Method and system for detecting hollowing of thermal insulation layer of high-rise building based on unmanned aerial vehicle

Abstract

The invention relates to a high-rise building heat insulation layer hollowing detection method and system based on an unmanned aerial vehicle. The method comprises the following steps: converting point cloud data of a building into a three-dimensional network model by using a Poisson reconstruction algorithm, and generating a spiral scanning unmanned aerial vehicle route; the unmanned aerial vehicle flies according to the route of the unmanned aerial vehicle, collects flight attitude data and reflection echo signals of the laser radar, and suppresses flight jitter interference through an anti-jitter signal processing algorithm; performing time-frequency domain analysis on the radar signals after noise reduction, extracting double-interface reflection coefficient features, inputting the double-interface reflection coefficient features into a support vector machine classifier to identify a hollowing region, and calculating hollowing position coordinates; and generating a three-dimensional thermal image of the building, and outputting a hollowing detection report. Anti-interference processing is carried out through an anti-jitter signal processing algorithm, so that hollowing feature misjudgment caused by jitter can be avoided; the whole process is automatic, and labor time consumption can be saved; automatic, precise and quantitative evaluation of hollowing detection is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of building detection, and in particular to a method and system for detecting hollowing of thermal insulation layers of high-rise buildings based on an unmanned aerial vehicle (UAV). Background Art

[0002] Hollowing of the insulation layer on the exterior walls of high-rise buildings is a common safety hazard, potentially leading to reduced insulation performance, water seepage, and even insulation loss. Traditional manual inspection methods rely on high-altitude work, using methods like tapping and infrared temperature measurement. These methods suffer from low efficiency (a single person can inspect less than 200 square meters per day), high risk, and limited inspection depth, making them inadequate for the large-scale inspection needs of modern high-rise buildings.

[0003] Existing drone inspection solutions mostly use optical cameras or lidar. Optical inspection is significantly affected by lighting conditions, with detection performance significantly reduced in rainy, overcast, and backlit environments. It also cannot penetrate decorative layers to identify internal hollows. While lidar can acquire 3D point cloud data, it lacks the ability to distinguish reflected signals from millimeter-level hollow interfaces, resulting in a missed detection rate exceeding 40%. Millimeter-wave radar, with its strong penetration (capable of penetrating 5-10cm thick decorative layers) and excellent interference immunity (unaffected by light and dust), has become a promising technology for internal building defect detection. However, its application to drone platforms faces multiple technical bottlenecks. First, drone flight jitter (frequency 5-20Hz) can cause radar beam pointing deviations, resulting in reflected signal errors exceeding ±15%, leading to misidentification of hollow features. Second, traditional GPS positioning accuracy is only at the meter level, which cannot meet the precise hollow location requirements (≤0.5m accuracy is required to match maintenance unit divisions). Third, existing solutions lack automated algorithms for the entire process, from building modeling and route planning to hollow location. Relying on manually annotated data, this limits inspection efficiency.

[0004] Therefore, the traditional method of detecting hollowing in the insulation layer of the exterior wall of high-rise buildings often has the problem of low detection accuracy because the flight vibration of the drone will interfere with the radar signal, resulting in low spatial positioning accuracy. Summary of the Invention

[0005] Based on this, in order to solve the above technical problems, a method and system for detecting hollowing in the insulation layer of high-rise buildings based on drones are provided, which can realize the automation, precision and quantitative evaluation of hollowing detection.

[0006] A method for detecting hollowing of thermal insulation layers of high-rise buildings based on an unmanned aerial vehicle (UAV), the method comprising:

[0007] Obtaining exterior wall point cloud data collected by a drone equipped with a laser radar flying around the building, and converting the point cloud data into a three-dimensional network model using a Poisson reconstruction algorithm;

[0008] Generate a spiral scanning drone route based on the three-dimensional network model, and set a laser radar scanning angle, flight data, and scanning overlap rate according to the drone route;

[0009] The UAV is equipped with a laser radar and flies along the UAV route to collect flight attitude data and a reflected echo signal of the laser radar. An anti-jitter signal processing algorithm is used to suppress flight jitter interference based on the flight attitude data and the reflected echo signal to obtain a noise-reduced radar signal.

[0010] Performing time-frequency domain analysis on the noise-reduced radar signal, extracting dual-interface reflection coefficient features and inputting them into a support vector machine classifier to identify hollow areas and calculate hollow position coordinates;

[0011] The hollowing position coordinates are mapped to the three-dimensional network model to generate a three-dimensional thermal imaging image of the building, and a hollowing detection report is output based on the three-dimensional thermal imaging image of the building.

[0012] In one embodiment, obtaining exterior wall point cloud data collected by flying around a building using a drone equipped with a laser radar includes:

[0013] Determine a data collection orbital path, and use a drone equipped with a laser radar to fly along the data collection orbital path;

[0014] A point cloud density reference value is determined, and the exterior wall of the building is scanned by the drone equipped with a laser radar to collect raw data greater than the point cloud density reference value as exterior wall point cloud data.

[0015] In one embodiment, a spiral scanning drone route is generated based on the three-dimensional network model, and a laser radar scanning angle, flight data, and scanning overlap rate are set according to the drone route, including:

[0016] Based on the three-dimensional network model, a spiral scanning drone route is generated by climbing from the bottom to the top of the building floor by floor;

[0017] Setting a scanning angle of the laser radar according to the drone route, wherein the scanning angle is such that the radar wave is incident vertically on the surface of the insulation layer of the high-rise building;

[0018] Setting flight data according to the UAV route, wherein the flight data includes flight altitude, flight speed, and flight heading angle;

[0019] The scanning overlap rate is set according to the UAV route, and the scanning overlap rate includes a horizontal overlap rate and a vertical overlap rate.

