Steel-concrete combined bridge crack monitoring system and method based on unmanned aerial vehicle

Abstract

The invention relates to the technical field of bridge engineering and information, and discloses a steel-concrete combined bridge crack monitoring system and method based on an unmanned aerial vehicle, and the system comprises a multi-rotor unmanned aerial vehicle, a ground multi-base-station RTK-GPS system, a stereoscopic vision collection module, a laser radar scanning module and a thermal infrared imager. The multi-rotor unmanned aerial vehicle carries each monitoring module to fly to a to-be-detected area, a ground multi-base-station RTK-GPS system generates a virtual base station through carrier phase difference, and centimeter-level positioning of the unmanned aerial vehicle is achieved; the stereoscopic vision acquisition module acquires a high-definition image, the laser radar scanning module generates point cloud data, and the stereoscopic vision acquisition module and the laser radar scanning module are fused to construct a textured three-dimensional model; and the thermal infrared imager identifies hidden cracks. Millimeter-level non-contact monitoring of the crack of the joint surface of the steel web and the concrete is achieved, personnel climbing is not needed, the positioning precision is high, the adaptability is high, the method is suitable for long-term health monitoring of a bridge, and reliable data support is provided for bridge structure performance prediction.

Description

Technical Field

[0001] This invention relates to the field of railway construction technology, and in particular to a system and method for monitoring cracks in steel-concrete composite bridges based on unmanned aerial vehicles (UAVs). Background Technology

[0002] Steel web and concrete composite beams are a common design method in bridges. The advantages are fast construction speed and reduced concrete usage, thus reducing the self-weight of the beam. However, steel and concrete have different expansion rates. Under weather conditions with large temperature fluctuations, expansion and contraction cracks are prone to occur at the connection between the steel web and the concrete. Rainwater can easily enter the beam and damage the beam structure, reducing the service life of the bridge.

[0003] Currently, the simplest and most commonly used method for crack observation is visual inspection. However, for larger bridges, it is impossible for personnel to climb onto all parts of the beam for observation. Other observation methods include contact monitoring such as crack needles, which install vibrating wire or electronic crack gauges on both sides of the crack to monitor width changes in real time; pre-embedded sensors, which embed fiber optic crack sensors at the joint surface during the construction phase to track crack development over a long period; and non-contact monitoring using image recognition technology and 3D laser scanning technology. For crack depth and internal defects, ultrasonic methods and infrared cameras are typically used.

[0004] All of the above monitoring methods require monitoring personnel to carry instruments and equipment close to the monitored area. For completed steel web-concrete composite bridges, monitoring personnel cannot get close to the connection between the steel web and the concrete to observe cracks. Summary of the Invention

[0005] To overcome the above shortcomings, this invention provides a crack monitoring system and method for steel-concrete composite bridges based on unmanned aerial vehicles (UAVs). By using a multi-rotor UAV equipped with a pair of high-definition optical cameras, lidar, and infrared thermal imagers, combined with ground multi-base station RTK-GPS differential positioning technology, millimeter-level non-contact monitoring of cracks at the interface between the steel web and the concrete can be achieved.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a UAV-based crack monitoring system for steel-concrete composite bridges, comprising a multi-rotor UAV, a ground multi-base station RTK-GPS system, a stereo vision acquisition module, a lidar scanning module, and an infrared thermal imager; the multi-rotor UAV is used to carry the monitoring equipment and fly to the area of ​​the steel-concrete composite beam to be inspected; the ground multi-base station RTK-GPS system consists of multiple ground base stations, generating a virtual base station through carrier phase differential to achieve centimeter-level positioning of the UAV; the stereo vision acquisition module is used to acquire high-definition images of the surface of the steel-concrete composite beam to identify cracks; the lidar scanning module works in conjunction with the stereo vision acquisition module to generate point cloud data of the steel-concrete composite beam structure, ensuring the integrity of the point cloud at the crack edges; the infrared thermal imager is used to detect abnormal heat conduction areas inside the steel-concrete composite beam caused by cracks and identify hidden cracks.

[0007] As a further description of the above technical solution: In the aforementioned ground-based multi-base station RTK-GPS system, the base station uses a weighted least squares method to fuse differential data from multiple base stations to generate virtual base station coordinates. The calculation formula is as follows: In the formula For base station coordinates, This refers to the distance from the drone to the base station.

