Unmanned aerial vehicle based highway bridge plate rubber bearing rapid inspection method and system

Abstract

The application discloses a kind of highway bridge slab rubber bearing rapid inspection method and system based on unmanned plane, method includes: through unmanned plane, the multi-angle image of bridge and its accessory facilities is shot, the three-dimensional real scene model of bridge surface with position information is established, the specific coordinate position of support is determined, unmanned plane support multi-point inspection flight route is planned;The planned flight route is introduced into unmanned plane remote controller, according to predetermined route, it is flown to bridge position, and high-definition camera device is used to carry out image data acquisition to slab rubber bearing;Unmanned plane will be collected image data real-time transmission back to ground control station, and ground control station carries out preliminary processing and storage to data;Image data that is preliminarily processed by ground control station is transmitted to data analysis processing unit, and the apparent state of slab rubber bearing is comprehensively evaluated, and detailed detection report is generated;The application can carry out targeted detection to bridge support and quickly and accurately identify the disease type and disease degree of support.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of highway bridge inspection, and more specifically relates to a highway bridge plate rubber support rapid inspection method and system based on a UAV. BACKGROUND

[0002] Highway bridges are an important part of modern transportation infrastructure, and their safety and stability are directly related to the efficiency of transportation and public safety. Plate rubber bearings are key components in bridge structures, mainly used to transfer loads, adapt to bridge displacement and rotation, and reduce vibration and impact. However, due to long-term exposure to environmental factors such as temperature changes, humidity, ultraviolet light, and load effects, the bearings are prone to aging, cracking, deformation, and other damage, which can affect the overall performance and safety of the bridge. Therefore, regular inspection and maintenance of the bridge plate rubber bearings is crucial. Traditional bearing inspection methods mainly rely on manual inspection, requiring technicians to climb the bridge structure for close observation and measurement. This method is not only inefficient, but also poses safety risks, especially in complex environments such as elevated bridges or river-crossing bridges. In addition, manual inspection has limited accuracy and coverage, making it difficult to comprehensively and quickly assess the condition of the bearings. Therefore, developing an efficient, safe, and accurate bearing inspection method has become an important demand in the field of bridge maintenance.

[0003] In recent years, with the rapid development of UAV technology, bridge inspection methods based on UAVs have gradually become a research hotspot. UAVs have the advantages of high flexibility, wide coverage, and safe operation, which can effectively make up for the shortcomings of traditional manual inspection. In the existing technology, some research has proposed bridge inspection schemes based on UAVs, such as the CN117406781A patent document, which discloses a bridge intelligent inspection system based on a UAV, including a server, a ground station, and a UAV. The ground station includes a modeling module that establishes a three-dimensional model of the bridge based on data collected when the UAV flies around the bridge, and a path planning module that plans and generates a cruising path for the UAV during inspection based on the three-dimensional model of the bridge. The UAV automatically flies around the bridge along the generated cruising path and collects image data, and the server analyzes the image data to find the disease location on the bridge. This invention establishes a bridge model and automatically plans an inspection route, thereby increasing the judgment of the overall structure of the bridge, avoiding obstacles that may be encountered during inspection, and minimizing the impact of the operator's control ability on the UAV during inspection, ensuring good effectiveness and efficiency of the inspection. However, it mainly focuses on the inspection of the overall structure of the bridge, and lacks detailed detection means for the inspection of the specific structure of the bearing. In addition, there are deficiencies in image data processing and analysis, which cannot quickly and accurately identify the disease type and degree of the bearing. SUMMARY

[0004] In order to overcome the above defects of the prior art or improve the prior art, the present application provides a highway bridge plate rubber support rapid inspection method and system based on a UAV, which determines the specific coordinate position of the support by constructing a three-dimensional real scene model of the bridge surface, plans a UAV support multi-point inspection flight route, takes images of the plate rubber support, analyzes the image data, comprehensively evaluates the apparent state of the plate rubber support, and quickly and accurately identifies the disease type and disease degree of the support.