[0020] In one embodiment, the flight attitude data includes pitch angle, roll angle, yaw angle, and acceleration; the reflected echo signal of the laser radar includes timestamp, distance, and phase.

[0021] In one embodiment, suppressing flight jitter interference based on the flight attitude data and the reflected echo signal by an anti-jitter signal processing algorithm to obtain a noise-reduced radar signal includes:

[0022] collecting a flight jitter frequency based on the flight attitude data;

[0023] The reflected echo signal is filtered in the time and space domain by an anti-jitter signal processing algorithm based on Kalman filtering, and the UAV flight jitter interference of the target frequency is suppressed according to the flight jitter frequency to obtain a noise-reduced radar signal.

[0024] In one embodiment, the radar signal after noise reduction is subjected to time-frequency domain analysis, and dual-interface reflection coefficient features are extracted and input into a support vector machine classifier to identify hollow areas, including:

[0025] Performing a fast Fourier transform on the noise-reduced radar signal and extracting a dual-interface frequency component;

[0026] Calculating a double-interface reflection coefficient difference according to the double-interface frequency components as a double-interface reflection coefficient feature;

[0027] The dual-interface reflection coefficient feature is input into a support vector machine classifier, and the dual-interface reflection coefficient difference is used as a standard threshold to identify the hollow area.

[0028] In one embodiment, calculating the coordinates of the hollow drum position includes:

[0029] Obtain the UWB coordinates of the drone at the time of detection, radar beam azimuth and pitch angle, and target distance;

[0030] Based on the UWB coordinates of the drone, the azimuth and pitch angles of the radar beam, and the target distance, the three-dimensional coordinates of the hollow drum are calculated through spatial geometric transformation;

[0031] The three-dimensional coordinates of the hollow drum are corrected based on the three-dimensional network model to obtain the coordinates of the hollow drum position.

[0032] In one embodiment, mapping the hollowing position coordinates to the three-dimensional network model to generate a three-dimensional thermal image of the building includes:

[0033] Mapping the hollow position coordinates to the three-dimensional network model, and generating a three-dimensional thermal image of the building through the Paraview visualization engine;

[0034] The three-dimensional thermal imaging image of the building displays different colors according to the difference in the dual-interface reflection coefficient.

[0035] In one embodiment, the hollowing detection report includes: a building hollowing distribution map, hollowing data, and hollowing detection efficiency data.

[0036] A high-rise building insulation hollowing detection system based on a drone, the system comprising:

[0037] A 3D network model creation module is used to obtain exterior wall point cloud data collected by a drone equipped with a laser radar flying around the building, and convert the point cloud data into a 3D network model using a Poisson reconstruction algorithm;

[0038] a route generation module, configured to generate a spiral scanning UAV route based on the three-dimensional network model, and set a scanning angle, flight data, and scanning overlap rate of a laser radar according to the UAV route;

[0039] A jitter interference suppression module is used to enable the UAV equipped with a laser radar to fly along the UAV route, collect flight attitude data and the reflected echo signal of the laser radar, and suppress flight jitter interference based on the flight attitude data and the reflected echo signal through an anti-jitter signal processing algorithm to obtain a noise-reduced radar signal;

[0040] A hollowing recognition module is used to perform time-frequency domain analysis on the noise-reduced radar signal, extract the dual-interface reflection coefficient feature and input it into a support vector machine classifier to identify the hollowing area and calculate the hollowing position coordinates;

[0041] The detection result display module is used to map the hollowing position coordinates into the three-dimensional network model, generate a three-dimensional thermal imaging image of the building, and output a hollowing detection report based on the three-dimensional thermal imaging image of the building.

[0042] The above-mentioned drone-based method and system for detecting hollowing in the thermal insulation layer of high-rise buildings converts the collected point cloud data into a three-dimensional network model, thereby generating a spiral scanning drone route, which can comprehensively collect high-rise building data and avoid the waste of resources for large-scale inspections and small-scale repairs; the anti-interference processing of radar signals through anti-jitter signal processing algorithms can avoid the misjudgment of hollowing features caused by jitter when the radar is used on a mobile platform; and the laser radar completes the point cloud collection and three-dimensional modeling of the entire building, generates a spiral scanning route, and the drone flies according to the planned route. The entire process is automated, saving manual time; and the automation, precision and quantitative evaluation of hollowing detection are realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 A diagram illustrating an application environment of a method for detecting hollowing in thermal insulation layers of high-rise buildings based on a drone in one embodiment;

[0044] Figure 2 1 is a flow chart of a method for detecting hollowing in the thermal insulation layer of a high-rise building based on a drone in one embodiment;

[0045] Figure 3 1 is a block diagram of a high-rise building insulation hollowing detection system based on a drone in one embodiment;

[0046] Figure 4 FIG. 1 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solutions and advantages of this application more clear, the following further describes this application in detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not intended to limit this application.

