Device and method for detecting cracks of vacuum tube of solar thermal collector

Abstract

The invention discloses a crack detection device and method for a vacuum tube of a solar thermal collector, and the device comprises a camera obscura, the interior of the camera obscura is fixedly connected with a plurality of groups of spacing plates at equal intervals, and the spacing plates are divided into a plurality of darkrooms in the camera obscura; the vacuum tube supporting mechanism comprises a plurality of supporting assemblies, each supporting assembly comprises a bearing table, three rotating assemblies are installed on each bearing table, a vacuum tube supporting opening is formed in the top end of each rotating assembly, a tube inlet is formed in the top wall of the camera obscura, and the tube inlets and the vacuum tube supporting openings are coaxially arranged; the image acquisition platform comprises a plurality of groups of acquisition units, each acquisition unit comprises a vertical displacement assembly installed in the darkroom, and an image acquisition device is installed on the vertical displacement assembly; a plurality of groups of positioning modules are arranged; the terminal system and the vertical displacement assembly are provided with position sensors. According to the method, the recognition and detection model containing the vacuum tube cracks is constructed based on the YOLOv8 network, and the cracks of the vacuum tube can be quickly recognized and detected.

Description

Technical Field

[0001] The invention relates to the technical field of solar vacuum tube product detection, and in particular to a solar collector vacuum tube crack detection device and method. Background Art

[0002] The vacuum tube water heater is mainly composed of a solar collector, a high-efficiency insulated water tank, and a water inlet and outlet pipe system. The solar collector is the key component of the vacuum tube water heater, which is generally composed of about 10 to 20 all-glass vacuum tubes, aluminum diffuse reflection plates, brackets, etc.

[0003] The all-glass vacuum tube is a new type of photothermal conversion device. Its structure is like a thermos bottle stretched to the size of a fluorescent tube. It has an inner glass tube with one end closed and a coaxial outer glass tube. The open ends of the inner and outer glass tubes are sealed together, and the interlayer between the inner and outer glass tubes is evacuated and sealed with a getter with a bracket. This getter with a bracket is used to support the closed round head of the inner glass tube. On the other hand, when the getter evaporates, it can be used to absorb the trace gas released by the vacuum tube during storage and operation. A selective absorption coating is also deposited on the outer surface of the inner glass tube using a special process. The absorption coating on the outer surface of the inner glass tube of the vacuum tube is used to absorb sunlight heat and reduce the radiation heat of the tube to a minimum. This absorption coating is a new type of sputtering selective absorption surface, with a solar absorption rate of α≥93% and a thermal emissivity of E≤6% at 100℃. The material of the all-glass vacuum collector tube is ultra-hard high-borosilicate glass 3.3, which can be used for a long time under various environmental conditions such as exposure, wind, rain, snow and hail (hail with a diameter less than or equal to 4 cm).

[0004] Solar collector vacuum tubes are key components in solar energy systems, and currently, cracks in vacuum tubes can only be identified manually. Therefore, there is an urgent need for a device and method for detecting and identifying cracks in solar collector vacuum tubes. Summary of the invention

[0005] The purpose of the present invention is to provide a solar collector vacuum tube crack detection device and method to solve the problems existing in the prior art.

[0006] To achieve the above object, the present invention provides the following solution: The present invention provides a solar collector vacuum tube crack detection device, comprising:

[0007] A dark box, wherein a plurality of partition plates are fixedly connected at equal intervals in the dark box, and the partition plates divide the dark box into a plurality of dark chambers;

[0008] A vacuum tube support mechanism, the vacuum tube support mechanism includes several groups of support components, the several groups of support components are respectively arranged in several darkrooms, the support components include a bearing platform, three groups of rotating components are installed on the bearing platform, a vacuum tube support port is installed on the top of the rotating component, a pipe inlet is opened on the top wall of the darkroom, the pipe inlet is coaxially arranged with the vacuum tube support port, the vacuum tube passes through the pipe inlet into the darkroom and is limitedly matched with the vacuum tube support port;

[0009] An image acquisition platform, the image acquisition platform comprising a plurality of acquisition units, the acquisition units comprising a vertical displacement assembly installed in the darkroom, an image acquisition device installed on the vertical displacement assembly, and the image acquisition device being arranged corresponding to the vacuum tube;

[0010] Positioning modules, the positioning modules are provided in a plurality of groups, the plurality of groups of positioning modules are respectively arranged on a plurality of groups of spacer plates, and the positioning modules are used to position the image acquisition device;

[0011] A terminal system, wherein the terminal system includes a Raspberry Pi, a display is fixedly connected to the top surface of the dark box, and the display, the positioning module, the image acquisition device, and the rotating assembly are all electrically connected to the terminal system;

[0012] Wherein, a position sensor is installed on the vertical displacement component.