[0005] To achieve the above object, according to one aspect of the present application, the present application provides a highway bridge plate rubber support rapid inspection method based on a UAV, comprising the following specific steps:

[0006] S1: Taking multi-angle images of the bridge and its auxiliary facilities by the UAV, establishing a three-dimensional real scene model of the bridge surface with position information, determining the specific coordinate position of the support, and planning a UAV support multi-point inspection flight route;

[0007] S2: Importing the planned flight route into the UAV remote controller, flying to the bridge position according to the predetermined route, and collecting image data of the plate rubber support by using a high-definition camera device;

[0008] S3: The UAV transmits the collected image data back to the ground control station in real time, and the ground control station preliminarily processes and stores the data;

[0009] S4: Transmitting the image data preliminarily processed by the ground control station to a data analysis processing unit, comprehensively evaluating the apparent state of the plate rubber support, and generating a detailed detection report;

[0010] S5: According to the detection report, marking and classifying the plate rubber support with problems, providing accurate basis for subsequent maintenance and repair decisions, and attaching suggestions based on data analysis.

[0011] Further, the step S1 comprises:

[0012] S11: Initializing and configuring the UAV body, remote controller, data storage unit, and sensor equipment;

[0013] S12: Dividing the shooting range according to the bridge to be measured and the surrounding real scene, including the bridge deck and its auxiliary facilities, the bridge pier, the beam bottom and the support connection, and setting appropriate flight height and overlap rate, following the order from left to right, from top to bottom, and from main body to auxiliary structure, and completing the shooting;

[0014] S13: arranging aerial images, using a multi-view image homonymy point matching method, calculating the camera space position and shooting angle, generating dense point cloud, connecting point cloud data, constructing triangular mesh model, mapping bridge material texture, establishing bridge surface three-dimensional real scene model, and determining the spatial position of bridge support;

[0015] S14: according to the determined rubber bearing position, dividing the support measuring section, and measuring points;

[0016] S15: planning a multi-point inspection route for the support, and planning a fixed-point shooting action of the unmanned aerial vehicle.

[0017] Further, the step S13 of establishing the bridge surface three-dimensional real scene model comprises:

[0018] S131: importing the multi-view image data, extracting the feature points with significant pixel position in the image, finding the homonymy points representing the same actual ground point in different images according to the collinearity equation of photogrammetry, using the matching information of a large number of homonymy points, and using the least square method to perform adjustment calculation to minimize the residual square sum between the homonymy point coordinates and the theoretical value calculated by the model, so as to calculate the high-precision three-dimensional coordinates of the ground point and the spatial position of the camera, and provide accurate camera parameter basis for subsequent three-dimensional modeling;

[0019] The collinearity equation is as follows:

[0020]

[0021] Wherein, (x, y) is the coordinate of the image point in the image plane coordinate system; (x0, y0) is the image principal point coordinate; f is the camera focal length; (X, Y, Z) is the object space rectangular coordinate of the ground point; (X s , Y s , Z s ) is the object space rectangular coordinate of the camera shooting moment position; a i ,b i ,c i (i=1, 2, 3) is the direction cosine composed of the camera heading angle, the lateral angle and the image rotation angle;

[0022] S132: after completing the overall adjustment of the region, performing multi-view image matching based on the obtained camera parameters, using the gray information and geometric constraint relationship of the image to perform pixel-by-pixel matching operation between adjacent images, mining the corresponding relationship from images with different angles, and then generating dense point cloud data to accurately represent the three-dimensional space shape of the object;

[0023] S133: using the dense point cloud data obtained by multi-view image matching, connecting adjacent points into a series of triangles to construct a three-dimensional irregular triangular mesh;

[0024] S134: according to the constructed three-dimensional irregular triangular mesh, further generating a three-dimensional white model through rendering;

[0025] S135: on the basis of the three-dimensional white model, establishing a mapping relationship from the model space to the image space according to the camera parameters and the three-dimensional coordinates of the model surface points, determining the corresponding image texture pixel position of each model surface point, mapping the texture information of the disease to the model surface, and in the texture mapping process, finding the corresponding texture area of each patch on the model surface in the image; when the texture area appears discontinuous or deformed, using a bilinear texture interpolation method to calculate the value of the target pixel in the texture space according to the values of the surrounding four known pixels, so that the transition of the texture on the model surface is more natural, and finally generating a three-dimensional real scene model of the bridge surface.