[0048] The hollowing detection method for high-rise building insulation layer based on drone provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Figure 1 As shown, the application environment includes a computer device 110 and a drone 120 equipped with a laser radar. The computer device 110 can communicate with the drone 120 over a network. The computer device 110 can obtain exterior wall point cloud data collected by a drone 120 equipped with a laser radar during a circumnavigation of the building, and convert the point cloud data into a three-dimensional network model using a Poisson reconstruction algorithm. The computer device 110 can generate a spiral scanning drone route based on the three-dimensional network model and set the laser radar's scanning angle, flight data, and scanning overlap rate according to the drone route. The drone 120 equipped with the laser radar flies along the drone route, collecting flight attitude data and the laser radar's reflected echo signal. The computer device 110 can suppress flight jitter interference based on the flight attitude data and the reflected echo signal using an anti-jitter signal processing algorithm to obtain a de-noised radar signal. The computer device 110 can perform time-frequency domain analysis on the de-noised radar signal, extracting dual-interface reflection coefficient features, which are input into a support vector machine classifier to identify hollow areas and calculate hollow location coordinates. The computer device 110 can map the hollow location coordinates to the three-dimensional network model to generate a three-dimensional thermal image of the building, and output a hollow detection report based on the three-dimensional thermal image of the building. The computer device 110 can be, but is not limited to, various personal computers, laptops, smartphones, robots, tablet computers, and other devices.

[0049] In one embodiment, Figure 2 As shown, a method for detecting hollowing of thermal insulation layers of high-rise buildings based on a drone is provided, comprising the following steps:

[0050] Step 202: Obtain the exterior wall point cloud data collected by a drone equipped with a laser radar flying around the building, and use the Poisson reconstruction algorithm to convert the point cloud data into a three-dimensional network model.

[0051] The computer device can control the drone equipped with the LiDAR to fly around the building. At this time, the LiDAR simultaneously collects point cloud data, thereby obtaining the collected exterior wall point cloud data.

[0052] Among them, the laser radar is a millimeter wave radar in the 77GHz frequency band, which transmits frequency modulated continuous wave (FMCW) at a frequency of 200Hz.

[0053] In one embodiment, a method for detecting hollowing in the insulation layer of a high-rise building based on a drone is provided, which may also include a process for collecting point cloud data. The specific process includes: determining a data collection orbiting path, and using a drone equipped with a laser radar to fly along the data collection orbiting path; determining a point cloud density reference value, and scanning the exterior wall of the building using a drone equipped with a laser radar, and collecting raw data greater than the point cloud density reference value as exterior wall point cloud data.

[0054] The data collection surrounding path may be a path for a drone equipped with a laser radar to collect point cloud data. The data collection surrounding path may be manually pre-set, and the drone may fly according to the data collection surrounding path.

[0055] Specifically, in this embodiment, a drone equipped with a laser radar (LiDAR) can be used to fly in a circular path to perform a 360° scan of the building's exterior wall, collecting point cloud data with a point cloud density of ≥50 points / square meter, covering the entire area of ​​the building's facade, including complex structures such as windows, balconies, and decorative lines.

[0056] After collecting point cloud data, a Poisson reconstruction algorithm can be used to generate a 3D mesh model with geographic coordinates, annotating the coordinates (X, Y, Z) of feature points such as each floor and window opening. Specifically, a computer device can use the Poisson reconstruction algorithm to convert the point cloud data into a 3D mesh model with geographic coordinates (WGS84 coordinate system). The model vertex coordinates are accurate to the centimeter level (XY plane error ≤ 5cm, Z axis error ≤ 8cm). The coordinates (X, Y, Z) of more than 200 feature points, such as floor boundaries and window opening centers, are manually annotated to form a detection reference coordinate system.

[0057] Step 204 , generating a spiral scanning drone route based on the three-dimensional network model, and setting the laser radar scanning angle, flight data, and scanning overlap rate according to the drone route.

[0058] Computer equipment can automatically generate a spiral scanning drone route in the inspection software based on the building's three-dimensional grid model, set the radar scanning angle to 45°-60°, ensure that the radar beam is vertically incident on the surface of the insulation layer, and the overlap rate of adjacent scanning strips is ≥30%.

[0059] In one embodiment, a method for detecting hollowing in the insulation layer of a high-rise building based on a drone is provided, and further, the process may include generating a spiral scanning drone route. The specific process includes: based on a three-dimensional network model, climbing floor by floor from the bottom to the top of the building, generating a spiral scanning drone route; setting a scanning angle of a laser radar according to the drone route, wherein the scanning angle is that the radar wave is vertically incident on the surface of the insulation layer of the high-rise building; setting flight data according to the drone route, wherein the flight data includes flight altitude, flight speed, and flight heading angle; and setting a scanning overlap rate according to the drone route, wherein the scanning overlap rate includes a horizontal overlap rate and a vertical overlap rate.

[0060] Based on the building's three-dimensional grid model, the computer equipment can automatically generate a spiral layered scanning route through detection software, climbing from the bottom of the building (2 meters from the ground) to the top floor by floor, with a single-circle scanning height difference of 3 meters, ensuring that there are no detection blind spots. In this embodiment, the beam emission angle can be set to 45°-60° relative to the vertical direction of the drone fuselage. Through geometric calculations, the radar wave is ensured to be perpendicular to the surface of the insulation layer, and the reflected signal intensity is increased by 20%-30%. Next, the horizontal overlap rate of adjacent routes is ≥30%, and the vertical overlap rate is ≥25%, to avoid missed detection due to drone flight trajectory errors. The flight altitude is maintained between 8-10 meters from the building facade, the speed is 1.5m / s±0.2m / s, and the heading angle error is ≤3°, ensuring that the radar beam incidence angle error is ≤5°.

[0061] In step 206, the drone equipped with the lidar flies according to the drone route, collects flight attitude data and the reflected echo signal of the lidar, and suppresses flight jitter interference based on the flight attitude data and the reflected echo signal through an anti-jitter signal processing algorithm to obtain a noise-reduced radar signal.

