Device and method for detecting apparent state of bottom surface of bridge

Abstract

The invention relates to the technical field of bridge bottom appearance state detection, in particular to a bridge bottom surface appearance state detection device and method, which comprises a main vehicle body, a first imaging device, a second imaging device, a photoelectric rotary encoder, a data acquisition and control computer and a power supply, the first imaging device, the second imaging device, the photoelectric rotary encoder, the data acquisition and control computer and the power supply are all arranged on the main vehicle body, and the power supply supplies power to the first imaging device, the second imaging device, the photoelectric rotary encoder and the data acquisition and control computer; the first imaging device, the second imaging device and the photoelectric rotary encoder are respectively connected with the data acquisition and control computer; the photoelectric rotary encoder is connected with the first imaging device and the second imaging device. The device can realize non-contact detection of the appearance state of the bottom surface of the beam body, is convenient to carry and assemble, can accurately identify appearance diseases of the bottom surface of the beam body, and avoids subjective deviation existing in manual visual detection.

Description

Technical Field

[0001] The present invention relates to the technical field of bridge bottom appearance state detection, and in particular to a bridge bottom surface appearance state detection device and method. Background Art

[0002] As a comprehensive transportation facility, dual-use road-rail bridges have the advantages of saving land resources, low construction cost, and fast construction progress. They are widely used in railway and road transportation networks. Dual-use road-rail bridges mostly adopt a double-layer structure with layered road and rail traffic, usually with the lower layer being the railway and the upper layer being the road.

[0003] During the operation of the highway-railway dual-use bridge, with the deterioration of the structure and materials and the influence of traffic loads, the upper highway bridge beams may suffer from defects such as steel rust and chipping, concrete cracking and damage. Once they fall and collide with the high-speed train passing on the lower level, it will seriously affect the safety of passengers and cargo transportation. Therefore, the surface condition of the bottom surface of the highway beam above the railway must be strictly inspected in accordance with relevant regulations and specifications.

[0004] At present, daily and regular inspections of the surface condition of the bottom surface of the upper highway beam are usually carried out by manual visual inspection with the help of auxiliary facilities. However, since the upper highway beam is located above the electrified railway contact network, the use of auxiliary facilities for manual visual inspection requires power outage of the electrified railway contact network. At the same time, it must be completed during the idle period of railway transportation (window point), which is extremely difficult to implement, and the inspection efficiency is low, making it inconvenient to carry out on a regular basis. Summary of the invention

[0005] In view of the above problems, the first purpose of the present invention is to provide a device for detecting the apparent state of the bottom surface of a bridge, which can not only realize non-contact detection of the apparent state of the bottom surface of the upper highway beam of a highway-railway dual-purpose bridge, but also realize non-contact detection of the apparent state of the bottom surface of a single-layer highway or railway bridge with a road surface under the bridge; the device is easy to carry and assemble, can be quickly installed in the railway skylight point, relies on railway track operation, and automatically detects lighting, and its work efficiency is much higher than current detection methods; in addition, the device can obtain panoramic photos of the bottom surface of the bridge, and accurately quantify the development history of the disease through a disease recognition module, thereby avoiding the subjective bias existing in manual visual inspection.

[0006] The second object of the present invention is to provide a method for detecting the apparent state of the bottom surface of a bridge.

[0007] The first technical solution adopted by the present invention is: a detection device for the apparent state of the bottom surface of a bridge, comprising a main vehicle body, a first imaging device, a second imaging device, a photoelectric rotary encoder, a data acquisition and control computer and a power supply, wherein the first imaging device, the second imaging device, the photoelectric rotary encoder, the data acquisition and control computer and the power supply are all placed on the main vehicle body, and the power supply supplies power to the first imaging device, the second imaging device, the photoelectric rotary encoder and the data acquisition and control computer respectively; the first imaging device, the second imaging device and the photoelectric rotary encoder are respectively connected to the data acquisition and control computer; the photoelectric rotary encoder is respectively connected to the first imaging device and the second imaging device;

[0008] Among them, the main vehicle body includes a frame, a push rod, a running wheel, a trigger wheel, an imaging device bracket, a camera base and a support plate; a movable groove is arranged in the middle of the frame; the trigger wheel is located at the bottom of one side of the frame, and the photoelectric rotary encoder is installed on the trigger wheel; the running wheel is located at the bottom of the other side of the frame; the bottom of the push rod is installed on the frame, and the top of the push rod is provided with a support plate, and the support plate is used to place the display screen of the data acquisition and control computer; the imaging device bracket is arranged on both sides of the frame, the bottom of the imaging device bracket is provided with a bottom plate, and the upper end of the imaging device bracket is provided with an arc-shaped long hole; the bottom plate of the imaging device bracket can move horizontally along the movable groove; bolts are arranged on the camera base, and the bolts of the camera base are connected with the long holes of the imaging device bracket; the first imaging device and the second imaging device are respectively installed on two camera bases.