[0026] Further, the step S14 comprises:

[0027] After generating the three-dimensional model of the bridge surface, the three-dimensional model of the bridge surface is imported into the unmanned aerial vehicle ground control station, the plate type rubber support to be measured is divided into different independent measuring sections according to the bridge design drawing and the position of the plate type rubber support in the three-dimensional model, taking the longitudinal center line of the bridge as the reference, combining the spacing of the support, the span of the beam body, the arrangement mode of the pier, and the key elements of the river navigation situation; in each measuring section, the top surface center, the edge, and the connecting part of the beam body of the support are selected as the representative measuring points according to the stress key parts and the vulnerable area situation.

[0028] Further, the step S3 comprises:

[0029] By using the dark channel prior algorithm, the model defogging algorithm, the image contrast adjustment, and the super-resolution reconstruction means to reduce the interference of the aerial photograph image; and by using the feature extraction and matching algorithm in S131, a plurality of aerial photograph images with overlapping areas are spliced, key feature points in the images are extracted, the corresponding relationship between the images is found by matching the feature points, and then the plurality of images are seamlessly spliced into a complete support panoramic view by using perspective transformation, thereby providing a macroscopic perspective for support state analysis.

[0030] Further, the step S4 comprises:

[0031] S41: uniform image size and input to the pre-trained visual image processing module;

[0032] S42: Automatically identify cracks, out-drum, deformation, void, and aging appearance defects on the surface of the support according to the disease category and severity through an image recognition algorithm and classify them;

[0033] S43: The visual image processing module generates a detection frame, frames the appearance defect area, and labels the relative position information for output;

[0034] S44: The health status evaluation module classifies and grades the appearance defects according to the appearance defect area, geometric size, and disease classification and grading standards to give the status evaluation results;

[0035] S45: The data storage and management module statistically summarizes the analysis results of the data analysis processing unit and issues a complete detection report.

[0036] Further, the appearance defects in step S44 are assigned scores according to the crack development morphology, out-drum length, shear deformation angle, and position string length characteristics, with a weight ratio of 4:2:2:2, and are classified into three grades of slight, moderate, and severe.

[0037] Further, the detection report in step S45 includes detailed listing of each detection index and status evaluation result of the plate rubber support, and highlighted marking and classification of supports with disease risks.

[0038] According to another aspect of the present application, the present application provides a highway bridge plate rubber support rapid inspection system based on a UAV, which is used to implement the steps of the highway bridge plate rubber support rapid inspection method based on a UAV described above, and includes:

[0039] The UAV flight platform is the core execution unit of the entire inspection system, which includes a UAV body, a sensor device, a high-definition camera device, and a data transmission module, and is used to complete the shooting of the bridge structure and the inspection of the support;

[0040] The ground control station is connected with the UAV flight platform through wireless communication, includes a remote controller and a data storage unit, is used to control the flight path and operation task of the UAV, receives and stores the collected image data from the UAV, and performs preliminary processing;

[0041] The data analysis processing unit is connected with the ground control station, is used to analyze the collected image data, judge the appearance defects of the support, give the status evaluation results, and generate a detection report.

[0042] Further, the data analysis processing unit includes:

[0043] Visual image processing module: through image recognition algorithm, the collected support image is analyzed, and the cracks, bulges, deformation, void, aging appearance defects are automatically recognized;

[0044] Health status evaluation module: combined with the image analysis result, the health status of the support is evaluated and graded, and the state evaluation result is given;

[0045] Data storage and management module: used for storing the collected images and analysis results in the database, generating detailed detection report, facilitating historical data query and comparison, and providing support for long-term health monitoring of the bridge.

[0046] Overall, compared with the prior art, the above technical solutions conceived by the present application can achieve the following beneficial effects:

[0047] 1. The unmanned aerial vehicle based rapid inspection method for highway bridge plate type rubber support of the present application can obtain support surface disease images by pre-constructing a bridge surface three-dimensional real scene model with position information, planning an unmanned aerial vehicle under-beam inspection path using the model, arranging different measurement section points, setting different shooting actions, and clearly defining specific task targets to be detected, thereby providing data support for support detection and disease analysis, improving support detection efficiency, and ensuring bridge safety performance and carrying capacity.

[0048] 2. The unmanned aerial vehicle based rapid inspection method for highway bridge plate type rubber support of the present application can automatically recognize and classify cracks, bulges, deformation, void, aging and other appearance defects on the surface of the support according to the image recognition algorithm of the data analysis processing module, and grade them according to the disease degree, so as to quickly and accurately develop a maintenance and repair scheme and ensure the safe operation of the bridge.

