and precipitation information corresponding to the drone; an autonomous obstacle avoidance unit for detecting the convex structure of the hyperbolic glass curtain wall and temporary obstacles through the fusion of millimeter-wave radar and binocular vision, and controlling the drone to perform autonomous obstacle avoidance when the distance to the convex structure and temporary obstacles is less than a preset distance; a multi-link redundant communication unit carried on the drone to transmit data in parallel with the controller using LoRa and 4G dual channels.

[0226] In some embodiments, the controller constructs a curtain wall performance degradation prediction model based on the historical inspection data corresponding to the hyperbolic glass curtain wall. The curtain wall performance degradation prediction model analyzes the defect expansion rate and the correlation of environmental factors corresponding to the hyperbolic glass curtain wall according to the inspection results, and the controller outputs an adaptive adjustment suggestion for the inspection cycle according to the defect expansion rate and the correlation of environmental factors.

[0227] It should be noted that those skilled in the art can clearly understand that, for the convenience and simplicity of description, the specific working processes of the above-described controller and each module can refer to the corresponding processes in the method embodiments described in the above-mentioned embodiments, and will not be elaborated herein.

[0228] The present application also provides a computer-readable storage medium storing a computer program, which when executed by a processor causes the processor to implement the steps of the automatic inspection method for high-rise hyperbolic glass curtain walls based on unmanned aerial vehicles provided in any embodiment of the present application.

[0229] Among them, the computer-readable storage medium may be an internal storage unit of the controller described in the foregoing embodiments, such as the hard disk or memory of the controller. The computer-readable storage medium may also be an external storage device of the controller, such as a plug-in hard disk equipped on the controller, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc.

[0230] It should be noted that those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described storage medium and each module can refer to the corresponding processes in the method embodiments described in the above embodiments, and will not be repeated here.

[0231] The above is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of various equivalent modifications or substitutions within the technical scope disclosed by the present application, and these modifications or substitutions should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. An automatic inspection system for high-rise hyperbolic glass curtain walls based on drones, characterized in that, Including: An unmanned aerial vehicle (UAV); A data acquisition module mounted on the UAV. The data acquisition module includes an ultrasonic sensor array, an infrared thermal imager, and an image acquisition module. The ultrasonic sensor array is used to transmit high-frequency ultrasonic signals to the hyperbolic glass curtain wall to be inspected and receive the reflected waves. The infrared thermal imager is used to capture the surface temperature distribution image of the structural adhesive of the hyperbolic glass curtain wall. The image acquisition module is used to collect the surface optical image of the hyperbolic glass curtain wall; A controller that is used to perform time-frequency domain analysis on the reflected waves to obtain the reflected wave signal intensity; the controller calculates the temperature gradient of the surface temperature distribution image of the structural adhesive to obtain the characteristics of the temperature anomaly area; the controller obtains the optical image texture according to the surface optical image; the controller outputs the inspection result according to the reflected wave signal intensity, the characteristics of the temperature anomaly area, and the optical image texture; the controller generates the UAV flight trajectory according to the three-dimensional model and curvature parameters of the hyperbolic glass curtain wall, generates the obstacle avoidance position information according to the environmental information set for the hyperbolic glass curtain wall, and updates the UAV flight trajectory according to the obstacle avoidance position information; the controller controls the UAV to perform automatic inspection on the hyperbolic glass curtain wall according to the UAV flight trajectory; the controller generates a fitness function according to the curvature change rate of the hyperbolic glass curtain wall, the corresponding real-time pose vector of the UAV, the estimated total inspection duration, and the battery endurance threshold; The controller generates the UAV flight trajectory based on the genetic algorithm according to the fitness function; The expression of the fitness function includes: ； Among them, is the fitness function, is the curvature smoothing weight coefficient, which is used to adjust the importance of the curvature change rate in the fitness function, is the total number of flight segments corresponding to the UAV flight trajectory, is the curvature change rate of the curtain wall in the flight segment, which represents the degree of change in the curvature of a certain section of the curtain wall in the flight path of the UAV and is used to measure the bending change of the path, is the estimated total inspection duration, is the battery life threshold, which represents the maximum allowable time for a single flight of the UAV, is the time efficiency weight coefficient, which adjusts the importance of the inspection time in the fitness function, , and are respectively , and UAV pose vectors at moments, including position and attitude information, and the second-order difference term is used to measure the acceleration change of the trajectory.