[0062] Among them, the flight attitude data includes pitch angle, roll angle, yaw angle, and acceleration; the reflected echo signal of the lidar includes timestamp, distance, and phase.

[0063] In one embodiment, a method for detecting hollowing in the insulation layer of a high-rise building based on a drone is provided, which may also include an anti-interference process. The specific process includes: collecting flight jitter frequency based on flight attitude data; performing spatiotemporal filtering on the reflected echo signal through an anti-jitter signal processing algorithm based on Kalman filtering, and suppressing the drone flight jitter interference of the target frequency according to the flight jitter frequency to obtain a noise-reduced radar signal.

[0064] The drone can fly according to the planned route and synchronously collect the flight attitude data of the inertial navigation system and the reflected echo signal of the millimeter wave radar. The reflected echo signal contains the timestamp (t), distance (d), phase The built-in anti-jitter signal processing algorithm filters the original signal in the time and space domain, suppressing the drone flight jitter interference with a frequency of 5-20Hz, and reducing the signal error from ±15% to within ±3%.

[0065] The drone is equipped with a customized 77GHz millimeter wave radar module that supports FMCW modulation mode and a transmission power of ≤10mW, which complies with electromagnetic radiation safety standards. The drone is integrated with a dual redundant inertial navigation system (INS) that collects real-time flight attitude data at a frequency of 100Hz, including three-axis acceleration (accuracy ±0.01m / s 2 ), three-axis angular velocity (accuracy ±0.1° / s), and Euler angles (pitch / roll / yaw, accuracy ±0.5°). The millimeter-wave radar module measures 150mm×100mm×80mm and weighs 1.2kg.

[0066] In this embodiment, a three-level anti-interference architecture can be used for anti-shake. In terms of hardware, the radar module uses a three-axis stabilized gimbal (load capacity 2kg, stability accuracy ±0.1°) to isolate the drone from vibration.

[0067] Specifically, the computer device can use the adaptive noise elimination algorithm based on the Kalman filter to perform the original radar echo signal (including the timestamp t, distance d, phase Parameters such as , etc.) perform joint filtering in the spatiotemporal domain, and the algorithm expression can be written as: Among them, S(t,d) is the original signal, S ′ (t, d) is the signal after noise reduction, f is the jitter interference model trained by historical data (including 5-20Hz vibration frequency components), w i Through iterative optimization of the adaptive gradient descent algorithm (learning rate 0.01), the signal error was finally suppressed from ±15% to within ±3%.

[0068] Step 208 , performing time-frequency domain analysis on the radar signal after noise reduction, extracting the dual-interface reflection coefficient feature and inputting it into a support vector machine classifier to identify the hollow area and calculate the hollow position coordinates.

[0069] In one embodiment, a provided drone-based method for detecting hollowing in the insulation layer of a high-rise building may also include a process for identifying hollowing areas. The specific process includes: performing a fast Fourier transform on the noise-reduced radar signal and extracting the dual-interface frequency component; calculating the dual-interface reflection coefficient difference based on the dual-interface frequency component as a dual-interface reflection coefficient feature; inputting the dual-interface reflection coefficient feature into a support vector machine classifier, and using the dual-interface reflection coefficient difference as a standard threshold to identify the hollowing area.

[0070] Computer equipment can perform time-frequency domain transformation on the radar echo after noise reduction, extract the dual-interface reflection coefficient characteristics, and identify the hollow area through the support vector machine (SVM) classifier. The judgment threshold is the reflection coefficient difference > 0.3dB and the area > 0.1m 2 In this embodiment, the training data set of the support vector machine (SVM) classifier contains radar echo features of a normal insulation layer and hollowing areas of varying degrees, and is generated by annotating measured samples.

[0071] Specifically, in this embodiment, the computer device can perform a fast Fourier transform (FFT) on the noise-reduced radar signal, converting the time-domain signal into frequency-domain features and extracting the double-peak frequency components of the insulation layer-air interface and the air-base layer interface. The computer device can then calculate the difference in reflection coefficients between the two interfaces (ΔΓ = Γ2 - Γ1), where Γ1 is the reflection coefficient of the insulation layer-air interface and Γ2 is the reflection coefficient of the air-base layer interface. A larger difference indicates a larger hollow gap.

[0072] When identifying hollow areas, a radial basis kernel function (RBF) support vector machine (SVM) classifier can be used. Based on a data set containing more than 100,000 samples, the judgment threshold is ΔΓ>0.3dB and the number of consecutive sampling points is ≥50 (corresponding to an area of ​​>0.1m 2 ); Among them, normal samples account for 60%, mild hollowing accounts for 25%, and moderate to severe hollowing accounts for 15%.

[0073] In one embodiment, a method for detecting hollowing in the insulation layer of a high-rise building based on a drone is provided, which may also include a process of calculating the coordinates of the hollowing position. The specific process includes: obtaining the UWB coordinates of the drone, the azimuth and pitch angles of the radar beam, and the target distance at the time of detection; calculating the three-dimensional coordinates of the hollowing through spatial geometric transformation based on the UWB coordinates of the drone, the azimuth and pitch angles of the radar beam, and the target distance; and correcting the three-dimensional coordinates of the hollowing based on a three-dimensional network model to obtain the coordinates of the hollowing position.