[0009] Preferably, a device for detecting the apparent state of the bottom surface of a bridge further comprises a single-point laser sensor, and the single-point laser sensor is respectively connected to a power supply and a data acquisition and control computer.

[0010] Preferably, an oblong hole is further provided in the middle of the frame, and after the bottom plate of the imaging device bracket is horizontally moved along the moving groove to a suitable position, it is connected and fixed to the oblong hole by bolts.

[0011] Preferably, the unilateral movement range of the imaging device bracket is 0 to 620 mm; the bolts of the camera base are connected to the arc-shaped long hole of the imaging device bracket, and the bolts of the camera base can move in the arc-shaped long hole of the imaging device bracket to achieve a rotation of plus or minus 65 degrees.

[0012] Preferably, the bottom of the push rod is connected to the frame by a pin.

[0013] The second technical solution adopted by the present invention is: a method for detecting the apparent state of the bottom surface of a bridge based on the detection device described in the first technical solution, comprising the following steps:

[0014] S100: The detection device described in the first technical solution is pushed under the bridge, the data acquisition and control computer controls the photoelectric rotary encoder to generate a pulse signal, the first imaging device and the second imaging device synchronously acquire images under the stimulation of the pulse signal, and send the acquired images to the data acquisition and control computer;

[0015] S200: The data acquisition and control computer automatically identifies the image and outputs the apparent defects at the bottom of the upper bridge.

[0016] Preferably, the data acquisition and control computer automatically identifies the image through a pre-trained defect recognition module and outputs the apparent defects at the bottom of the upper bridge.

[0017] Preferably, the disease recognition module is constructed and trained in the following manner:

[0018] S10: Construct training data set based on bridge bottom disease samples;

[0019] S20: performing image enhancement on the image data in the training data set;

[0020] S30: Input the image-enhanced training data set into the YOLOv5 model for training to obtain a disease recognition module.

[0021] Preferably, the step S200 further includes:

[0022] The data acquisition and control computer performs mileage positioning based on pulse signals, and collects longitudinal continuous images according to the initial mileage and pulse counting mileage positioning; and uses a single-point laser sensor to detect the bridge truss, divides the collected longitudinal continuous images based on the bridge truss, automatically identifies the divided images, and outputs the apparent defects at the bottom of the bridge.

[0023] Preferably, the apparent damage at the bottom of the upper bridge includes damage type, location and geometric attribute information.

[0024] Beneficial effects of the above technical solution:

[0025] (1) The present invention provides a device for detecting the apparent state of the bottom surface of a bridge, which can not only realize non-contact detection of the bottom surface of the upper highway beam of a highway-railway dual-purpose bridge, but also realize non-contact detection of the apparent state of the bottom surface of a single-layer highway or railway bridge with a road surface under the bridge; the device is easy to carry and assemble, can be quickly installed in the railway skylight point, relies on railway track operation, and automatically detects with lighting, and its work efficiency is much higher than that of current detection methods; in addition, the device can obtain panoramic photos of the bottom surface of the bridge, and accurately quantify the development history of the disease through a disease recognition module, thereby avoiding the subjective bias existing in manual visual detection.

[0026] (2) The present invention provides a device for detecting the surface condition of the bottom surface of a bridge, which adopts high-definition imaging technology to overcome the inconvenience and low efficiency of manual inspection, and realizes the acquisition of the surface condition image of the bottom of the highway bridge. At the same time, it realizes the recognition and extraction of typical surface defect features based on the defect recognition module.