[0008] As a further description of the above technical solution: the stereo vision acquisition module includes a pair of high-definition optical cameras, the camera distortion is <1%, the cameras are fixed at a 15° angle to the gimbal, and the baseline distance is 15cm; it also includes an image preprocessing unit, which performs distortion correction based on Zhang's calibration method, and then performs grayscale equalization processing.

[0009] As a further description of the above technical solution: the data acquired by the lidar scanning module and the stereo vision acquisition module are fused to construct a textured three-dimensional model, which is used to intuitively display the steel-concrete composite beam structure and crack distribution.

[0010] As a further description of the above technical solution: the system also includes a data processing module, which uses a U-Net++ network model to automatically identify cracks in the acquired image data. The U-Net++ network model training set contains no fewer than 1000 images of bridge cracks under different types and working conditions; the data processing module establishes a crack development prediction model. In the formula The initial width, For temperature sensitivity coefficient, The temperature difference is used to predict the future development trend of cracks.

[0011] As a further description of the above technical solution: the base station shell of the ground multi-base station RTK-GPS system is made of 316L stainless steel; or it adopts PPK post-processing mode, combined with multiple ground base station control points, to overcome the monitoring problem in areas with severe satellite signal obstruction such as canyons.

[0012] As a further description of the above technical solution: the multi-rotor UAV has a built-in SSD to store images, realizing data storage in network-free environments, and is suitable for monitoring steel-concrete composite beams in mountainous areas and other areas without networks.

[0013] A method for monitoring cracks in steel-concrete composite bridges based on unmanned aerial vehicles (UAVs) includes the following steps:

[0014] Step S1: Multi-base station RTK-GPS dynamic differential positioning steps: Before the UAV flies, the ground multi-base station completes initialization and generates a virtual base station through carrier phase differential. During the flight, the UAV receives differential data in real time to achieve centimeter-level positioning.

[0015] Step S2: Stereo vision and LiDAR point cloud fusion modeling step: The UAV flies to the area to be detected of the steel-concrete composite beam. The stereo vision acquisition module and the LiDAR scanning module work synchronously to acquire image and point cloud data. After data fusion, a textured 3D model is constructed.

[0016] Step S3: Automatic Crack Identification and Trend Prediction: The data processing module uses the U-Net++ network model to automatically identify cracks in the acquired images, and predicts the future development trend of cracks based on the identification results and the crack development prediction model.

[0017] As a further description of the above technical solution: step S2 also includes image preprocessing of the acquired data, namely, distortion correction based on Zhang's calibration method, grayscale equalization processing, and noise reduction and other preprocessing operations on the lidar point cloud data.

[0018] As a further description of the above technical solution: In step S3, the U-Net++ network model periodically updates the training set of the crack identification model to improve the model's identification accuracy and adaptability, while also optimizing the crack development prediction model by combining data such as ambient temperature and humidity.

[0019] The present invention has the following beneficial effects:

[0020] 1. In this invention, a multi-rotor drone equipped with a pair of high-definition optical cameras, lidar, and infrared thermal imagers, combined with ground multi-base station RTK-GPS differential positioning technology, can achieve millimeter-level non-contact monitoring of cracks at the interface between the steel web and the concrete. There is no need for monitoring personnel to climb the beams and slabs. The observation is conducted remotely by the drone, which completely avoids the risks of high-altitude operations and improves monitoring efficiency.

[0021] 2. In this invention, no network is required at the bridge location, supporting offline data storage and analysis in network-free environments. The base station is waterproof, lightning-proof, and wind-resistant, making it suitable for various complex scenarios such as mountainous areas, sea crossings, and canyons, thus enhancing environmental adaptability. By predicting crack trends through a crack development prediction model and combining it with bridge structural performance analysis, a scientific basis is provided for bridge maintenance and reinforcement, extending the service life of the bridge. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of data transmission in the monitoring system of the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Example 1

[0025] Reference Figure 1 This invention provides an embodiment of a crack monitoring system for steel-concrete composite bridges based on unmanned aerial vehicles (UAVs). The system comprises a multi-rotor UAV, a ground-based multi-base station RTK-GPS system, a stereo vision acquisition module, a lidar scanning module, an infrared thermal imager, and a data processing module. These components work collaboratively to achieve full-dimensional crack monitoring.