[0074] Among them, the UWB coordinates of the drone are collected through the UWB positioning network. The UWB positioning network deploys ≥3 base stations within 20-50 meters around the building, and forms a real-time positioning system with the tags carried by the drone. The ranging accuracy is ≤0.1 meter and the refresh rate is ≥50Hz, realizing sub-meter relative positioning between the drone and the building with an error of ≤0.8 meter.

[0075] The computer device can calculate the hollowing position coordinates through spatial geometric transformation based on the UWB positioning coordinates (X0, Y0, Z0) of the drone at the time of detection, the radar beam pointing angle, and the target distance d. Among them, the radar beam pointing angle includes the radar beam azimuth angle θ in the horizontal direction and the pitch angle in the vertical direction. The calculation formula of the hollow drum position coordinates can be expressed as: X = X0 + d·sinθ·cosφ; Y = Y0 + d·sinθ·sinφ; Z = Z0 + d·cosθ.

[0076] Then, the computer equipment can be combined with the building's three-dimensional grid model to perform coordinate correction, with the final positioning accuracy ≤0.2m.

[0077] In step 210 , the hollowing position coordinates are mapped into a three-dimensional network model to generate a three-dimensional thermal image of the building, and a hollowing detection report is output based on the three-dimensional thermal image of the building.

[0078] In one embodiment, the computer device can map the hollow position coordinates into a three-dimensional network model and generate a three-dimensional thermal image of the building through the Paraview visualization engine; wherein the three-dimensional thermal image of the building displays different colors according to the difference in the dual-interface reflection coefficient.

[0079] 3D thermal images of buildings generated by the ParaView visualization engine can be rendered with transparency, using different colors to encode the severity of hollowing. For example, yellow indicates mild, orange indicates moderate, and red indicates severe. Specifically, in the yellow area, ΔΓ = 0.3-0.6dB corresponds to mild hollowing (gaps of 1-3mm); in the orange area, ΔΓ = 0.6-0.9dB corresponds to moderate hollowing (gaps of 3-5mm); and in the red area, ΔΓ > 0.9dB corresponds to severe hollowing (gaps > 5mm). The color codes correspond to the following reflectance difference ranges: yellow (0.3-0.6dB), orange (0.6-0.9dB), and red (> 0.9dB).

[0080] In one embodiment, the hollowing detection report includes: a building hollowing distribution map, hollowing data, and hollowing detection efficiency data. Specifically, the computer device can output the detection report containing coordinates, area, and depth, and the format can be PDF; specifically, it includes an overview map of the building facade hollowing distribution (marking the number of hollows and area ratio on each floor); a single hollowing detail table (coordinates, area, reflectance difference, recommended repair plan); and detection efficiency statistics (flight time, coverage area, data efficiency ≥ 95%).

[0081] The present application provides a method for detecting hollowing in the insulation layer of a high-rise building based on a drone. Through the fusion technology of laser SLAM pre-modeling and UWB ultra-wideband positioning, a centimeter-level coordinate benchmark for the building facade and a sub-meter-level real-time positioning network for drones (ranging accuracy ≤ 0.1 meters, positioning error ≤ 0.8 meters) are constructed. Combined with spatial geometric transformation and model correction algorithm, the hollowing position accuracy is ultimately controlled to ≤ 0.2 meters. Compared with the meter-level error of traditional GPS, it can accurately match the building facade maintenance unit, avoiding the waste of resources for large-scale inspections and small-scale repairs. For example, in the inspection of a 30-story residential building, it can accurately locate a certain insulation board area on a specific floor, and the maintenance efficiency is improved by more than 50%.

[0082] Through the hardware isolation of the three-axis stabilized gimbal (stable accuracy of ±0.1°) and the optimization of the Kalman filter algorithm, a "hardware + software" dual anti-shake mechanism is formed, effectively suppressing radar signal errors caused by 5-20Hz vibrations during drone flight. Measured data shows that the original signal error is reduced from ±15% to within ±3%, and the echo signal-to-noise ratio (SNR) is improved by 10-15dB, solving the problem of misjudgment of hollow characteristics caused by jitter in traditional millimeter-wave radar applications on mobile platforms. For example, in a flight environment simulating a level 5 wind speed (8.0-10.7m / s), it can still stably capture the reflection signal differences of 3mm-level hollow interfaces.

[0083] Moreover, the laser radar completes the point cloud collection and 3D modeling of the entire building within 2 hours, the detection software generates a spiral scanning route in 3 minutes, and the drone flies automatically according to the planned route (no manual control is required). A single flight (30 minutes of flight time) can cover an inspection area of ​​500-800 square meters, and the daily workload reaches 2000-3000 square meters, which is 20 times the efficiency of manual inspection. At the same time, the automatic generation of the inspection report takes ≤15 minutes, which reduces the manpower input by 80% compared with the traditional method of manual data annotation. Taking a high-rise building with a construction area of ​​15,000 square meters as an example, this application only takes 6-8 working days to complete the full facade inspection, while traditional manual inspection takes 4-6 months, saving time.

[0084] In one embodiment, a method for detecting hollowing of thermal insulation layers of high-rise buildings based on a drone provided in this application is implemented, and the implementation scenario and equipment configuration are as follows:

[0085] Taking a 32-story residential building (98 meters high and 12,000 square meters of facade area) as an example, the following equipment is used for detection:

[0086] UAV platform: Six-axis multi-rotor UAV (wheelbase 850mm, flight time 35 minutes, payload 3kg), equipped with dual-redundant INS inertial navigation system (positioning accuracy centimeter level).