[0027] (3) The detection device provided by the present invention is a trolley, which adopts a modular design, which is convenient for portable disassembly and transportation; the trolley is pushed on the railway track, collects images of a certain range of visual field directly above the running line, and realizes automatic identification of information such as mileage, location, and geometric attributes of the defect by training an intelligent recognition model; it provides an effective means for the inspection of the bottom of highway bridges of dual-use highway and railway bridges, solves the disadvantages of low efficiency and many missed inspections of manual inspections, greatly improves the inspection efficiency, and provides automated and intelligent inspection equipment and technical means for railway engineering sites; after modifying the running wheels, it can also be used for non-contact inspection of the apparent state of the bottom surface of a single-layer highway or railway bridge with a road surface under the bridge.

[0028] (4) The detection device provided by the present invention meets the requirements for detecting the apparent state of the bottom of a common highway-railway dual-purpose bridge. The bridge width is 24-36m, and the clearance between the upper and lower floors is 12-16m; its detection speed is 0-10km / h.

[0029] (5) The detection device provided by the present invention adopts a dual-channel imaging module to achieve high-definition imaging of the appearance status of the bottom of the bridge. Each channel camera can move horizontally by a certain displacement (0 to 620 mm) and rotate by a certain angle (plus or minus 65 degrees), that is, it has a rotation function to meet the coverage of the bottom view of the bridge with various structural forms and design sizes. The synchronous and equidistant continuous sampling of the dual-channel camera is achieved through the control of the photoelectric rotary encoder to ensure that the image resolution collected from different running tracks is consistent, and the multi-channel images can be spliced ​​into a complete bridge bottom image.

[0030] (6) The present invention realizes image positioning according to the initial mileage and the pulse counting mileage, and realizes image segmentation according to the detection of the truss by the single-point laser sensor; the present invention has the function of online detection of defects, realizes simultaneous detection and identification, and after the detection is completed, provides typical defect type, location and geometric attribute information.

[0031] (7) This detection device can intelligently identify cracks, peeling, damage and other types of defects on the painted surface of the steel structure at the bottom of the bridge and the surface of the concrete structure. The model structure is specially designed based on the general model, with high recognition speed and good recognition accuracy, and is innovative or unique. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 A block diagram of a device for detecting the surface state of a bridge bottom surface provided by one embodiment of the present invention;

[0033] Figure 2 A front view of a main vehicle body provided by an embodiment of the present invention;

[0034] Figure 3 A left side view of a main vehicle body provided by one embodiment of the present invention;

[0035] Figure 4 A top view of a main vehicle body provided for one embodiment of the present invention;

[0036] Figure 5 A is a schematic diagram of a pin provided in one embodiment of the present invention;

[0037] Figure 6 A schematic diagram B of a pin provided in one embodiment of the present invention;

[0038] Figure 7 A schematic diagram of the connection between an imaging device support and a camera base provided by an embodiment of the present invention;

[0039] Figure 8 A front view of an imaging device support provided by an embodiment of the present invention;

[0040] Fig. 9 A left side view of an imaging device support provided by one embodiment of the present invention;

[0041] Fig.10 A top view of an imaging device support provided by one embodiment of the present invention;

[0042] Fig.11 A front view of a camera base provided by one embodiment of the present invention;

[0043] Fig.12 A right side view of a camera base provided for one embodiment of the present invention;

[0044] Fig.13 A schematic diagram of a data acquisition and control computer for acquiring image data provided by an embodiment of the present invention;

[0045] Fig.14 A schematic diagram of the field of view of a dual-channel imaging device provided by an embodiment of the present invention (double-track railway);

[0046] Fig.15 A schematic diagram of the statistics of the number of defects and the distribution of the bounding boxes in a bridge bottom sample library provided by an embodiment of the present invention;

[0047] Fig.16 A schematic diagram of a confusion matrix of a bridge bottom disease verification set provided by an embodiment of the present invention;

[0048] Among them, 100-the first imaging device or the second imaging device; 1-the trigger wheel; 2-the frame, 21-the moving groove, 22-the oblong hole; the push rod 3; 4-the running wheel; 5-the imaging device bracket, 51-the bottom plate, 52-the arc-shaped oblong hole; 6-the camera base, 61-the bolt; 7-the support plate; 8-the pin shaft. DETAILED DESCRIPTION

[0049] The present invention is further described below through specific embodiments. It should be pointed out that, for ordinary technicians in this field, several modifications and improvements can be made without departing from the principle of the present invention, and these should also be regarded as belonging to the protection scope of the present invention.