[0026] (1) The multi-rotor UAV adopts a six-axis UAV, which can withstand winds of level 6 (wind speed ≤12m / s). When the wind speed exceeds this, it will automatically return to home to ensure monitoring stability. Its hovering accuracy is high (horizontal ±0.1m, vertical ±0.05m), supports low-altitude and low-speed flight (5-10m / s), and can be close to the surface of the structure (distance 2-5m) for shooting. It is suitable for fine inspection of small and medium span bridges (span ≤100m) or complex structures (variable cross section, box girder interior). The UAV has a built-in large-capacity SSD storage module (≥128GB), which can store 4 hours of 4K images (about 12,000 images). It supports data storage in environments without network, and is suitable for monitoring needs in mountainous areas and other areas without network. At the same time, it has a built-in high-precision inertial attitude measurement system (INS), which can measure the attitude angle of the UAV in real time and correct its position to make up for the position deviation caused by the rapid movement of the UAV between two differential signals.

[0027] (2) The ground multi-base station RTK-GPS system consists of multiple ground base stations. The base station shell is made of 316L stainless steel, which has good corrosion resistance and is suitable for special environments such as cross-sea bridges. The base station is installed at the measurement control point. The installation position is stable and can withstand wind force 6. In areas with wind force exceeding level 6, the system must be shut down and the base station host must be removed. The system must be re-erected after the wind stops.

[0028] The base station generates a virtual base station using carrier phase differential method. The coordinates of the virtual base station are calculated by fusing differential data from multiple base stations using weighted least squares method. The calculation formula is as follows: In the formula For base station coordinates, The distance from the drone to the base station is used. The ground base station receives satellite signals in real time and calculates the error. It then sends the corrected data to the drone via wireless links such as O4 industry version image transmission. The drone combines its own GNSS signals to calculate its position in real time, achieving centimeter-level positioning (horizontal ±1cm +1ppm, vertical ±1.5cm +1ppm). The positioning accuracy is verified as follows:

[0029] In tests on a 500m span bridge, the horizontal error was ≤1.5cm and the vertical error was ≤2cm (total station comparison data). Traditional single-base station positioning accuracy decreases with distance (error ≥10cm at 5km); In this invention with multiple base stations, the error remains stable within ±1cm over a 5km range (see table below).

[0030] Number of base stations Error at 1km Error at 5km Data update rate Traditional single base station 1 ±3cm ±12cm 1Hz This invention has multiple base stations. 3 ±1cm ±1.5cm 5Hz

[0031] The system supports dual-frequency BeiDou / GPS with an initialization time of less than 5 seconds. It can also use PPK post-processing mode and combine ground base station control points at both ends of the bridge for positioning, effectively overcoming the monitoring difficulties in areas with severe satellite signal obstruction, such as canyons.

[0032] (3) The stereo vision acquisition module includes a pair of high-definition optical cameras and an image preprocessing unit; the resolution of the pair of high-definition optical cameras is ≥48 million pixels, ensuring that a single photo can identify a crack of ≥0.1mm; the camera distortion is <1%, the two cameras are fixed at a 15° angle to the gimbal, and the baseline distance is 15cm, which meets the requirements of stereo vision triangulation; it is equipped with an 85mm focal length fixed lens (equivalent to full frame), which is suitable for vertical orthophoto shooting, as well as a 24-70mm zoom lens (such as DJI Zenmuse X7), which can adapt to different distance shooting needs; it has global shutter (to avoid motion blur, which is necessary when the flight speed is >5m / s) and RAW format storage (to retain the original image data for easy later spectral analysis), and the parameters of the two cameras are consistent to shoot stereo image pairs;

[0033] The image preprocessing unit performs distortion correction on the acquired images based on Zhang's calibration method, and then optimizes the image quality through grayscale equalization processing, providing a clear data source for subsequent crack identification.

[0034] (4) The lidar scanning module uses a 16-line lidar with a range of 100m, an accuracy of ±3cm, a scanning frequency of 20Hz, and a point cloud density of ≥100 points / ㎡ to ensure the integrity of the point cloud at the crack edge. This module works in conjunction with the stereo vision acquisition module to generate point cloud data of the steel-concrete composite beam structure. After the point cloud data it collects is fused with the image data from the stereo vision acquisition module, a textured 3D model can be constructed to intuitively display the steel-concrete composite beam structure and crack distribution, while also assisting the UAV in autonomous obstacle avoidance and crack spatial positioning.