[0049] 3. The unmanned aerial vehicle based rapid inspection method for highway bridge plate type rubber support of the present application can reduce interference by preliminary processing of aerial photography images through the ground control station, and seamlessly splice multiple images into a complete support panoramic view through image stitching by feature extraction and matching algorithm, thereby providing a macroscopic perspective for support state analysis and improving detection accuracy.

[0050] 4. The data analysis processing unit of the unmanned aerial vehicle based rapid inspection system for highway bridge plate type rubber support of the present application can realize small sample learning for new data sets and task requirements, so as to quickly complete training on a large scale of extensive data and generalize application to downstream tasks, thereby improving the generalization ability of the model. In different scene requirements, even if the support damage situation category is not clear or multiple damage types may be involved, the support defects can be accurately detected, and the method can be applied to realize bridge support detection of various different configurations and different spans. BRIEF DESCRIPTION OF DRAWINGS

[0051] Figure 1 A flow chart of a rapid patrol inspection method for highway bridge plate rubber bearings based on a UAV is provided for an embodiment of the present application.

[0052] Figure 2 A data analysis processing flow chart of a data analysis processing unit of a rapid patrol inspection method for highway bridge plate rubber bearings based on a UAV is provided for an embodiment of the present application.

[0053] Figure 3 A schematic diagram of a rapid patrol inspection system for highway bridge plate rubber bearings based on a UAV is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0054] In order to make the purpose, technical solutions and advantages of the present application clearer and more apparent, the present application will be further described in detail below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and do not limit the present application. In addition, the technical features involved in each embodiment of the present application described below can be combined with each other as long as they do not conflict with each other.

[0055] Embodiment 1

[0056] As shown in Figure 3 , an embodiment of the present application provides a rapid patrol inspection system for highway bridge plate rubber bearings based on a UAV, which comprises:

[0057] The UAV flight platform is the core execution unit of the entire patrol inspection system, which comprises a UAV body, a sensor device, a high-definition camera device and a data transmission module, which are used to complete the shooting of the bridge structure and the patrol inspection task of the bearing;

[0058] The high-definition camera device is used to obtain clear images of the plate rubber bearing to capture appearance defects such as cracks, bulges, deformation, voids and aging on the surface of the bearing. The data transmission module transmits the collected images and data to the ground control station in real time, and at the same time, allows the ground control station to control and adjust the task of the UAV in real time;

[0059] The ground control station is connected with the UAV flight platform through wireless communication, which comprises a remote controller and a data storage unit, and is used to control the flight path and operation task of the UAV, and receive and store the collected images and data from the UAV and perform preliminary processing;

[0060] The data analysis processing unit is connected with the ground control station, which is used to analyze the collected image data, judge the appearance defects of the bearing and give a state evaluation result, and generate a detection report. It comprises:

[0061] Visual image processing module: through image recognition algorithm, the collected support image is analyzed, and the appearance defects such as crack, bulge and deformation are automatically recognized. The recognition accuracy can reach millimeter level, and the disease part can be accurately positioned.

[0062] Health status evaluation module: combined with the image analysis result, the health status of the support is evaluated and graded, and the state evaluation result is given, which provides basis for subsequent maintenance and repair.

[0063] Data storage and management module: used for storing the collected images and analysis results in the database, generating detailed detection report, facilitating historical data query and comparison, and providing support for long-term health monitoring of bridge.

[0064] Example 2

[0065] As Figure 1 shown, based on the above system, the embodiment of the application provides a rapid patrol inspection method for highway bridge plate rubber support based on unmanned aerial vehicle, which comprises the following steps:

[0066] S1: the multi-view image of the bridge and its auxiliary facilities is shot by the unmanned aerial vehicle, the three-dimensional real scene model of the bridge surface with position information is established, the specific coordinate position of the support is determined, the unmanned aerial vehicle support multi-point patrol flight route is planned, and all plate rubber support positions of the bridge to be detected can be covered.

