Concrete structure cracking risk area cross-scale visual diagnosis method and device

Abstract

The application discloses a concrete structure cracking risk area cross-scale visual diagnosis method and device, and the method comprises the following steps: collecting a macro image of a cracking risk target area and a fluorescence excitation image in a fluorescence excitation area; performing three-dimensional real scene modeling to obtain a non-fluorescence excitation real scene model and a fluorescence excitation real scene model; using the characteristic that the morphology of a hidden crack is amplified when the hidden crack is subjected to fluorescence imaging, embedding the fluorescence imaging morphology of the hidden crack into the non-fluorescence excitation real scene model to obtain a crack cross-scale visual three-dimensional real scene model; and according to the comprehensive development law of macro cracks and hidden cracks in the crack cross-scale visual three-dimensional real scene model, evaluating and diagnosing the cracking risk degree of the concrete structure according to regions. Through the cooperative detection of macro cracks and micro cracks, the cross-scale visual detection of the concrete structure cracking is realized; and the cracking risk area is taken as a unit to evaluate the destructiveness of the cracking risk area to the concrete structure, so that the evaluation result is more reliable.

Description

Technical Field

[0001] This invention belongs to the field of concrete structure cracking risk diagnosis technology, and specifically relates to a cross-scale visualization diagnosis method and device for concrete structure cracking risk zones. Background Technology

[0002] Cracking is the most common form of damage to concrete structures, and accurate diagnosis of structural cracking risk is crucial for the safety and prevention of concrete structures. Currently, cracking risk assessment of concrete structures is generally limited to judging the hazard of a single crack. However, the growth of a single crack has a significant uncertainty regarding the degree of structural damage. Furthermore, assessments mainly focus on the stage-based damage caused by large-scale macroscopic cracks. In cracking risk identification, the focus is primarily on detecting macroscopic cracks that have already caused damage. The occurrence and development of microscopic latent cracks characterize the potential deterioration trend of their location. By comprehensively investigating the development patterns of both macroscopic and microscopic cracks, and systematically analyzing the cracking damage currently experienced and will subsequently face in the cracked area, the degree of cracking risk at the structural region level can be accurately assessed. Although Chinese patent CN117309837A discloses a fluorescent selective excitation flight detection and location technology for latent cracks in concrete under real-world conditions, which can detect latent cracks at the microscopic scale under engineering conditions, it is limited to detecting latent cracks and cannot yet address the detection of macroscopic cracks. Summary of the Invention

[0003] To address the problems existing in the prior art, the present invention aims to provide a cross-scale visual diagnostic method and apparatus for crack risk zones in concrete structures.

[0004] To achieve the above objectives, the present invention adopts the following technical solution:

[0005] The first aspect of this invention provides a cross-scale visual diagnostic method for crack risk zones in concrete structures, comprising the following steps:

[0006] S1. Use a spraying drone to carry clean water to clean the structural surfaces of the target area at risk of cracking;

[0007] S2. A detection drone is used to collect macroscopic images of the target area at risk of cracking, and macroscopic cracks with a width of more than 0.1 mm are detected in the target area; in this invention, cracks with an opening width greater than 0.1 mm are defined as macroscopic cracks.

[0008] S3. After the drone to be detected completes the detection of macroscopic cracks with a width of more than 0.1 mm in the target area of ​​crack risk, the liquid spraying drone is used to spray the fluorescent excitation solution on the concentrated area and key area of ​​macroscopic cracks detected by the drone in the target area for fluorescent excitation. Then, the drone is used to collect the fluorescent excitation image in the fluorescent excitation area to detect hidden cracks with a width of less than 0.1 mm marked by fluorescent morphology.

[0009] S4. The data processing unit in the ground remote control platform uses the obtained macroscopic image and fluorescence excitation image to perform three-dimensional real scene modeling, respectively, to obtain non-fluorescent excitation real scene model and fluorescence excitation real scene model;

[0010] S5. Taking advantage of the characteristic that the morphology of hidden cracks is magnified during fluorescence development, the deep learning image recognition technology based on the segmented attention model in the data processing unit of the ground remote control platform is used to extract the fluorescence-excited development morphology and coordinate position information of hidden cracks from the fluorescence-excited real scene model. Then, the fluorescence-excited development morphology of hidden cracks is embedded into the non-fluorescence-excited real scene model according to the coordinate position information to construct a cross-scale visualized three-dimensional real scene model of cracks.