[0035] (5) The infrared thermal imager has a resolution of ≥640×512 pixels and a temperature measurement accuracy of ±2℃. Its working principle is to use the temperature gradient difference caused by abnormal heat conduction at the crack to identify hidden cracks such as voids inside the concrete, so as to achieve full-type coverage monitoring of cracks.

[0036] (6) The data processing module uses the U-Net++ network model to automatically identify cracks in the collected image data. The training set of this model contains 100,000 images of bridge cracks under different types and working conditions. The crack identification accuracy is ≥95%, and the false detection rate is <3%. At the same time, a crack development prediction model is established: In the formula The initial width, For temperature sensitivity coefficient, The temperature difference is used to predict the future development trend of cracks by combining data such as ambient temperature and humidity.

[0037] Example 2

[0038] A method for monitoring cracks in steel-concrete composite bridges based on unmanned aerial vehicles (UAVs) includes the following steps:

[0039] Step S1: Equipment Preparation Stage

[0040] a1: Drone preparation: Select a six-axis or higher multi-rotor drone with a built-in high-precision GPS system that supports multiple satellite systems such as the US GPS, GLOMASS, and BeiDou, and has a built-in high-precision inertial positioning system (INS) that can measure the drone's attitude angles and correct its position in real time.

[0041] a2: Preparation of stereo vision acquisition module: Select two high-definition cameras with completely identical parameters to form a camera pair, ensuring that the captured stereo image pair can generate stereo photos and three-dimensional color graphics of cracks;

[0042] a3: Other equipment preparation: Check the working status of the lidar and infrared thermal imager to ensure that the equipment is operating normally; prepare the PC and dedicated analysis tools, and install the data analysis and processing APP.

[0043] Step S2: Base Station Installation Phase

[0044] b1: Layout of Measurement Control Points: For bridges with multiple ground-based measurement control points, one control point is placed on each bank of the river. For bridges with smaller station numbers, one control point is placed on each bank, for a total of four control points. For extra-large bridges with a span of over 500m, two control points are placed on the left and two on the right of each bank, for a total of eight control points on both banks. Control points must be at least 20m away from the bridge structure, with soil compaction ≥93%, avoiding underground pipelines, high-voltage power lines, and mobile base stations. The horizontal position of the control points is determined after rigorous adjustment using a total station or GPS static measurement. The elevation data of the control points is obtained after rigorous adjustment using geometric leveling or precise trigonometric leveling with a total station. The control points are re-measured and corrected annually to ensure no displacement or settlement.

[0045] b2: Base station setup: Following the RTK setup method, a mobile app is used to connect to the base station electronic centering base via Bluetooth to set the elevation angle and sampling interval. The base station electronic centering base has a round bubble, a long bubble, and an electronic bubble, which is integrated with the GPS receiver and has a built-in display screen that can display information such as the electronic bubble centering status, the number of received satellites, PDOP, elevation angle, sampling interval, GPS time, latitude and longitude, etc. Leveling and centering are achieved through laser centering, and the instrument height is automatically set using laser ranging, and the height difference between the ranging center and the GPS phase center is automatically added.

[0046] Step S3: Monitoring and Data Acquisition Phase

[0047] c1: Drone Flight Control: Using a drone remote control device and display (a tablet or mobile app can be used as the remote control display), the drone is manually controlled to fly slowly along the connection between the bridge's steel web and concrete. Workers observe the transmitted images to check for cracks. The drone should maintain a distance of 2-3 meters from the beam surface during flight; if the distance is less than 1 meter, an obstacle avoidance procedure should be initiated to move it away from the beam surface.

[0048] c2: Data Acquisition: After a suspected crack is detected, the stereo vision acquisition module captures stereo image pairs, the lidar scanning module moves to scan the suspected crack, and the infrared thermal imager detects simultaneously; all image data, point cloud data, infrared detection data, and the drone's position and status information are stored together in the drone's built-in SSD storage module.

[0049] Step S4: Data Processing and Analysis Stage

[0050] d1: Data preprocessing: Import the data from the SSD storage module into the PC. The image data is processed by the image preprocessing unit based on Zhang's calibration method for distortion correction and grayscale equalization. The lidar point cloud data is preprocessed for noise reduction.