[0013] According to the solar collector vacuum tube crack detection device provided by the present invention, the rotating assembly includes a rotating drive motor fixedly connected to the supporting platform, the output shaft of the rotating drive motor is fixedly connected to a worm gear, a worm is vertically rotatably connected to the supporting platform, the worm wheel and the worm are matched in transmission, and the top end of the worm is fixedly connected to the vacuum tube support.

[0014] According to the solar collector vacuum tube crack detection device provided by the present invention, the vertical displacement assembly includes a linear motor vertically arranged in the darkroom, a movable slide is installed on the linear motor, and the image acquisition device is installed on the movable slide.

[0015] According to the solar collector vacuum tube crack detection device provided by the present invention, the positioning module includes photoelectric sensors, and the photoelectric sensors are arranged in three groups at equal intervals on the partition plate.

[0016] According to the solar collector vacuum tube crack detection device provided by the present invention, a light source is installed on the partition plate.

[0017] A method for detecting cracks in a solar collector vacuum tube comprises the following steps:

[0018] S1, build an image acquisition platform for solar collector vacuum tubes, develop image acquisition software through Qt, and collect image information of vacuum tubes on the acquisition platform through the image acquisition software;

[0019] S2, preprocessing and data enhancement of the collected vacuum tube image information, and using Label Img to mark the vacuum tube image cracks and their location information to obtain a data set;

[0020] S3, trains the data set based on the YOLOv8 model, and identifies the cracks of the vacuum tube and detects the location information;

[0021] S4, saving the trained YOLOv8 model in torch script format, and using the trained model to detect cracks in the vacuum tube image to be identified;

[0022] S5, integrates the trained model into the software system developed based on Qt and deploys the software to the Raspberry Pi;

[0023] S6, the Raspberry Pi sends instructions to each actuator to identify the crack of the vacuum tube under test and display its location information.

[0024] According to the solar collector vacuum tube crack detection method provided by the present invention, in S1, developing image acquisition software through Qt specifically includes the following steps:

[0025] S11, using the C++ secondary development API of the camera of the image acquisition device to complete the camera call in the Qt development platform;

[0026] S12, using Qt's QLabel class to display the vacuum tube crack image and the detection results on the display screen of the Raspberry Pi;

[0027] S13, controlling the movement of the vertical displacement component and the rotation component, using software to control the output of the GPIO port of the Raspberry Pi, when starting to detect the vacuum tube, the Raspberry Pi controls the rotation component to drive the vacuum tube to rotate, thereby fully detecting the vacuum tube;

[0028] S14, configure the C++ version of Libtorch in the Qt development environment, and use Libtorch to call the trained model;

[0029] S15, build the software system under Release, package it and release it, and deploy it to the Raspberry Pi.

[0030] According to the solar collector vacuum tube crack detection method provided by the present invention, in S2, preprocessing the collected vacuum tube crack image information specifically includes the following steps:

[0031] S21, remove irrelevant content according to the coordinate range of the image, and Resize it to obtain 224×224 pixels compatible with the YOLOv8 network;

[0032] S22, use Label Img to manually annotate the vacuum tube crack images, convert the vacuum tube crack image data into the YOLO dataset format, and divide the dataset into three groups: training set, validation set and test set;

[0033] S23, performing data enhancement on the training set data, wherein the enhancement methods include random rotation and cropping to improve the target detection performance.

[0034] According to the solar collector vacuum tube crack detection method provided by the present invention, in S3, the specific method of training the data set based on the YOLOv8 model includes:

[0035] S31, uses the Backbone module for feature extraction and adopts a series of convolutional and deconvolutional layers, while using residual connections and bottleneck structures to reduce the size of the network and improve performance;

[0036] S32, uses the Neck module to perform multi-scale feature fusion, fuses feature maps from different stages of Backbone, and enhances feature representation capabilities;

[0037] S33, uses the Head module to perform the final target detection and classification tasks, and fuses the feature maps from different stages of Backbone to enhance the feature representation capability.

[0038] The present invention discloses the following technical effects:

[0039] The present invention uses an image acquisition device to collect images of cracks in the vacuum tube of the solar collector, and the camera can effectively obtain image information of the vacuum tube containing cracks; at the same time, the present invention constructs a vacuum tube crack recognition and detection model based on the YOLOv8 network, which can realize rapid recognition and detection of cracks in the vacuum tube;

[0040] The present invention performs image preprocessing on the collected image information of vacuum tube cracks, uses Label Img to manually annotate the vacuum tube crack images, and converts the vacuum tube crack image data into the YOLO data set format. The data set is divided into three groups: a training set, a validation set, and a test set. The training set data is enhanced, and the enhancement methods include random rotation, cropping, etc., to improve the target detection performance.