[0011] S6. Using a ground-based remote control platform, the regional cracking risk assessment unit is employed to diagnose the degree of cracking risk in concrete structures based on the comprehensive development patterns of macroscopic and latent cracks in the cross-scale visualized 3D real-world model of cracks. The assessment principles and corresponding levels are as follows:

[0012] a. High-risk cracking zone, with a total of more than 8 macroscopic and latent cracks per m. 2 Among them, hidden cracks account for no more than 50%, and the maximum width of macroscopic cracks is greater than 0.5 mm;

[0013] b. Moderate risk zone for cracking, with a total of more than 4 macroscopic and hidden cracks per m. 2 Among them, macroscopic cracks exist and account for no more than 50%, with the maximum width of macroscopic cracks greater than 0.3 mm; or the total number of macroscopic and hidden cracks is greater than 4 / m. 2 Among them, the proportion of hidden cracks shall not exceed 50%, and the maximum width of macro cracks shall be 0.3-0.5 mm; or the total number of macro cracks and hidden cracks shall be 5-8 per m. 2 Of these, hidden cracks account for no more than 50%, and the maximum crack width is greater than 0.5 mm;

[0014] c. Potential cracking risk zone: Only macroscopic cracks exist, with 2-4 macroscopic cracks per m. 2 The maximum crack width is greater than 0.3 mm; or the total number of macroscopic cracks and hidden cracks is greater than 4 / m. 2 Among them, macroscopic cracks account for no more than 50%, and the maximum width of macroscopic cracks is less than or equal to 0.3 mm.

[0015] Preferably, in step S2, the use of a detection drone to collect macroscopic images of the target area at risk of cracking, and the detection of macroscopic cracks wider than 0.1 mm in the target area, specifically includes the following steps:

[0016] The drone is driven to fly to the shooting position and continuously takes pictures of the target area starting from the right apex of the crack risk target area. While taking pictures, the drone hovers. After taking pictures, the drone moves horizontally to the left at the same distance from the target area, with each movement not exceeding 3 meters. After moving, the drone hovers and takes pictures of the target area. When the drone completes taking pictures at the left apex of the target area, it descends vertically at the same distance from the target area, with each descent not exceeding 2 meters. After descending, it continues to hover and take pictures to the right of the target area. When the drone reaches the right boundary of the target area, it descends again. The drone repeats this hovering and taking pictures until the entire target area is covered. The shooting position is 5 to 9 meters away from the crack risk target area.

[0017] Preferably, in step S3, the method of using a spraying drone to carry a fluorescent excitation solution to spray the solution onto the macroscopic crack concentration area and key area detected by the detection drone in the target area for fluorescent excitation, and then using the detection drone to collect fluorescent excitation images in the fluorescent excitation area to detect hidden cracks less than 0.1 mm wide marked by fluorescent morphology, specifically includes the following steps:

[0018] A spraying drone, loaded with fluorescent excitation solution, flies to the spraying location and sprays solution to fully cover the macroscopic crack concentration area and critical area identified by the detection drone in the target area. The spraying rate is controlled to ensure that the fluorescent excitation solution flows freely along the surface of the excited area. Then, after the spraying drone begins spraying from the initial excited area, the detection drone flies to the shooting position and continuously hovers to take pictures of the fluorescently excited area. The detection drone is controlled to ensure that the excited area being photographed is within 30s to 60s after the fluorescent excitation, until the captured image completely covers the fluorescently excited area. The spraying location is 2-5m away from the macroscopic crack concentration area and critical area, and the shooting position is 5-9m away from the fluorescently excited area, which is the same distance as the distance from the detection drone to the target area in step S2.

[0019] Preferably, the horizontal overlap between adjacent macroscopic images obtained by the detection drone in continuous photography is 70%–90%, and the vertical overlap is 60%–80%; the horizontal overlap between adjacent fluorescence excitation images is 70%–90%, and the vertical overlap is 60%–80%.

[0020] Preferably, in step S5, the microscopic morphology of the hidden cracks is magnified using fluorescence imaging, and the fluorescence-excited development morphology and coordinate location information of the hidden cracks are extracted from the fluorescence-excited real-world model using image recognition technology based on a segmented attention model deep learning in the data processing unit of the ground remote control platform. Specifically, this includes the following steps:

[0021] First, the fluorescence-excited real-world image is input into the channel attention model to extract the attention features along the channel dimension. Then, the spatial attention model applies max pooling and average pooling operations to the input data along the channel dimension to achieve dimensionality reduction and feature fusion of the input features. Each element in the extracted fluorescence-excited real-world image corresponds to the attention weight information at its spatial location, thereby obtaining the fluorescence excitation development morphology and coordinate position information of the hidden cracks in the fluorescence-excited real-world image.