[0051] d2: 3D modeling: The data processing module fuses the preprocessed image data and point cloud data to construct a textured 3D model, which intuitively presents the spatial distribution of the bridge structure and cracks;

[0052] d3: Crack Identification: The data processing module calls the U-Net++ network model to automatically identify cracks in the fused image data and outputs the crack's location (calculated using the UAV's spatial coordinates and attitude angles to determine the crack's three-dimensional coordinates), length, width, and depth parameters.

[0053] d4: Trend prediction: Combining the initial width data of the identified cracks with the temperature, humidity and other data collected by environmental monitoring equipment, the crack development prediction model is used to predict the future development trend of the cracks.

[0054] d5: Model optimization: Regularly update the training set of the U-Net++ network model, incorporate newly acquired crack image data, and continuously improve the model's recognition accuracy and scene adaptability; optimize the parameters of the crack development prediction model based on actual monitoring data to improve prediction accuracy.

[0055] Step S5: Data Archiving and Decision Support Phase

[0056] e1: Data archiving: Upload monitoring data and analysis results to a remote server for archiving, and record in detail the location and time of crack appearance to provide a reference for subsequent bridge design;

[0057] e2: Structural performance analysis: The design unit combines other monitoring data such as bridge temperature, air pressure, traffic flow, vibration and structural displacement to analyze the structural performance of the bridge and predict its fatigue life, remaining load-bearing capacity, etc.

[0058] e3: Decision Support: Through the user interaction module, users can query monitoring data, analysis results and prediction information. The system provides users with bridge maintenance suggestions and decision support based on the analysis and prediction results, such as whether structural reinforcement or component replacement is required. If the cracks further enlarge or penetrate the thickness of the beam concrete, the system will promptly report to the bridge design unit for remedial and reinforcement measures.

[0059] Example 3

[0060] The process of crack monitoring for a 50m span steel-concrete composite bridge is as follows:

[0061] S1. Equipment Preparation: Select a six-axis multi-rotor drone equipped with a Sony RX1R II high-definition camera (39.9 megapixels), a 16-line LiDAR, and a FLIR VUE PRO-R infrared thermal imager; the drone has a built-in 128GB SSD storage module that supports RAW format storage; prepare 4 ground base stations powered by 12V car starter batteries; install dedicated data analysis software and the U-Net++ network model on the PC.

[0062] S2. Base Station Installation: Four measurement control points are set up on both sides of the bridge, 25m away from the bridge structure, with soil compaction of 95%, avoiding underground pipelines and high-voltage towers; the horizontal position of the control points is determined by GPS static measurement, and the elevation data is obtained by geometric leveling; the base station is fixed at the control point, and the elevation angle is set to 30° and the sampling interval is 1s using a mobile APP; laser centering and leveling are used to ensure that the electron bubble is centered; the lightning protection grounding resistance of the base station is checked to be 3Ω, and the waterproof treatment of the interface meets the requirements.

[0063] S3. Monitoring and Data Acquisition: Start the drone and base station. After the base station completes initialization, a virtual base station is generated. The drone receives differential signals to achieve centimeter-level positioning. Manually control the drone to fly along the connection between the steel web and the concrete at a speed of 8m / s and a distance of 3m from the beam surface. After a suspected crack is detected, the drone hovers, the camera takes stereo images, the lidar scans the suspected area, and the infrared thermal imager detects simultaneously. All data is stored in the SSD module.

[0064] S4. Data Processing and Analysis: Import the data from the SSD into the PC. After distortion correction and grayscale equalization, the image is denoised and the point cloud data is fused to construct a textured 3D model. The U-Net++ network model identifies the crack length as 2.5m, width as 0.3mm, and depth as 1.2cm. Combined with the ambient temperature data (temperature difference of 15℃), the crack development prediction model is substituted into the data and predicts that the crack width will reach 0.45mm after 6 months.

[0065] S5. Decision Support: Upload the monitoring results and analysis report to the remote server. The bridge design unit, in conjunction with other monitoring data, recommends increasing the observation frequency in this area to once every 3 months. No reinforcement treatment is required for the time being.