[0067] Specifically, it includes the following steps:

[0068] S11: the unmanned aerial vehicle body, remote controller, data storage unit and sensor equipment are initialized and configured;

[0069] S12: according to the measured bridge and surrounding real scene, the shooting range is divided, including the bridge deck and its auxiliary facilities, bridge pier, beam bottom and support connection, etc., and appropriate flight height and overlap rate are set, the sequence from left to right, from top to bottom, first main body and then auxiliary structure is followed, and shooting is completed; specifically, it includes:

[0070] According to the measured bridge structure characteristics and surrounding building environment, the measurement range is circled, the appropriate flight height is set, the image overlap rate is guaranteed to be greater than 70%, the five-direction oblique photography flight route is planned to shoot the bridge deck system and its superstructure. After completing five-direction photography, the unmanned aerial vehicle is controlled by the flight hand, the height is reduced in steps, the bridge is overflown, and the manual surrounding supplementary shooting is completed;

[0071] According to the characteristics of the bridge side and its auxiliary facilities, the image overlap rate is guaranteed to be greater than 70%, the "bow" type flight route is planned to patrol the bridge body. After completing the bow-shaped patrol, the height of the flight vehicle is reduced below the beam bottom, and the multi-point sweeping supplementary shooting is completed;

[0072] According to the bridge pier structure characteristics, the specific route is planned, the image overlap rate is guaranteed to be greater than 70%, the zigzag route is planned, the unmanned aerial vehicle is manually controlled in the occasion where the pier is relatively dense, the positioning and obstacle avoidance functions are closed to take pictures;

[0073] After the shooting is completed, the unmanned aerial vehicle is controlled to pass under the beam, the support information is supplemented at a safe distance, and the sensor equipment and the support position are kept level during the navigation process. Combined with the laser range finder and the unmanned aerial vehicle positioning function, the support position information is obtained.

[0074] S13: The aerial image is arranged, a multi-view image same-name point matching method is used, the camera space position and the shooting angle are solved, a dense point cloud is generated, the point cloud data is connected, a triangular mesh model is constructed, the bridge material texture is mapped, a bridge surface three-dimensional real scene model is established, and the bridge support space position is determined. Specifically, the following steps are included:

[0075] S131: According to the collinearity equation of photogrammetry, the multi-view image data is imported, the feature points including the corner points, edge points and other pixel position significant points in the image are extracted, the same-name point matching method is used to find the same actual ground point in different images. With the matching information of a large number of same-name points, the least square method is used for adjustment calculation, so that the residual sum of squares between the point coordinates and the theoretical value calculated by the model is minimized, so that the high-precision ground point three-dimensional coordinates and the space position and posture of the camera are solved, and accurate camera parameter basis is provided for subsequent three-dimensional modeling. The collinearity equation is as follows:

[0076]

[0077] Wherein, (x, y) is the coordinate of the image point in the image plane coordinate system; (x0, y0) is the image principal point coordinate; f is the camera focal length. (X, Y, Z) is the object space rectangular coordinate of the ground point; (X s , Y s , Z s ) is the object space rectangular coordinate of the camera position at the moment of shooting; a i ,b i ,c i (i=1, 2, 3) are direction cosines composed of the camera heading angle, the lateral angle and the image rotation angle.

[0078] S132: After completing the overall adjustment of the region, multi-view image matching is performed based on the camera parameters obtained, the gray information and geometric constraints of the image are used, and the pixel-by-pixel matching operation is performed between adjacent images. The corresponding relationship is mined from the images of different angles, and then the dense point cloud data is generated, the three-dimensional space shape of the object is accurately represented, and rich data support is provided for constructing the high-precision bridge three-dimensional model.

[0079] S133: Dense point cloud data obtained by multi-view image matching is used to construct a three-dimensional irregular triangular mesh. According to the spatial distribution of the point cloud, adjacent points are connected to form a series of triangles, and no other points in the point cloud are contained in the circumscribed circle of each triangle, so as to ensure the shape quality of the mesh and not affect the precision and stability of the model. All the triangular meshes form a network structure covering the surface of the object, fitting the complex three-dimensional geometry of the bridge surface, facilitating the maintenance of topological relationship and geometric calculation in subsequent model processing.

[0080] S134: According to the constructed three-dimensional irregular triangular mesh, a three-dimensional white model is further generated. The three-dimensional white model is a three-dimensional geometric model without texture information. It is based on the triangular mesh and presents the geometric shape of the model in a visual way through rendering. In the rendering process, according to the topological relationship and geometric coordinates of the triangular mesh, the position and orientation of each triangular patch in three-dimensional space are calculated, and then the lighting model is used to determine the light and dark effect of the model surface, generating a white model with only geometric shape, preparing for the subsequent addition of bridge texture material.