[0022] Preferably, in step S5, during the process of embedding the fluorescence-excited development pattern of the hidden crack into the non-fluorescent-excited real-world model according to the coordinate position information, if there are macroscopic cracks distributed at the location to be replaced in the non-fluorescent-excited real-world model, then the fluorescence-excited development crack to be replaced at that location is actually a macroscopic crack, and the fluorescence image is discarded and not replaced.

[0023] A second aspect of the present invention provides an apparatus for implementing the above-described cross-scale visual diagnostic method for crack risk zones in concrete structures, the apparatus comprising:

[0024] The detection drone is used to perform aerial image acquisition, which is divided into two stages. The first stage is to take pictures of the target area with crack risk, which is used to detect macroscopic cracks with a width of more than 0.1 mm in the target area with crack risk. The second stage is to take pictures of the fluorescent excitation area, which is used to detect hidden cracks with a width of less than 0.1 mm in the fluorescent excitation area.

[0025] The spraying drone is used to perform drone-borne liquid cleaning and drone-borne liquid stimulation. Specifically, the drone-borne liquid cleaning involves spraying clean water onto the target area at risk of cracking to clean the structural surface. Specifically, the drone-borne liquid stimulation involves spraying fluorescent stimulation solution onto the concentrated area of ​​macroscopic cracks and critical areas to stimulate fluorescence.

[0026] The ground-based remote control platform includes an unmanned aerial vehicle (UAV) control unit, a data processing unit, and a regional cracking risk assessment unit.

[0027] The UAV control unit is used to control the operating parameters of the liquid spraying UAV and the detection UAV;

[0028] The data processing unit is connected to the detection drone via wireless transmission technology. It is used to construct non-fluorescent excitation real-world models and fluorescent excitation real-world models respectively using 3D modeling software based on macroscopic images and fluorescence excitation images collected by the detection drone and transmitted wirelessly to the data processing center. It is also used to extract the fluorescence excitation development morphology and coordinate position information of the hidden cracks from the fluorescent excitation real-world model based on the magnified morphology of the hidden cracks after fluorescence development, combined with deep learning image recognition technology based on a segmented attention model. Then, the fluorescence excitation development morphology of the hidden cracks is embedded into the non-fluorescent excitation real-world model according to the coordinate position information to construct a cross-scale visualized 3D real-world model of the cracks.

[0029] The regional cracking risk assessment unit is used to evaluate the degree of cracking risk of concrete structures by region based on the comprehensive development law of macro cracks and hidden cracks in the cross-scale visualized 3D reality model of cracks.

[0030] The cross-scale visualization diagnostic method and apparatus for cracking risk zones in concrete structures provided by this invention have the following beneficial effects:

[0031] (1) This invention uses the crack risk zone as a unit to evaluate the crack risk of concrete structure, and incorporates hidden cracks into the crack risk evaluation. It combines the comprehensive development law of macro cracks and hidden cracks to evaluate the destructiveness of the crack risk zone to the concrete structure. Compared with evaluating the destructiveness of a single macro crack, the evaluation results are more reliable.

[0032] (2) This invention takes the development of hidden cracks with a width of less than 0.1 mm as an important basis for evaluating the degree of hazard of crack risk zone. It considers both the existing destructiveness of macro cracks and the potential subsequent destructiveness of hidden cracks. The gradient development pattern presented at the crack scale is used as the core indicator to characterize the degree of regional hazard. It systematically considers the hazard of crack risk zone to concrete structure from the perspective of dynamic development. Compared with the evaluation of the destructiveness of single macro cracks, it more effectively locks the source area of ​​cracking and damage of concrete structure.

[0033] (3) This invention utilizes the characteristic that the morphology of hidden cracks is magnified during fluorescence development, and combines image recognition technology based on segmented attention model deep learning to extract the fluorescence development morphology and location coordinates of hidden cracks from the fluorescence-excited real-world model (obtained by 3D real-world modeling using fluorescence-excited images in the fluorescence-excited region). It successfully embeds the fluorescence-excited development morphology of hidden cracks into the corresponding position in the non-fluorescence-excited real-world model (obtained by 3D real-world modeling using macroscopic images of the crack risk target area), thus obtaining a cross-scale visualized 3D real-world model of cracks. By utilizing the magnification of hidden cracks in their fluorescence development morphology, hidden cracks can be distinguished from macroscopic cracks with a width greater than 0.1 mm in the cross-scale visualized 3D real-world model of cracks through their fluorescence development morphology. Through the magnification effect of fluorescence morphology, cross-scale visualization of 3D real-world modeling of macroscopic cracks and hidden cracks is achieved, laying a data foundation for regional crack risk assessment. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 This is a schematic diagram illustrating the use of a spraying drone loaded with clean water to clean the structural surface of a target area at risk of cracking.