[0050] The contents not described in detail in the specification of the present invention belong to the common knowledge of the professionals in this field.

[0051] Example 1

[0052] like Figure 1 As shown, the present invention discloses a detection device for the apparent state of the bottom surface of a bridge, the detection device is a trolley, including a main body, a first imaging device, a second imaging device 100, a photoelectric rotary encoder, a single-point laser sensor, a data acquisition and control computer and a power supply, the first imaging device, the second imaging device, the photoelectric rotary encoder, the single-point laser sensor, the data acquisition and control computer and the power supply are all placed on the main body, the power supply respectively supplies power to the first imaging device, the second imaging device, the photoelectric rotary encoder, the single-point laser sensor and the data acquisition and control computer; the first imaging device, the second imaging device, the photoelectric rotary encoder and the single-point laser sensor are respectively connected to the data acquisition and control computer; the photoelectric rotary encoder is connected to the first imaging device and the second imaging device, and the first imaging device and the second imaging device constitute a dual-channel imaging device.

[0053] Among them, Figure 2 , Figure 3 and Figure 4 As shown, the main vehicle body includes a frame 2, a push rod 3, a running wheel 4, a trigger wheel 1, an imaging device bracket 5, a camera base 6 and a support plate 7; a moving groove 21 and an oblong hole 22 are designed in the middle of the frame 2; the trigger wheel 1 is located at the bottom of one side of the frame 2, and a photoelectric rotary encoder is installed on the trigger wheel 1, and the photoelectric rotary encoder is used to provide a synchronous trigger signal and record the mileage of the trolley at the same time; the running wheel 4 is located at the bottom of the other side of the frame 2, and the running wheel 4 of the trolley can be disassembled and replaced by bolt connection to meet the needs of rail running and normal road running; the push rod 3 is used to push the trolley to move, the bottom of the push rod 3 is installed on the frame 2, and the top of the push rod 3 is provided with a support plate 7, and the support plate 7 is used to place the display screen of the data acquisition and control computer, and the bottom of the push rod and the frame 2 are connected as shown in the figure Figure 5 and Figure 6 (The units of the dimensions in the figure are all millimeters) The pin shaft 8 connection shown can be quickly disassembled and assembled on site without the need for tools.

[0054] like Figure 7 , Figure 8 , Fig. 9 and Fig.10 As shown, the bottom of the imaging device bracket 5 is provided with a bottom plate 51, and the upper end of the imaging device bracket 5 is provided with an arc-shaped long hole 52; the bottom plate 51 of the imaging device bracket 5 can be horizontally moved along the movable groove 21 to a suitable position, and then connected and fixed with the long circular hole 22 by bolts; the imaging device bracket 5 is preferably two, and the two imaging device brackets 5 are arranged on both sides of the frame 2 and can move on both sides; the unilateral movement range of the imaging device bracket 5 is 0 to 620mm; the power supply, electronic equipment, etc. placed on the frame can be adjusted according to the position of the imaging device bracket 5; as shown in FIG. Fig.11 and Fig.12 As shown, the camera base 6 is provided with a bolt 61, and the bolt 61 of the camera base 6 is connected to the arc-shaped long hole 52 of the imaging device bracket 5. The bolt 61 of the camera base 6 can move in the arc-shaped long hole 52 of the imaging device bracket 5 to achieve a rotation of plus or minus 65 degrees, that is, the camera base can rotate a certain angle (plus or minus 65 degrees) to achieve the function of rotating the camera angle. The first imaging device and the second imaging device 100 are respectively installed on the two camera bases 6.

[0055] Furthermore, in one embodiment, handles are welded on both sides of the frame 2 to facilitate carrying up and down the track; the imaging equipment brackets on both sides are detachable structures, and two people can disassemble or install the up and down track work within 10 minutes; the frame 2 is made of aluminum alloy material and is lightweight, and electronic equipment such as dual-channel imaging equipment, photoelectric rotary encoders, single-point laser sensors, data acquisition and control computers and power supplies can be placed on the frame; the trolley has the advantages of being intelligent, efficient, and labor-saving.