[0081] S135: Based on the three-dimensional white model, the mapping relationship from the model space to the image space is established according to the camera parameters and the three-dimensional coordinates of the model surface points, and the corresponding image texture pixel position of each model surface point is determined. The texture information of the disease is mapped to the model surface. In the texture mapping process, for each patch on the model surface, find its corresponding texture area in the image. When the texture area appears discontinuous or deformed, use the bilinear texture interpolation method to calculate the value of the target pixel in the texture space according to the values of the surrounding four known pixels, so that the transition of the texture on the model surface is more natural, and finally generate a three-dimensional real scene model of the bridge surface;

[0082] S136: The spatial position of the bridge support is determined by combining the professional measurement data and the spatial coordinate information in the model.

[0083] S14: According to the determined position of the plate rubber support, the support measurement section is divided, and the measurement points are determined.

[0084] Specifically, after generating the three-dimensional model of the bridge surface, the three-dimensional model of the bridge surface is imported into the unmanned aerial vehicle ground control station. First, according to the position of the plate rubber support in the three-dimensional model and the bridge design drawing, taking the longitudinal center line of the bridge as the reference, combining the key elements such as the spacing of the support, the span of the beam body, the arrangement mode of the pier, and the navigation condition of the river, the measured plate rubber support is divided into different independent measurement sections. In each measurement section, the top center of the support, the edge, and the connection part with the beam body can be selected as representative measurement points according to the stress key parts and the vulnerable area.

[0085] S15: A multi-point inspection route of the support is planned, and a UAV fixed-point shooting action is planned, and specific parameters including a camera shooting angle, a gimbal pitch range, a lens zoom ratio, optical shooting, and infrared shooting are set.

[0086] S2: The planned flight route is imported into a UAV remote controller, and the UAV is flown to the bridge position according to the predetermined route, and high-definition camera devices are used to collect image data of the plate rubber support, specifically including the following steps:

[0087] S21: The UAV body, remote controller, data storage unit, and sensor device are initialized and configured.

[0088] S22: The UAV is launched at the predetermined position, and the support detection route task is performed.

[0089] S23: A high-definition zoom camera is used to shoot clear appearance images of the plate rubber support.

[0090] During the shooting process, the aperture size, shutter speed, and sensitivity need to be adjusted reasonably according to the light conditions, such as sunny, cloudy, backlight, etc., to ensure that the shooting picture is clear, bright, and has high color reproduction.

[0091] S3: The UAV transmits the collected image data back to the ground control station in real time, and the ground control station performs preliminary processing and storage of the data; specifically, including the following contents:

[0092] The aerial images may be subject to various interferences, such as rain and fog interference, strong light interference, lens defocus, etc. The interference can be reduced by using dark channel prior algorithm, model de-fogging algorithm, image contrast adjustment, super-resolution reconstruction, etc. When the ground control station receives multiple aerial images with overlapping areas, the images can also be spliced by the aforementioned feature extraction and matching algorithm, the key feature points in the images are extracted, the corresponding relationship between the images is found by matching the feature points, and then the multiple images are seamlessly spliced into a complete support panoramic image by perspective transformation, providing a macroscopic perspective for support state analysis.

[0093] A solid state disk or a network storage device with fast read and write speed and large storage space is used to store data, and the data is managed by classification and naming according to a reasonable organization architecture.

[0094] S4: The image data preliminarily processed by the ground control station is transmitted to a data analysis processing unit, the apparent state of the plate rubber support is comprehensively evaluated, and a detailed detection report is generated; specifically, as shown in the following table, including the following specific steps: Figure 2

[0095] S41: The image size is unified and input to a pre-trained visual image processing module. ​

[0096] S42: Automatically identify and classify the appearance defects such as cracks, bulging, deformation, void, aging, etc. on the surface of the support according to the disease categories and severity by image recognition algorithm.

[0097] S43: The visual image processing module generates a detection frame, frames the appearance defect area, and labels the relative position information for output.

[0098] S44: The health status evaluation module grades the appearance defects according to the appearance defect area, geometric size, and disease classification grading standards to give the status evaluation results. According to the crack development morphology, bulging length, shear deformation angle, position string length, etc., the four detection indexes are assigned according to the weight ratio of 4:2:2:2 to distinguish into three grades of slight, moderate, and severe, generate the status evaluation results, and provide suggestions and maintenance schemes.