2. The system according to claim 1, wherein The data acquisition module includes: A polarization filter disposed at the front end of the image acquisition module for suppressing the surface reflection interference of the hyperbolic glass curtain wall.

3. The system according to claim 1, wherein The controller generates an ultrasonic signal attenuation coefficient according to the reflected wave signal intensity and the ultrasonic signal intensity; The controller calculates the proportion of the temperature anomaly area according to the characteristics of the temperature anomaly area and the area corresponding to the hyperbolic glass curtain wall; The controller determines the crack morphology characteristics according to the optical image texture; The controller outputs the inspection result according to the ultrasonic signal attenuation coefficient, the proportion of the temperature anomaly area, and the crack morphology characteristics; the inspection result includes the defect type, defect location, and defect severity score.

4. The system according to claim 3, characterized in that, The controller inputs the ultrasonic signal attenuation coefficient, the proportion of the temperature anomaly area, and the crack morphology curvature corresponding to the crack morphology characteristics into a deep learning model based on the attention mechanism, and the deep learning model outputs the inspection result.

5. The system according to claim 1, wherein The expression of the curvature smoothing weight coefficient includes: ； Among them, is the curvature smoothing weight coefficient, is the average curvature change gradient, representing the average value of the absolute values of the curvature change rates of all flight segments, reflecting the bending degree of the overall path, is the curvature change rate threshold, the critical value used to trigger weight adjustment. When the average curvature exceeds , increases.

6. The system according to claim 1, wherein It also includes: A meteorological perception unit for real-time monitoring of the wind speed information and precipitation information corresponding to the UAV; An autonomous obstacle avoidance unit that detects the convex structure and temporary obstacles of the hyperbolic glass curtain wall through the fusion of millimeter-wave radar and binocular vision, and controls the UAV to perform autonomous obstacle avoidance when the distance to the convex structure and temporary obstacles is less than the preset distance. The multi-link redundant communication unit is mounted on the UAV and transmits data in parallel with the controller through LoRa and 4G dual channels.

7. The system according to claim 1, wherein The controller constructs a curtain wall performance degradation prediction model based on the historical inspection data corresponding to the hyperbolic glass curtain wall. The curtain wall performance degradation prediction model analyzes the defect expansion rate and the correlation of environmental factors corresponding to the hyperbolic glass curtain wall according to the inspection results, and the controller outputs an adaptive adjustment suggestion for the inspection cycle according to the defect expansion rate and the correlation of environmental factors.

8. An automatic inspection method for high-rise hyperbolic glass curtain walls based on drones, characterized in that, A controller applied to the system according to any one of claims 1-7, the method comprising: Performing time-frequency domain analysis on the reflected waves received by the ultrasonic sensor array to obtain the reflected wave signal intensity; Calculating the temperature gradient of the surface temperature distribution image captured by the infrared thermal imager to obtain the characteristics of the temperature anomaly area; Obtaining the optical image texture according to the surface optical image collected by the image acquisition module; Generating a UAV flight trajectory according to the three-dimensional model and curvature parameters of the hyperbolic glass curtain wall; Generating obstacle avoidance position information according to the environmental information set by the hyperbolic glass curtain wall, and updating the UAV flight trajectory according to the obstacle avoidance position information; Controlling the UAV to automatically inspect the hyperbolic glass curtain wall according to the UAV flight trajectory.

Images