[0036] Figure 2 This is a schematic diagram illustrating the use of a detection drone to photograph macroscopic cracks wider than 0.1 mm in a target area at risk of cracking.

[0037] Figure 3 This is a schematic diagram illustrating the use of a liquid-spraying drone to spray a fluorescent excitation solution for fluorescence excitation.

[0038] Figure 4 This is a schematic diagram illustrating the use of a detection drone to capture fluorescence excitation images of hidden cracks less than 0.1 mm wide in the fluorescence excitation region.

[0039] Figure 5 A schematic diagram of a macroscopic crack detection route designed for a 3D real-scene model of a concrete dam;

[0040] Figure 6 To create a cross-scale visualization 3D real-world model image of cracks, which is obtained by embedding the fluorescence-excited development morphology of hidden cracks extracted from the fluorescence-excited real-world model into the non-fluorescence-excited real-world model according to the coordinate position information.

[0041] Figure 7 Images (1m) of macroscopic cracks and latent cracks with fluorescent (excitation) morphology in a 3D reality model for cross-scale visualization of cracks. 2 ). Detailed Implementation

[0042] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0043] The present application will be further described below with reference to the accompanying drawings and specific embodiments.

[0044] This invention provides a cross-scale visual diagnostic device for crack risk zones in concrete structures, comprising a spraying drone 1, a detection drone 2, and a ground-based remote control platform, wherein...

[0045] The liquid spraying drone 1 includes a drone carrier platform mounted on the liquid spraying drone. The carrier platform includes a liquid storage tank and spraying components. The liquid storage tank holds either clean water or (environmentally friendly) fluorescent excitation solution. The spraying components include a horizontal single spray bar and a horizontal alloy nozzle 4. The spraying state of the horizontal alloy nozzle 4 is adjustable, enabling full coverage from atomized spraying to water jet spraying. The liquid spraying drone 1 is used to perform drone-based liquid cleaning and drone-based liquid excitation. Specifically, the drone-based liquid cleaning involves spraying clean water onto the crack-risk target area using the drone 1 to clean the structural surface. Specifically, the drone-based liquid excitation involves spraying a fluorescent excitation solution onto the area to be fluorescently excited, i.e., the macroscopic crack concentration area and critical area detected by the detection drone, using the fluorescent excitation solution to excite fluorescence.

[0046] The detection drone 2 is equipped with a visible light camera 3 (with a resolution of no less than 20 million pixels) and an ultraviolet light source. The detection drone 2 is used to perform drone aerial image acquisition, which is divided into two stages. The first stage is to capture images of the crack risk target area to detect macroscopic cracks with a width of more than 0.1 mm in the crack risk target area. The second stage is to capture images of the fluorescence excitation area to detect hidden cracks with a width of less than 0.1 mm in the fluorescence excitation area.

[0047] The ground-based remote control platform includes an unmanned aerial vehicle (UAV) control unit, a data processing unit, and a regional cracking risk assessment unit.

[0048] The drone control unit is used to control the operating parameters of the liquid spraying drone 1 and the detection drone 2;

[0049] The data processing unit is connected to the detection drone 2 via wireless transmission technology. It is used to transmit macroscopic images and fluorescence-excited images from the detection drone 2 to the data processing center via wireless transmission. It then uses 3D modeling software to construct non-fluorescent-excited real-world models and fluorescence-excited real-world models, respectively. Based on the magnified morphology of the hidden cracks after fluorescence development, it uses deep learning image recognition technology based on a segmented attention model to extract the fluorescence-excited development morphology and coordinate position information of the hidden cracks from the fluorescence-excited real-world model. Then, it embeds the fluorescence-excited development morphology of the hidden cracks into the non-fluorescent-excited real-world model according to the coordinate position information to construct a cross-scale visualized 3D real-world model of the cracks.

[0050] The regional cracking risk assessment unit is used to evaluate the degree of cracking risk of concrete structures by region based on the comprehensive development law of macro cracks and hidden cracks in the cross-scale visualized 3D reality model of cracks.