[0099] S45: The data storage and management module statistically summarizes the analysis results of the data analysis processing unit to issue a complete detection report, which includes detailed listing of each detection index and status evaluation result of the plate rubber support, and highlights the support with disease risk for classification, so as to quickly and accurately develop maintenance and repair scheme to ensure the safe operation of the bridge.

[0100] S5: According to the detection report, mark and classify the plate rubber support with problems to provide accurate basis for subsequent maintenance and repair decision, and attach the suggested measures based on data analysis. If the support is in the slight disease stage, it is recommended to be listed as an observation object for regular review and monitoring; if it is in the moderate disease stage, a targeted maintenance scheme is proposed; if it is in the severe disease stage, an alarm is immediately given to the relevant management department.

[0101] Those skilled in the art will readily understand that the above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modification, equivalent replacement, and improvement within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for rapid inspection of highway bridge slab rubber bearings based on unmanned aerial vehicles, characterized in that, The method comprises the following specific steps: S1: shooting the bridge and its auxiliary facilities by the unmanned aerial vehicle to establish a three-dimensional real scene model of the bridge surface with position information, determining the specific coordinate position of the support, and planning the multi-point inspection flight route of the unmanned aerial vehicle support; S2: importing the planned flight route into the remote controller of the unmanned aerial vehicle, flying to the bridge position according to the predetermined route, and collecting image data of the plate rubber support by using the high-definition camera device; S3: the unmanned aerial vehicle transmits the collected image data back to the ground control station in real time, and the ground control station preliminarily processes and stores the data; S4: transmitting the image data preliminarily processed by the ground control station to the data analysis processing unit, comprehensively evaluating the apparent state of the plate rubber support, and generating a detailed detection report; S5: according to the detection report, marking and classifying the plate rubber support with problems, and providing accurate basis for subsequent maintenance and repair decision; The step S1 comprises: S11: initializing and configuring the unmanned aerial vehicle body, remote controller, data storage unit and sensor equipment; S12: according to the to-be-measured bridge and the surrounding real scene, dividing the shooting range, including the bridge deck and its auxiliary facilities, the bridge pier, the beam bottom and the support connection, and setting appropriate flight height and overlap rate, following the order from left to right, from top to bottom, and from main body to auxiliary structure, and completing shooting; S13: arranging the aerial image, using a multi-view image homonymy point matching method, solving the camera space position and shooting angle, generating dense point cloud, connecting point cloud data, constructing a triangular mesh model, mapping the bridge material texture, establishing a three-dimensional real scene model of the bridge surface, and determining the spatial position of the bridge support; S14: according to the determined position of the plate rubber support, dividing the support measurement section and measuring point; S15: planning the support multi-point inspection route, and planning the unmanned aerial vehicle fixed-point shooting action; The step S13 comprises: S131: importing the multi-view image data, extracting the feature points with significant pixel position in the image, finding the image points representing the same actual ground point in different images by the homonymy point matching method according to the collinearity equation of photogrammetry, using the matching information of a large number of homonymy points, and using the least square method for adjustment calculation to make the residual sum of squares between the image point coordinates and the theoretical values calculated by the model minimum, so as to solve the high-precision ground point three-dimensional coordinates and the space position and posture of the camera, and provide accurate camera parameter basis for subsequent three-dimensional modeling; The collinearity equation is as follows: wherein, is the coordinate of the image point in the image plane coordinate system; is the image principal point coordinate; is the camera focal length; is the object space rectangular coordinate of the ground point; is the object space rectangular coordinate of the camera position at the moment of shooting; is the direction cosine composed of the camera heading angle, the side angle and the picture rotation angle. S132: after completing the overall adjustment of the region, performing multi-view image matching based on the obtained camera parameters, using the gray information and geometric constraint relationship of the image to perform pixel-by-pixel matching operation between adjacent images, mining the corresponding relationship from images with different angles, and then generating dense point cloud data to accurately represent the three-dimensional space shape of the object; S133: using the dense point cloud data obtained by multi-view image matching to connect adjacent points into a series of triangles to construct a three-dimensional irregular triangular mesh. S134: According to the constructed three-dimensional irregular triangle net, further generate three-dimensional white model through rendering; S135: On the basis of three-dimensional white model, according to camera parameters and three-dimensional coordinates of model surface points, the mapping relationship from model space to image space is established, the corresponding image texture pixel position of each model surface point is determined, the texture information of disease is mapped to the model surface, and in the texture mapping process, for each surface patch of the model surface, the corresponding texture area in the image is found;When the texture area is discontinuous or deformed, the bilinear texture interpolation method is used, the value of the target pixel is calculated according to the values of the surrounding four known pixels in the texture space, so that the transition of the texture on the model surface is more natural, and finally the three-dimensional real scene model of the bridge surface is generated; The step S14 comprises: After generating the three-dimensional model of the bridge surface, the three-dimensional model of the bridge surface is imported into the unmanned aerial vehicle ground control station, the plate type rubber support to be measured is divided into different independent measuring sections according to the bridge design drawing and the position of the plate type rubber support in the three-dimensional model, the longitudinal center line of the bridge is taken as the reference, the spacing of the support, the span of the beam body, the arrangement mode of the pier, and the key elements of the river navigation situation are combined, and the plate type rubber support to be measured is divided into different independent measuring sections; The step S3 comprises: By using dark channel prior algorithm, model defogging algorithm, image contrast adjustment and super resolution reconstruction means to reduce the interference of aerial images, and by using the feature extraction and matching algorithm in S131, the key feature points in the image are extracted, and then the corresponding relationship between the images is found by matching the feature points, and then the multiple images are seamlessly spliced into a complete support panoramic view by using perspective transformation, so that a macroscopic perspective for support state analysis is provided.