Claims

1. A crack monitoring system for steel-concrete composite bridges based on unmanned aerial vehicles (UAVs), characterized in that: The system includes a multi-rotor drone, a ground-based multi-base station RTK-GPS system, a stereo vision acquisition module, a lidar scanning module, and an infrared thermal imager. The multi-rotor drone carries the monitoring equipment and flies to the area of ​​the steel-concrete composite beam to be inspected. The ground-based multi-base station RTK-GPS system consists of multiple ground base stations and generates a virtual base station through carrier phase differential to achieve centimeter-level positioning of the drone. The stereo vision acquisition module acquires high-definition images of the surface of the steel-concrete composite beam to identify cracks. The lidar scanning module works in conjunction with the stereo vision acquisition module to generate point cloud data of the steel-concrete composite bridge structure, ensuring the integrity of the point cloud at the crack edges. The infrared thermal imager detects areas of abnormal heat conduction caused by cracks inside the steel-concrete composite beam and identifies hidden cracks.

2. The UAV-based crack monitoring system for steel-concrete composite bridges according to claim 1, characterized in that: In the aforementioned ground-based multi-base station RTK-GPS system, the base station uses a weighted least squares method to fuse differential data from multiple base stations to generate virtual base station coordinates. The calculation formula is as follows: In the formula For base station coordinates, This refers to the distance from the drone to the base station.

3. The UAV-based crack monitoring system for steel-concrete composite bridges according to claim 1, characterized in that: The stereo vision acquisition module includes a pair of high-definition optical cameras with distortion of <1%, fixed at a 15° angle to the gimbal, and a baseline distance of 15cm; it also includes an image preprocessing unit that performs distortion correction based on Zhang's calibration method, and then performs grayscale equalization processing.

4. The UAV-based crack monitoring system for steel-concrete composite bridges according to claim 1, characterized in that: The data acquired by the lidar scanning module and the stereo vision acquisition module are fused to construct a textured 3D model, which is used to intuitively display the steel-concrete composite beam structure and crack distribution.

5. A crack monitoring system for steel-concrete composite bridges based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that: The system also includes a data processing module, which uses a U-Net++ network model to automatically identify cracks in the acquired image data. The U-Net++ network model training set contains no fewer than 1000 images of bridge cracks under different types and working conditions. The data processing module also includes a crack development prediction model. In the formula The initial width, For temperature sensitivity coefficient, The temperature difference is used to predict the future development trend of cracks.

6. The UAV-based crack monitoring system for steel-concrete composite bridges according to claim 1, characterized in that: The base station shell of the ground multi-base station RTK-GPS system is made of 316L stainless steel; or it adopts PPK post-processing mode, combined with ground base station control points at both ends of the bridge, with no less than four, to overcome the monitoring problem in areas with severe satellite signal obstruction such as canyons.

7. A crack monitoring system for steel-concrete composite bridges based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that: The multi-rotor UAV has a built-in SSD to store images, enabling data storage in network-free environments and making it suitable for monitoring steel-concrete composite bridges in mountainous and other network-free areas.

8. A method for monitoring cracks in steel-concrete composite bridges based on unmanned aerial vehicles (UAVs), characterized in that: Includes the following steps: Step S1: Multi-base station RTK-GPS dynamic differential positioning steps: Before the UAV flies, the ground multi-base station completes initialization and generates a virtual base station through carrier phase differential. During the flight, the UAV receives differential data in real time to achieve centimeter-level positioning. Step S2: Stereo vision and LiDAR point cloud fusion modeling step: The UAV flies to the area to be detected of the steel-concrete composite beam. The stereo vision acquisition module and the LiDAR scanning module work synchronously to acquire image and point cloud data. After data fusion, a textured 3D model is constructed. Step S3: Automatic Crack Identification and Trend Prediction: The data processing module uses the U-Net++ network model to automatically identify cracks in the acquired images, and predicts the future development trend of cracks based on the identification results and the crack development prediction model.

9. A method for monitoring cracks in steel-concrete composite bridges based on unmanned aerial vehicles (UAVs) according to claim 8, characterized in that: Step S2 also includes image preprocessing of the acquired data, namely distortion correction based on Zhang's calibration method, grayscale equalization processing, and noise reduction of the lidar point cloud data.

10. A method for monitoring cracks in steel-concrete composite bridges based on unmanned aerial vehicles (UAVs) according to claim 8, characterized in that: In step S3, the U-Net++ network model periodically updates the training set of the crack identification model to improve the model's identification accuracy and adaptability. At the same time, it combines data such as ambient temperature and humidity to optimize the crack development prediction model.

Images