[0051] The cross-scale visualization diagnostic method for cracking risk zones in concrete structures based on the above-mentioned device includes the following steps:

[0052] Step 1: Since there are usually attached substances on the surface of the actual concrete structure, in order to reduce the obstruction of the surface of the crack in the target area, the first step is to use a spraying drone 1 loaded with clean water to clean the structural surface of the target area at risk of cracking.

[0053] Specifically, refer to Figure 1 The drone 1, loaded with clean water, is driven by the drone control unit in the ground remote control platform to take off and reach a distance of 2 to 5 meters from the target area of ​​the concrete structure at risk of cracking. It sprays clean water onto the structural surface of the target area to clean it. After the target area is completely cleaned, the drone loaded with clean water returns.

[0054] Step 2: Use the detection drone 2 to collect macroscopic images of the target area at risk of cracking, and detect macroscopic cracks with a width of 0.1 mm or more in the target area;

[0055] Specifically, refer to Figure 2The detection drone 2, driven by the drone control unit in the ground remote control platform, takes off and enters RTK mode. It reaches a distance of 5-9 meters from the target area at risk of cracking in the concrete structure. A visible light camera 3 mounted on the detection drone 2 continuously photographs the target area, starting from the right apex. During photography, the detection drone 2 hovers. After taking a picture, the drone control unit drives the detection drone 2 to move horizontally to the left at the same distance from the target area, with each movement not exceeding 3 meters. After moving, the detection drone 2 hovers and photographs the target area. After completing a photograph at the left apex of the target area, the detection drone 2 descends vertically at the same distance from the target area. After descending, it continues to hover and photograph to the right of the target area, with each descent not exceeding 2 meters. When the detection drone 2 reaches the right boundary of the target area, it descends again. The detection drone 2 repeats this hovering and photographing process until the entire macroscopic image covers the target area. The collected macroscopic images are then transmitted in real-time to the data processing unit in the ground remote control platform via wireless transmission technology. The horizontal overlap between adjacent macroscopic images is 70%-90%, and the vertical overlap is 60%-80%.

[0056] Step 3: After the detection drone 2 completes the detection of macroscopic cracks wider than 0.1 mm in the target area, the spraying drone 1, loaded with fluorescent excitation solution, sprays the solution onto the concentrated macroscopic crack areas and critical areas identified by the detection drone 2 in the target area for fluorescence excitation. Then, the detection drone 2 collects fluorescence excitation images of the fluorescently excited areas to detect hidden cracks less than 0.1 mm wide marked by fluorescent morphology. Here, the present invention defines the concentrated macroscopic crack area as an area with no less than 4 macroscopic cracks wider than 0.1 mm per square meter in the crack risk target area; and the critical area is defined as the area controlling the structural safety of concrete components in the concentrated macroscopic crack areas and crack risk target areas identified by the detection drone 2. For example, when the concrete component is a dam, the drone photographs the area around the dam's spillway, the spillway gate pier, and the dam shoulder area; when the concrete component is a civil engineering beam, the drone photographs the mid-span area and the support area.

[0057] Specifically, refer to Figure 3-4The spraying drone 1, driven by the drone control unit in the ground remote control platform, takes off after loading the fluorescent excitation solution. It reaches a distance of 2-5 meters from the area to be excited (i.e., the concentrated macroscopic crack area and critical area), and sprays the fluorescent excitation solution to achieve full coverage. The spraying rate is controlled to ensure the fluorescent excitation solution flows freely along the surface of the excited area. 50 seconds after the spraying drone 1 begins spraying, the detection drone 2, driven by the drone control unit in the ground remote control platform, flies to a distance of 5-9 meters from the fluorescent excitation area. A visible light camera 3 mounted on the detection drone 2 continuously hovers and takes pictures from the area initially excited by the spraying drone 1 until the image completely covers the fluorescent excitation area. The obtained fluorescent excitation images are transmitted in real time to the data processing unit in the ground remote control platform via wireless transmission technology. The horizontal overlap between adjacent fluorescent excitation images is 70%-90%, and the vertical overlap is 60%-80%.