[0087] Millimeter-wave radar module: 77GHz FMCW radar (ranging accuracy 0.05m, scanning angle 45°-60°), integrated three-axis stabilized gimbal (weight 1.2kg, stabilization accuracy ±0.1°) and anti-shake signal processing unit.

[0088] Positioning system: laser SLAM radar (point cloud density 100 points / ㎡), 4 UWB base stations (deployed at the four corners of the building, with a spacing of 60 meters and a ranging accuracy of 0.08m).

[0089] Software system: independently developed detection software (including route planning, signal processing, and hollow drum recognition modules).

[0090] The specific process of implementing the UAV-based high-rise building insulation hollowing detection method is as follows:

[0091] Pre-modeling of building facades (4 hours):

[0092] Data collection: The drone flew around the building twice at a speed of 2m / s, while the lidar simultaneously collected point cloud data of the exterior walls, covering complex structures such as balconies and bay windows, and accumulated more than 1.2 million point cloud coordinates.

[0093] Model construction: A Poisson reconstruction algorithm is used to generate a three-dimensional grid model with geographic coordinates, with an XY plane accuracy of 4 cm and a Z-axis accuracy of 6 cm. Feature points such as floor lines (each floor is 3 meters high) and window opening centers (960 in total) are marked to form a detection reference coordinate system.

[0094] Drone route planning (takes 15 minutes):

[0095] Route generation: The inspection software automatically generates a spiral ascending route based on the building model, starting from a height of 2 meters above the ground and rising 3 meters per circle, with a total of 40-story scanning paths planned (including the rooftop equipment layer);

[0096] Parameter settings: flight altitude: 9 meters from the facade, speed 1.8m / s, scanning angle 50° (vertical incidence on the insulation layer);

[0097] Strip overlap rate: 35% horizontally and 30% vertically to ensure no missed inspection areas.

[0098] Dynamic anti-interference data acquisition (single machine, single day):

[0099] Flight operations: The drone flies automatically according to the route, with each flight lasting 30 minutes and covering 550 square meters. Five flights are carried out per day (total coverage of 2,750 square meters);

[0100] Signal Processing: The inertial navigation system collects flight attitude data (e.g., pitch angle ±2°, roll angle ±1.5°) at a 100Hz frequency. The millimeter-wave radar transmits a frequency-modulated continuous wave (FMCW) at 200Hz. The raw echo signal is processed by a Kalman filter algorithm, reducing jitter error from 12.7% to 2.3% and improving the signal-to-noise ratio by 12dB.

[0101] Hollowing feature recognition and spatial positioning (data processing takes 2 hours):

[0102] Hollowing Identification: Time-frequency analysis was performed on the denoised signal to extract the difference in dual-interface reflection coefficients. Using the SVM classifier, 47 hollowing areas were identified (28 of which were mild, 13 were moderate, and 6 were severe). For example, a hollowing area with a reflection coefficient difference of 0.7 dB and an area of ​​0.25 m2 was identified as moderate hollowing.

[0103] Coordinate calculation: Using UWB positioning data (such as the coordinates of the drone at a certain moment: X = 100.23m, Y = 200.56m, Z = 30.18m), radar beam angle (θ = 45°, ) and distance (d=9.12m), calculate the hollow coordinates: X=100.23+9.12·sin45°·cos85°≈100.91mY=200.56+9.12·sin45°·sin85°≈206.98mZ=30.18+9.12·cos45°≈36.69m; after correction combined with the building model, the positioning accuracy is 0.18m, corresponding to the third insulation board on the southwest side of the 12th floor of the building.

[0104] Visualization of test results (report generation takes 10 minutes):

[0105] 3D thermal imaging: 47 hollow areas were marked in the building model, with the red, severe hollow areas concentrated at the top of the building (9th to 12th floors). This is consistent with engineering experience that high wind pressure can cause the insulation layer to loosen.

[0106] The inspection report shows a total hollow area of ​​28.6 square meters, accounting for 0.24% of the facade area, with severe hollowing accounting for 12.8%. It is recommended that red areas be repaired immediately, orange areas be re-inspected within three months, and yellow areas be included in annual monitoring.

[0107] The hollowing detection method for high-rise building insulation layers based on drones in this application is compared with traditional manual inspection and selected drone lidar inspection. The comparison results are as follows:

[0108]

[0109] The entire implementation process of this example took eight working days (including modeling, flight, and data processing), a 95% reduction compared to traditional manual inspection (which takes five months), and achieved over 10 times higher positioning accuracy. Through anti-shake algorithms and multi-source positioning technology, the system successfully identified tiny hollows at the 3mm level, which are difficult to detect using traditional methods. This provides precise quantitative evidence for building maintenance, significantly reducing the risks and repair costs of high-altitude operations.

[0110] It should be understood that, although the various steps in the above flow chart are shown in sequence as indicated by the arrows, these steps are not necessarily performed in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and these steps can be performed in other orders. Moreover, at least a portion of the steps in the above flow chart may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily performed at the same time, but can be performed at different times, and the execution order of these sub-steps or stages is not necessarily to be performed in sequence, but can be performed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0111] In one embodiment, Figure 3 As shown, a high-rise building insulation layer hollowing detection system based on a drone is provided, comprising: a three-dimensional network model creation module 310, a route generation module 320, a jitter interference suppression module 330, a hollowing recognition module 340 and a detection result display module 350, wherein:

[0112] A 3D network model creation module 310 is used to obtain exterior wall point cloud data collected by a drone equipped with a laser radar flying around the building, and convert the point cloud data into a 3D network model using a Poisson reconstruction algorithm;

[0113] The route generation module 320 is used to generate a spiral scanning UAV route based on the three-dimensional network model and set the laser radar scanning angle, flight data, and scanning overlap rate according to the UAV route;

[0114] The jitter interference suppression module 330 is used to collect flight attitude data and reflected echo signals of the lidar when the drone is equipped with a lidar and fly along the drone's route. The anti-jitter signal processing algorithm is used to suppress flight jitter interference based on the flight attitude data and reflected echo signals to obtain a de-noised radar signal.