[0056] Since both road-rail dual-purpose bridges are designed based on comprehensive considerations of local environment, traffic efficiency, usage scenarios and other factors, the overall structural form, material, length, width, number and position of tracks are all different. Therefore, the present invention has made special designs in the image acquisition module and the trolley pushing mode to take into account the actual conditions of various bridge structures, pushing conditions and other actual conditions.

[0057] Furthermore, in one embodiment, if the running wheels are made of rubber, they can be used for non-contact detection of the apparent state of the bottom surface of a single-layer highway or railway bridge with a road surface under the bridge.

[0058] The railway tracks of the highway-railway dual-purpose bridge usually have 4 or 2 lines, so it is necessary to push the trolley on different lines to collect the bridge bottom image with a suitable field of view above the track at an orthographic angle as much as possible; combined with the characteristics of railway skylight operations, it is necessary to cover the detection of each line, and also consider the assembly and movement of the trolley, portability and promotion efficiency.

[0059] Example 2

[0060] The present invention discloses a method for detecting the apparent state of the bottom surface of a bridge, comprising the following steps:

[0061] S100: The detection device described in Example 1 is pushed under the bridge, the data acquisition and control computer controls the photoelectric rotary encoder to generate a pulse signal, the first imaging device and the second imaging device synchronously acquire images under the stimulation of the pulse signal, and send the acquired images to the data acquisition and control computer;

[0062] If it is a dual-use bridge for road and railway (the upper layer is a road and the lower layer is a railway), the detection device (i.e., a trolley) is pushed on the railway track below the bridge; if it is a single-layer road or railway bridge, the detection device (i.e., a trolley) is pushed on the road surface below the bridge; the image of the bottom surface of the bridge is collected by the first imaging device and the second imaging device 100;

[0063] like Fig.13 As shown, the data acquisition and control computer is connected to the dual-channel imaging device 100 and the photoelectric rotary encoder respectively; wherein, the data acquisition and control computer is a high-performance portable computer, and the data acquisition and control computer is used to control the photoelectric rotary encoder to generate a pulse signal, and the dual-channel imaging device 100 (i.e., a dual-channel color linear array camera) synchronously collects images under the stimulation of the pulse signal, and sends the collected images to the data acquisition and control computer; the data acquisition and control computer automatically identifies the collected images and outputs the apparent defects at the bottom of the upper bridge.

[0064] The dual-channel imaging device is used to collect images of the bottom surface of the upper beam of the highway-railway dual-purpose bridge. The dual-channel imaging device includes a first imaging device and a second imaging device 100. The first imaging device and the second imaging device 100 both use a high-resolution color industrial camera-high-brightness light source as core components. The dual-channel layout is used to realize image detection of a certain field of view directly above the running track. The field of view of the dual-channel imaging device is as follows: Fig.14 shown.

[0065] S200: the data acquisition and control computer automatically identifies the image and outputs the apparent defects at the bottom of the upper bridge, so as to detect the apparent state of the bottom surface of the bridge;

[0066] A defect recognition module is provided in the data acquisition and control computer. The defect recognition module automatically recognizes the defects at the bottom of the bridge based on the collected images and outputs the apparent defects at the bottom of the bridge to detect the apparent state of the bottom surface of the bridge. The apparent defects detected and identified include attribute information such as defect type, location, geometry, etc. Defect types include, for example, concrete cracks, spalling, water stains, and steel rust and cracking.

[0067] The Defect Recognition module analyzes the collected images and uses a deep convolution algorithm to identify different types of structural surface defects and defects. This function is crucial to improving the efficiency and accuracy of bridge inspections because it can significantly speed up the inspection process through automation technology, making it more efficient than traditional manual visual inspections.

[0068] The disease identification module is constructed in the following manner:

[0069] S10: Construct training data set based on bridge bottom disease samples;

[0070] In order to improve the accuracy of accurate detection and identification of apparent damage on the bottom of the bridge, a variety of disease sample sets were established. Cracks, whitening, water leakage, rust, rust spots, paint peeling, paint peeling, exposed reinforcement, water stains, voids, concrete peeling or damage, concrete mesh cracks and other categories were annotated through the large-scale data annotation tool CVAT to form a bridge bottom disease sample library, that is, the training data set.