[0058] Step 4: The data processing unit in the ground remote control platform uses the received macroscopic images and fluorescence excitation images to perform three-dimensional real-scene modeling using three-dimensional modeling software, respectively, to obtain non-fluorescent excitation real-scene model and fluorescence excitation real-scene model;

[0059] Step 5: Utilizing the characteristic that the morphology of hidden cracks is magnified during fluorescence development, the deep learning image recognition technology based on the segmented attention model in the data processing unit of the ground remote control platform is used to extract the fluorescence-excited development morphology and coordinate position information of hidden cracks from the fluorescence-excited real scene model. Then, the fluorescence-excited development morphology of hidden cracks is embedded into the non-fluorescence-excited real scene model according to the coordinate position information to construct a cross-scale visualized three-dimensional real scene model of cracks.

[0060] Specifically, firstly, the fluorescence-excited real-world image is input into the channel attention model to extract the attention features along the channel dimension. Then, the spatial attention model applies max pooling and average pooling operations to the input data along the channel dimension to achieve dimensionality reduction and feature fusion of the input features. Each element in the extracted fluorescence-excited real-world image corresponds to the attention weight information at its spatial location. This allows the acquisition of the fluorescence excitation development morphology and coordinate position information of the hidden cracks in the fluorescence-excited real-world image. The fluorescence excitation development morphology of the hidden cracks is then embedded into the non-fluorescence-excited real-world model according to the coordinate position information. During this process, if macroscopic cracks are distributed at the location to be replaced in the non-fluorescence-excited real-world model, the fluorescence-developed crack to be replaced at that location is actually a macroscopic crack, and the fluorescence image is discarded without replacement.

[0061] Step 6: Using the regional cracking risk assessment unit in the ground-based remote control platform, the degree of cracking risk in the concrete structure is evaluated by region based on the comprehensive development pattern of macroscopic and latent cracks in the cross-scale visualized 3D reality model of cracks. The evaluation principles and corresponding levels are as follows:

[0062] a. High-risk cracking zone, with a total of more than 8 macroscopic and latent cracks per m. 2 Among them, hidden cracks account for no more than 50%, and the maximum width of macroscopic cracks is greater than 0.5 mm;

[0063] b. Moderate risk zone for cracking, with a total of more than 4 macroscopic and hidden cracks per m. 2 Among them, macroscopic cracks exist and account for no more than 50%, with the maximum width of macroscopic cracks greater than 0.3 mm; or the total number of macroscopic and hidden cracks is greater than 4 / m. 2 Among them, the proportion of hidden cracks shall not exceed 50%, and the maximum width of macro cracks shall be 0.3-0.5 mm; or the total number of macro cracks and hidden cracks shall be 5-8 per m. 2 Of these, hidden cracks account for no more than 50%, and the maximum crack width is greater than 0.5 mm;

[0064] c. Potential cracking risk zone: Only macroscopic cracks exist, with 2-4 macroscopic cracks per m. 2 The maximum crack width is greater than 0.3 mm; or the total number of macroscopic cracks and hidden cracks is greater than 4 / m. 2 Among them, macroscopic cracks account for no more than 50%, and the maximum width of macroscopic cracks is less than or equal to 0.3 mm.

[0065] Example 1

[0066] Using the method of this invention, a cross-scale visual diagnosis of cracking risk zone is performed on a concrete dam as an example.

[0067] Reference Figure 5 This diagram illustrates a macroscopic crack detection route designed using a 3D real-scene model of a concrete dam as an example (where point A is the starting point of aerial photography and point B is the ending point). The area enclosed by the black box in the diagram represents the critical area (multiple side-by-side spillways) where hidden cracks need to be detected. The fluorescent excitation development morphology of the hidden crack in the spillway (gate pier) at point A03, a critical area, is embedded into a non-fluorescent excitation real-scene model (obtained through 3D real-scene modeling using macroscopic images), thus constructing a cross-scale visualized 3D real-scene model of the cracks. Figure 6 As shown, based on the comprehensive development pattern of macroscopic and latent cracks in the constructed cross-scale visualization 3D real-world model of cracks, statistics are presented. Figure 6 The black border area in the middle (set to 1m) 2 The number of cracks in the data is shown in the figure. Figure 7 , Figure 7 The results showed that there were a total of 5 cracks in the area, of which cracks 1 to 4 were hidden cracks and crack 5 was a macroscopic crack. Therefore, the area can be diagnosed as a potential cracking risk zone.

[0068] This invention is not limited to the specific embodiments described above. Any modifications made by those skilled in the art based on the above concept without creative effort are within the scope of protection of this invention.