[0115] The hollowing recognition module 340 is used to perform time-frequency domain analysis on the radar signal after noise reduction, extract the dual-interface reflection coefficient feature and input it into the support vector machine classifier to identify the hollowing area and calculate the hollowing position coordinates;

[0116] The detection result display module 350 is used to map the hollowing position coordinates into the three-dimensional network model, generate a three-dimensional thermal image of the building, and output a hollowing detection report based on the three-dimensional thermal image of the building.

[0117] In one embodiment, the three-dimensional network model creation module 310 is also used to determine a data acquisition orbit path, and use a drone equipped with a laser radar to fly along the data acquisition orbit path; determine a point cloud density reference value, and use a drone equipped with a laser radar to scan the exterior wall of the building, and collect original data greater than the point cloud density reference value as exterior wall point cloud data.

[0118] In one embodiment, the route generation module 320 is also used to generate a spiral scanning drone route based on a three-dimensional network model, climbing from the bottom to the top of the building floor by floor; setting the scanning angle of the lidar according to the drone route, and the scanning angle is that the radar wave is vertically incident on the surface of the insulation layer of the high-rise building; setting the flight data according to the drone route, and the flight data includes the flight altitude, flight speed, and flight heading angle; setting the scanning overlap rate according to the drone route, and the scanning overlap rate includes the horizontal overlap rate and the vertical overlap rate.

[0119] In one embodiment, the flight attitude data includes pitch angle, roll angle, yaw angle, and acceleration; the reflected echo signal of the laser radar includes timestamp, distance, and phase.

[0120] In one embodiment, the jitter interference suppression module 330 is also used to collect flight jitter frequency based on flight attitude data; the reflected echo signal is filtered in the time and space domain through an anti-jitter signal processing algorithm based on Kalman filtering, and the UAV flight jitter interference of the target frequency is suppressed according to the flight jitter frequency to obtain a noise-reduced radar signal.

[0121] In one embodiment, the hollowing recognition module 340 is also used to perform fast Fourier transform on the radar signal after noise reduction and extract the dual-interface frequency component; calculate the dual-interface reflection coefficient difference based on the dual-interface frequency component as the dual-interface reflection coefficient feature; input the dual-interface reflection coefficient feature into the support vector machine classifier, and use the dual-interface reflection coefficient difference as the standard threshold to identify the hollowing area.

[0122] In one embodiment, the hollowing recognition module 340 is also used to obtain the UWB coordinates of the drone, the azimuth and pitch angles of the radar beam, and the target distance at the detection time; based on the UWB coordinates of the drone, the azimuth and pitch angles of the radar beam, and the target distance, the three-dimensional coordinates of the hollowing are calculated through spatial geometric transformation; the three-dimensional coordinates of the hollowing are corrected based on the three-dimensional network model to obtain the hollowing position coordinates.

[0123] In one embodiment, the detection result display module 350 is also used to map the hollow position coordinates into a three-dimensional network model, and generate a three-dimensional thermal image of the building through the Paraview visualization engine; wherein the three-dimensional thermal image of the building displays different colors according to the difference in the dual-interface reflection coefficient.

[0124] In one embodiment, the hollowing detection report includes: a building hollowing distribution map, hollowing data, and hollowing detection efficiency data.

[0125] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 4 As shown. The computer device includes a processor, memory, network interface, display screen and input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with an external terminal via a network connection. When the computer program is executed by the processor, a method for detecting hollowing of the thermal insulation layer of a high-rise building based on a drone is implemented. The display screen of the computer device can be a liquid crystal display or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a key, trackball or touchpad provided on the computer device housing, or an external keyboard, touchpad or mouse.

[0126] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than shown in the figure, or combine certain components, or have a different component arrangement.

[0127] In one embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of a method for detecting hollowing of thermal insulation layers of high-rise buildings based on a drone are implemented.

[0128] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of a method for detecting hollowing of thermal insulation layers of high-rise buildings based on a drone are implemented.

[0129] Those skilled in the art will appreciate that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0130] The technical features of the above embodiments can be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0131] The above-described embodiments merely represent several implementation methods of the present application. While the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art could make various modifications and improvements without departing from the spirit of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of the present patent application shall be determined by the appended claims.

Claims

1. A method for detecting hollowing of thermal insulation layers in high-rise buildings based on drones, characterized in that: The method comprises: Obtaining exterior wall point cloud data collected by a drone equipped with a laser radar flying around the building, and converting the point cloud data into a three-dimensional network model using a Poisson reconstruction algorithm; Generate a spiral scanning drone route based on the three-dimensional network model, and set a laser radar scanning angle, flight data, and scanning overlap rate according to the drone route; The UAV is equipped with a laser radar and flies along the UAV route to collect flight attitude data and a reflected echo signal of the laser radar. An anti-jitter signal processing algorithm is used to suppress flight jitter interference based on the flight attitude data and the reflected echo signal to obtain a noise-reduced radar signal. Performing time-frequency domain analysis on the noise-reduced radar signal, extracting dual-interface reflection coefficient features and inputting them into a support vector machine classifier to identify hollow areas and calculate hollow position coordinates; The hollowing position coordinates are mapped to the three-dimensional network model to generate a three-dimensional thermal imaging image of the building, and a hollowing detection report is output based on the three-dimensional thermal imaging image of the building.