[0071] The accumulated bridge bottom disease sample library contains 13 kinds of diseases. The number of diseases is unevenly distributed, ranging from 10 to 100 samples. The length-width ratio of the boundary box of the disease is also different. The distribution of the number of diseases and the distribution of the boundary box of the disease are as follows: Fig.15 As shown; the bar chart on the left shows in detail the distribution of different types of structural problems (such as cracks, rust, and water stains) in the training data set; different bars represent different defect types, among which "rust spots" have the largest number; these data are important inputs before model training and are used to adjust and optimize the performance of the model; Fig.15 The right figure in the middle also shows the distribution of all disease bounding boxes. Fig.15 It can be seen that the bounding boxes are distributed in different lengths and of various types, which provides guidance for adjusting the parameters of the model.

[0072] S20: performing image enhancement on the image data in the training data set;

[0073] In order to meet the requirements of the visual detection method, it is necessary to enhance the bridge bottom picture under low light to obtain a high-recognition, high-contrast picture to meet the downstream target detection task; the image enhancement algorithm process is shown in Table 1:

[0074] Table 1 Image enhancement algorithms

[0075]

[0076]

[0077] The image is enhanced by the above-mentioned image enhancement method to obtain an enhanced picture.

[0078] S30: Input the image-enhanced training data set into the YOLOv5 model for training to obtain a disease recognition module.

[0079] The basic architecture of the YOLOv5 model includes multiple convolutional layers, residual layers, and upsampling layers. In order to integrate the attention mechanism module, the attention mechanism module is inserted after the last convolutional layer of each residual block. This deployment strategy is based on the following assumptions: the output of the residual block contains rich contextual information. This method mainly strengthens important features and suppresses unimportant features through element-level adaptive modulation, thereby optimizing the network's expressiveness. The specific implementation algorithm is shown in Table 2.

[0080] Table 2 Attention mechanism module

[0081]

[0082]

[0083] The results of disease identification module are as follows: Fig.16 As shown by Fig.16 It can be seen that the disease recognition module has a good overall recognition effect on diseases.

[0084] Aiming at the defects of poor lighting environment at the bottom of highway-railway dual-use bridges, uneven distribution of defects, different characteristics of defects, large variations in defect sizes, and high false alarm and missed alarm rates of existing algorithms, the present invention proposes an image brightness adjustment algorithm, which is beneficial to target recognition and manual verification process of downstream tasks; combining the YOLOv5 model and a simple but effective attention module, the 3D attention weights in space and channels are considered without adding additional model parameters, so as to realize the accurate recognition of defects by the model.

[0085] Further, in one embodiment, if Figure 1 and Fig.13 As shown, the data acquisition and control computer is connected to the single-point laser sensor, and the data acquisition and control computer controls the single-point laser sensor to detect the position of the truss for image segmentation.

[0086] The data acquisition and control computer is also used for mileage positioning based on pulse signals, and positioning of the collected longitudinal continuous images according to the initial mileage and the pulse counting mileage; and using a single-point laser sensor to detect the bridge truss, and segmenting the collected longitudinal continuous images based on the bridge truss (there is a certain overlap rate between two adjacent images during segmentation, and there is no need to worry that the segmentation will separate the defects), and the segmented images are automatically identified using the defect recognition module to output the apparent defects at the bottom of the bridge; the present invention uses a method of first positioning, then segmenting and identifying the image, which is helpful for locating the defect image and can effectively improve the recognition efficiency and accuracy of the defect recognition module.

[0087] The present invention realizes the identification of typical defects such as cracks, peeling, rust, missing bolts, etc. on the bottom of the bridge through digital acquisition and control computers while collecting data; detection and identification can output attribute information such as defect type, location, geometry, etc.; it serves as a basis for staff to judge whether to carry out repairs and remediation; defect identification is mainly aimed at the bottom of steel structure bridges and concrete bridges. The bottom of steel structure bridges is usually painted with different colors, so the effectiveness of the defect identification model for multiple tasks should be emphasized.

[0088] The present invention is described in detail above in combination with specific implementation methods and exemplary examples. The above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments; the above description cannot be understood as limiting the present invention. Those skilled in the art understand that, without departing from the spirit and scope of the present invention, the technical solution of the present invention and its implementation methods may be subjected to a variety of equivalent substitutions, modifications or improvements, all of which fall within the scope of the present invention; the protection scope of the present invention shall be subject to the attached claims.