Claims

1. A cross-scale visual diagnostic method for cracking risk zones in concrete structures, characterized in that, Includes the following steps: S1. Use a spraying drone to carry clean water to clean the structural surfaces of the target area at risk of cracking; S2. Use a detection drone to collect macroscopic images of the target area at risk of cracking, and detect macroscopic cracks with a width of 0.1 mm or more in the target area; S3. After the drone to be detected completes the detection of macroscopic cracks with a width of more than 0.1 mm in the target area of ​​crack risk, the liquid spraying drone is used to spray the fluorescent excitation solution on the concentrated area and key area of ​​macroscopic cracks detected by the drone in the target area for fluorescent excitation. Then, the drone is used to collect the fluorescent excitation image in the fluorescent excitation area to detect hidden cracks with a width of less than 0.1 mm marked by fluorescent morphology. S4. The data processing unit in the ground remote control platform uses the obtained macroscopic image and fluorescence excitation image to perform three-dimensional real scene modeling, respectively, to obtain non-fluorescent excitation real scene model and fluorescence excitation real scene model; S5. Taking advantage of the characteristic that the morphology of hidden cracks is magnified during fluorescence development, the deep learning image recognition technology based on the segmented attention model in the data processing unit of the ground remote control platform is used to extract the fluorescence-excited development morphology and coordinate position information of hidden cracks from the fluorescence-excited real scene model. Then, the fluorescence-excited development morphology of hidden cracks is embedded into the non-fluorescence-excited real scene model according to the coordinate position information to construct a cross-scale visualized three-dimensional real scene model of cracks. S6. Using a ground-based remote control platform, the regional cracking risk assessment unit is employed to diagnose the degree of cracking risk in concrete structures based on the comprehensive development patterns of macroscopic and latent cracks in the cross-scale visualized 3D real-world model of cracks. The assessment principles and corresponding levels are as follows: a. High-risk cracking zone, with a total of more than 8 macroscopic and latent cracks per m. 2 Among them, hidden cracks account for no more than 50%, and the maximum width of macroscopic cracks is greater than 0.5 mm; b. Moderate risk zone for cracking, with a total of more than 4 macroscopic and hidden cracks per m. 2 Among them, macroscopic cracks exist and account for no more than 50%, with the maximum width of macroscopic cracks greater than 0.3 mm; or the total number of macroscopic and hidden cracks is greater than 4 / m. 2 Among them, the proportion of hidden cracks shall not exceed 50%, and the maximum width of macro cracks shall be 0.3-0.5 mm; or the total number of macro cracks and hidden cracks shall be 5-8 per m. 2 Of these, hidden cracks account for no more than 50%, and the maximum crack width is greater than 0.5 mm; c. Potential cracking risk zone: Only macroscopic cracks exist, with 2-4 macroscopic cracks per m. 2 The maximum crack width is greater than 0.3 mm; or the total number of macroscopic cracks and hidden cracks is greater than 4 / m. 2 Among them, macroscopic cracks account for no more than 50%, and the maximum width of macroscopic cracks is less than or equal to 0.3 mm.

2. The cross-scale visual diagnostic method for cracking risk zones in concrete structures according to claim 1, characterized in that, In step S2, the use of a detection drone to collect macroscopic images of the target area at risk of cracking, and to detect macroscopic cracks wider than 0.1 mm in the target area, specifically includes the following steps: The drone is driven to fly to the shooting position and continuously takes pictures of the target area starting from the right apex of the crack risk target area. While taking pictures, the drone hovers. After taking pictures, the drone moves horizontally to the left at the same distance from the target area, with each movement not exceeding 3 meters. After moving, the drone hovers and takes pictures of the target area. When the drone completes taking pictures at the left apex of the target area, it descends vertically at the same distance from the target area, with each descent not exceeding 2 meters. After descending, it continues to hover and take pictures to the right of the target area. When the drone reaches the right boundary of the target area, it descends again. The drone repeats this hovering and taking pictures until the entire target area is covered. The shooting position is 5 to 9 meters away from the crack risk target area.

3. The cross-scale visual diagnostic method for cracking risk zones in concrete structures according to claim 2, characterized in that, In step S3, the method of using a spraying drone to spray fluorescent excitation solution onto the macroscopic crack concentration areas and critical areas identified by the detection drone in the target area for fluorescence excitation, and then using the detection drone to collect fluorescence excitation images in the fluorescent excitation areas to detect hidden cracks less than 0.1 mm wide marked by fluorescent morphology, specifically includes the following steps: A spraying drone, loaded with fluorescent excitation solution, flies to the spraying location and sprays solution to fully cover the macroscopic crack concentration area and critical area identified by the detection drone in the target area. The spraying rate is controlled to ensure that the fluorescent excitation solution flows freely along the surface of the excited area. Then, after the spraying drone starts spraying solution from the first excited area, the detection drone flies to the shooting position and continuously hovers to take pictures of the fluorescent excited area. The detection drone is controlled to ensure that the excited area being photographed is within 30s to 60s after the fluorescent excitation, until the image completely covers the fluorescent excited area. The spraying location is 2 to 5m away from the macroscopic crack concentration area and critical area, and the shooting position is 5 to 9m away from the fluorescent excited area, which is the same distance as the detection drone from the target area in step S2.