2. The method according to claim 1, characterized in that, The step S4 comprises: S41: unify the image size and input into the pre-trained visual image processing module; S42: through the image recognition algorithm, the cracks, bulging, deformation, void, and aging appearance defects on the surface of the support are automatically recognized and classified according to the disease categories and severity; S43: the visual image processing module generates a detection frame, selects the appearance defect area, and labels the relative position information for output; S44: the health state evaluation module grades the appearance defects according to the appearance defect area, geometric size disease classification and grading standard, and gives the state evaluation result; S45: the data storage and management module statistically analyzes the analysis results of the data analysis processing unit, and issues a complete detection report.

3. The rapid inspection method for highway bridge plate rubber bearings based on unmanned aerial vehicles (UAVs) according to claim 2, characterized in that, In step S44, the appearance defects are scored according to the crack development morphology, bulging length, shear deformation angle, position string length characteristics, and the four detection indexes are scored according to the weight ratio of 4:2:2:2, and are divided into three grades of slight, moderate and severe.

4. The method according to claim 3, characterized in that, In step S45, the detection report comprises: detailed listing of each detection index and state evaluation result of each plate type rubber support, and highlighting marking and classification of supports with disease risk.

5. A UAV-based rapid inspection system for highway bridge slab rubber bearing, for implementing the steps of a UAV-based rapid inspection method for highway bridge slab rubber bearing according to any one of claims 1-4, characterized in that, Comprise: The unmanned aerial vehicle flight platform is a core execution unit of the whole inspection system, and includes an unmanned aerial vehicle body, a sensor device, a high-definition camera device, and a data transmission module, which are used to complete the shooting of the bridge structure and the inspection of the support; The ground control station is connected with the unmanned aerial vehicle flight platform through wireless communication, and includes a remote controller and a data storage unit, which are used to control the flight path and operation task of the unmanned aerial vehicle, receive and store the collected image data from the unmanned aerial vehicle, and perform preliminary processing; The data analysis and processing unit is connected with the ground control station, and is used to analyze the collected image data, judge the appearance defects of the support, give a state evaluation result, and generate a detection report.

6. The unmanned aerial vehicle based highway bridge slab rubber bearing rapid patrol inspection system according to claim 5, characterized in that, The data analysis and processing unit comprises: a visual image processing module: through an image recognition algorithm, the collected support image is analyzed, and cracks, bulges, deformation, voids, and aging appearance defects are automatically identified; a health state evaluation module: combined with the image analysis result, the health state of the support is evaluated and graded, and a state evaluation result is given; a data storage and management module: used to store the collected image and analysis result in a database, generate a detailed detection report, and facilitate the query and comparison of historical data, thereby providing support for long-term health monitoring of the bridge.

Images