Claims

1. A device for detecting the apparent state of the bottom surface of a bridge, characterized in that: It includes a main vehicle body, a first imaging device, a second imaging device, a photoelectric rotary encoder, a data acquisition and control computer and a power supply. The first imaging device, the second imaging device, the photoelectric rotary encoder, the data acquisition and control computer and the power supply are all placed on the main vehicle body. The power supply supplies power to the first imaging device, the second imaging device, the photoelectric rotary encoder and the data acquisition and control computer respectively; the first imaging device, the second imaging device and the photoelectric rotary encoder are respectively connected to the data acquisition and control computer; the photoelectric rotary encoder is respectively connected to the first imaging device and the second imaging device; The main vehicle body comprises a vehicle frame (2), a push rod (3), a running wheel (4), a trigger wheel (1), an imaging device bracket (5), a camera base (6) and a support plate (7); a movable groove (21) is arranged in the middle of the vehicle frame (2); the trigger wheel (1) is located at the bottom of one side of the vehicle frame (2), and the photoelectric rotary encoder is installed on the trigger wheel (1); the running wheel (4) is located at the bottom of the other side of the vehicle frame (2); the bottom of the push rod (3) is installed on the vehicle frame (2), and the top of the push rod (3) is provided with a support plate (7), and the support plate (7) is used to place data acquisition and control calculation The imaging device support (5) is provided on both sides of the vehicle frame (2); a bottom plate (51) is provided at the bottom of the imaging device support (5); an arc-shaped long hole (52) is provided at the upper end of the imaging device support (5); the bottom plate (51) of the imaging device support (5) can move horizontally along the movable groove (21); a bolt (61) is provided on the camera base (6); the bolt (61) of the camera base (6) is connected to the long hole (52) of the imaging device support (5); the first imaging device and the second imaging device (100) are respectively mounted on the two camera bases (6).

2. The detection device according to claim 1, characterized in that: It also includes a single-point laser sensor, which is connected to a power supply and a data acquisition and control computer respectively.

3. The detection device according to claim 1, characterized in that: An oblong hole (22) is also provided in the middle of the vehicle frame (2); after the bottom plate (51) of the imaging device bracket (5) is horizontally moved along the moving groove (21) to a suitable position, it is connected and fixed to the oblong hole (22) by bolts.

4. The detection device according to claim 1, characterized in that: The unilateral movement range of the imaging device bracket (5) is 0 to 620 mm; the bolt (61) of the camera base (6) is connected to the arc-shaped long hole (52) of the imaging device bracket (5), and the bolt (61) of the camera base (6) can move in the arc-shaped long hole (52) of the imaging device bracket (5) to achieve a rotation of plus or minus 65 degrees.

5. The detection device according to claim 1, characterized in that: The bottom of the push rod is connected to the vehicle frame (2) by a pin shaft (8).

6. A method for detecting the apparent state of the bottom surface of a bridge, comprising the following steps: S100: Pushing the detection device according to any one of claims 1 to 5 under the bridge, the data acquisition and control computer controlling the photoelectric rotary encoder to generate a pulse signal, the first imaging device and the second imaging device synchronously acquiring images under the stimulation of the pulse signal, and sending the acquired images to the data acquisition and control computer; S200: The data acquisition and control computer automatically identifies the image and outputs the apparent defects at the bottom of the upper bridge.

7. The detection method according to claim 6, characterized in that: The data acquisition and control computer automatically identifies the image through a pre-trained defect recognition module and outputs the apparent defects at the bottom of the upper bridge.

8. The detection method according to claim 7, characterized in that: The disease recognition module is constructed and trained in the following manner: S10: Construct training data set based on bridge bottom disease samples; S20: performing image enhancement on the image data in the training data set; S30: Input the image-enhanced training data set into the YOLOv5 model for training to obtain a disease recognition module.

9. The detection method according to claim 6, characterized in that: The step S200 further includes: The data acquisition and control computer performs mileage positioning based on pulse signals, and collects longitudinal continuous images according to the initial mileage and pulse counting mileage positioning; and uses a single-point laser sensor to detect the bridge truss, divides the collected longitudinal continuous images based on the bridge truss, automatically identifies the divided images, and outputs the apparent defects at the bottom of the bridge.

10. The detection method according to claim 6, characterized in that: The apparent defects at the bottom of the upper bridge include defect type, location and geometric attribute information.

Images