4. The cross-scale visual diagnosis method for cracking risk zones in concrete structures according to claim 2 or 3, characterized in that, The horizontal overlap between adjacent macroscopic images obtained by the detection drone in continuous photography is 70%–90%, and the vertical overlap is 60%–80%; the horizontal overlap between adjacent fluorescent excitation images of hidden cracks is 70%–90%, and the vertical overlap is 60%–80%.

5. The cross-scale visual diagnostic method for cracking risk zones in concrete structures according to claim 1, characterized in that, In step S5, utilizing the characteristic that the morphology of hidden cracks is magnified during fluorescence development, the data processing unit of the ground remote control platform uses deep learning image recognition technology based on a segmented attention model to extract the fluorescence-excited development morphology and coordinate position information of hidden cracks from the fluorescence-excited real-world model. This specifically includes the following steps: First, the fluorescence-excited real-world image is input into the channel attention model to extract the attention features along the channel dimension. Then, the spatial attention model applies max pooling and average pooling operations to the input data along the channel dimension to achieve dimensionality reduction and feature fusion of the input features. Each element in the extracted fluorescence-excited real-world image corresponds to the attention weight information at its spatial location, thereby obtaining the fluorescence excitation development morphology and coordinate position information of the hidden cracks in the fluorescence-excited real-world image.

6. The cross-scale visual diagnosis method for cracking risk zones in concrete structures according to claim 1, characterized in that, In step S5, during the process of embedding the fluorescence imaging morphology of the hidden crack into the non-fluorescent excitation real-world model according to the coordinate position information, if there are macroscopic cracks distributed at the location to be replaced in the non-fluorescent excitation real-world model, then the fluorescence imaging crack to be replaced at that location is actually a macroscopic crack, and the fluorescence image is discarded and not replaced.

7. An apparatus for implementing the cross-scale visual diagnostic method for cracking risk zones in concrete structures as described in any one of claims 1-6, characterized in that, The device includes: The detection drone is used to perform aerial image acquisition, which is divided into two stages. The first stage is to take pictures of the target area with crack risk, which is used to detect macroscopic cracks with a width of more than 0.1 mm in the target area with crack risk. The second stage is to take pictures of the fluorescent excitation area, which is used to detect hidden cracks with a width of less than 0.1 mm in the fluorescent excitation area. The liquid spraying drone is used to perform drone-based liquid spraying cleaning and drone-based liquid spraying stimulation. Specifically, the drone-based liquid spraying cleaning involves spraying clean water onto the target area at risk of cracking to clean the structural surface. Specifically, the drone-based liquid spraying stimulation involves spraying a fluorescent stimulation solution onto the macroscopic crack concentration area and critical area identified by the detection drone in the target area to perform fluorescent stimulation. The ground-based remote control platform includes an unmanned aerial vehicle (UAV) control unit, a data processing unit, and a regional cracking risk assessment unit. The UAV control unit is used to control the operating parameters of the liquid spraying UAV and the detection UAV; The data processing unit is connected to the detection drone via wireless transmission technology. It is used to construct non-fluorescent excitation real-world models and fluorescent excitation real-world models respectively using 3D modeling software based on macroscopic images and fluorescence excitation images collected by the detection drone and transmitted wirelessly to the data processing center. It is also used to extract the fluorescence excitation development morphology and coordinate position information of the hidden cracks from the fluorescent excitation real-world model based on the magnified morphology of the hidden cracks after fluorescence development, combined with deep learning image recognition technology based on a segmented attention model. Then, the fluorescence excitation development morphology of the hidden cracks is embedded into the non-fluorescent excitation real-world model according to the coordinate position information to construct a cross-scale visualized 3D real-world model of the cracks. The regional cracking risk assessment unit is used to evaluate the degree of cracking risk of concrete structures by region based on the comprehensive development law of macro cracks and hidden cracks in the cross-scale visualized 3D reality model of cracks.

Images