2. The method for detecting hollowing of thermal insulation layers of high-rise buildings based on drones according to claim 1 is characterized in that: Obtain exterior wall point cloud data collected by flying around the building using a drone equipped with a lidar, including: Determine a data collection orbital path, and use a drone equipped with a laser radar to fly along the data collection orbital path; A point cloud density reference value is determined, and the exterior wall of the building is scanned by the drone equipped with a laser radar to collect raw data greater than the point cloud density reference value as exterior wall point cloud data.

3. The method for detecting hollowing of thermal insulation layer of high-rise buildings based on drone according to claim 1 is characterized in that: Generate a spiral scanning drone route based on the three-dimensional network model, and set the laser radar scanning angle, flight data, and scanning overlap rate according to the drone route, including: Based on the three-dimensional network model, a spiral scanning drone route is generated by climbing from the bottom to the top of the building floor by floor; Setting a scanning angle of the laser radar according to the drone route, wherein the scanning angle is such that the radar wave is incident vertically on the surface of the insulation layer of the high-rise building; Setting flight data according to the UAV route, wherein the flight data includes flight altitude, flight speed, and flight heading angle; The scanning overlap rate is set according to the UAV route, and the scanning overlap rate includes a horizontal overlap rate and a vertical overlap rate.

4. The method for detecting hollowing of thermal insulation layers of high-rise buildings based on drones according to claim 1 is characterized in that: The flight attitude data includes pitch angle, roll angle, yaw angle, and acceleration; the reflected echo signal of the laser radar includes timestamp, distance, and phase.

5. The method for detecting hollowing of thermal insulation layers of high-rise buildings based on drones according to claim 1 is characterized in that: The anti-jitter signal processing algorithm is used to suppress flight jitter interference based on the flight attitude data and the reflected echo signal to obtain a noise-reduced radar signal, including: collecting a flight jitter frequency based on the flight attitude data; The reflected echo signal is filtered in the time and space domain by an anti-jitter signal processing algorithm based on Kalman filtering, and the UAV flight jitter interference of the target frequency is suppressed according to the flight jitter frequency to obtain a noise-reduced radar signal.

6. The method for detecting hollowing of thermal insulation layers of high-rise buildings based on drones according to claim 1 is characterized in that: Performing time-frequency domain analysis on the noise-reduced radar signal, extracting the dual-interface reflection coefficient feature and inputting it into a support vector machine classifier to identify hollow areas, including: Performing a fast Fourier transform on the noise-reduced radar signal and extracting a dual-interface frequency component; Calculating a double-interface reflection coefficient difference according to the double-interface frequency components as a double-interface reflection coefficient feature; The dual-interface reflection coefficient feature is input into a support vector machine classifier, and the dual-interface reflection coefficient difference is used as a standard threshold to identify the hollow area.

7. The method for detecting hollowing of thermal insulation layers of high-rise buildings based on drones according to claim 6 is characterized in that: Calculate the coordinates of the hollow drum position, including: Obtain the UWB coordinates of the drone at the time of detection, radar beam azimuth and pitch angle, and target distance; Based on the UWB coordinates of the drone, the azimuth and pitch angles of the radar beam, and the target distance, the three-dimensional coordinates of the hollow drum are calculated through spatial geometric transformation; The three-dimensional coordinates of the hollow drum are corrected based on the three-dimensional network model to obtain the coordinates of the hollow drum position.

8. The method for detecting hollowing of thermal insulation layers of high-rise buildings based on drones according to claim 7 is characterized in that: Mapping the hollow position coordinates to the three-dimensional network model to generate a three-dimensional thermal image of the building includes: Mapping the hollow position coordinates to the three-dimensional network model, and generating a three-dimensional thermal image of the building through the Paraview visualization engine; The three-dimensional thermal imaging image of the building displays different colors according to the difference in the dual-interface reflection coefficient.

9. The method for detecting hollowing of thermal insulation layers of high-rise buildings based on drones according to claim 1 is characterized in that: The hollowing detection report includes: building hollowing distribution map, hollowing data, and hollowing detection efficiency data.

10. A high-rise building insulation layer hollowing detection system based on drones, characterized in that: The system comprises: A 3D network model creation module is used to obtain exterior wall point cloud data collected by a drone equipped with a laser radar flying around the building, and convert the point cloud data into a 3D network model using a Poisson reconstruction algorithm; a route generation module, configured to generate a spiral scanning UAV route based on the three-dimensional network model, and set a scanning angle, flight data, and scanning overlap rate of a laser radar according to the UAV route; A jitter interference suppression module is used to enable the UAV equipped with a laser radar to fly along the UAV route, collect flight attitude data and the reflected echo signal of the laser radar, and suppress flight jitter interference based on the flight attitude data and the reflected echo signal through an anti-jitter signal processing algorithm to obtain a noise-reduced radar signal; A hollowing recognition module is used to perform time-frequency domain analysis on the noise-reduced radar signal, extract the dual-interface reflection coefficient feature and input it into a support vector machine classifier to identify the hollowing area and calculate the hollowing position coordinates; The detection result display module is used to map the hollowing position coordinates into the three-dimensional network model, generate a three-dimensional thermal imaging image of the building, and output a hollowing detection report based on the three-dimensional thermal imaging image of the building.

Images