[0041] The present invention makes a fully automatic improvement on a traditional vacuum tube device for detecting a solar collector, utilizes a Raspberry Pi with a display screen to monitor the working conditions of all devices, and enables the detection device to be centrally controlled through the Raspberry Pi display screen. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0043] Figure 1 Axonometric measurement of the solar collector vacuum tube crack detection device of the present invention Figure I ;

[0044] Figure 2 Axonometric measurement of the solar collector vacuum tube crack detection device of the present invention Figure II ;

[0045] Figure 3 It is a schematic diagram of the internal structure of the supporting platform of the solar collector vacuum tube crack detection device of the present invention;

[0046] Figure 4 The present invention is a flow chart of the method for detecting cracks in a solar collector vacuum tube.

[0047] Among them, 1. dark box; 2. display screen; 3. image acquisition device; 4. linear motor; 5. moving slide; 6. vacuum tube; 7. light source; 8. photoelectric sensor; 9. support platform; 10. rotary drive motor; 11. vacuum tube support; 12. partition plate; 13. position sensor; 14. pipe inlet; 15. Raspberry Pi; 16. worm gear; 17. worm. DETAILED DESCRIPTION

[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0049] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0050] Reference Figure 1-Figure 4 The present invention provides a solar collector vacuum tube crack detection device, comprising:

[0051] A dark box 1, wherein a plurality of partition plates 12 are fixedly connected at equal intervals in the dark box 1, and the partition plates 12 divide the dark box 1 into a plurality of dark rooms;

[0052] The vacuum tube 6 support mechanism includes several groups of support components, which are respectively arranged in several darkrooms. The support components include a bearing platform 9, on which three groups of rotating components are installed. A vacuum tube support port 11 is installed on the top of the rotating component. A pipe inlet 14 is opened on the top wall of the darkroom 1. The pipe inlet 14 is coaxially arranged with the vacuum tube support port 11. The vacuum tube 6 penetrates into the darkroom 1 through the pipe inlet 14 and is limitedly matched with the vacuum tube support port 11.

[0053] An image acquisition platform, the image acquisition platform includes a plurality of acquisition units, the acquisition units include a vertical displacement assembly installed in a darkroom, an image acquisition device 3 is installed on the vertical displacement assembly, and the image acquisition device 3 is arranged corresponding to the vacuum tube 6;

[0054] Positioning modules, the positioning modules are provided in several groups, and the several groups of positioning modules are respectively arranged on several groups of spacer plates 12, and the positioning modules are used to position the image acquisition device 3;

[0055] The terminal system includes a Raspberry Pi 15, a display is fixedly connected to the top surface of the dark box 1, and the display, the positioning module, the image acquisition device 3, and the rotating component are all electrically connected to the terminal system;

[0056] Wherein, a position sensor 13 is installed on the vertical displacement component.

[0057] In this embodiment, the hardware configuration of the Raspberry Pi 15 is a CPU Intel Core i7-1210M 2.50 GHz processor, RAM (8G), and a Windows 64-bit operating system.

[0058] The dark box 1 adopts a rectangular stainless steel box body, and the stainless steel box body is sprayed with black paint.

[0059] The image acquisition device 3 is a camera, which adopts a H1605 CMOS camera with 8 million pixels;

[0060] A further optimized solution is that the rotating assembly includes a rotating drive motor 10 fixedly connected to the supporting platform 9, the output shaft of the rotating drive motor 10 is fixedly connected to a worm gear 16, a worm 17 is vertically rotatably connected to the supporting platform 9, the worm gear 16 and the worm 17 are matched with each other in transmission, and the top of the worm 17 is fixedly connected to a vacuum tube support 11.

[0061] According to a further optimized solution, the vertical displacement assembly includes a linear motor 4 vertically arranged in a darkroom, a movable slide 5 is mounted on the linear motor 4 , and the image acquisition device 3 is mounted on the movable slide 5 .

[0062] According to a further optimized solution, the positioning module includes photoelectric sensors 8 , and three groups of photoelectric sensors 8 are arranged at equal intervals on the spacer plate 12 .

[0063] According to a further optimized solution, a light source 7 is mounted on the partition plate 12 .

[0064] A method for detecting cracks in a solar collector vacuum tube comprises the following steps:

[0065] S1, building an image acquisition platform for the solar collector vacuum tube 6, developing image acquisition software through Qt, and collecting image information of the vacuum tube 6 on the acquisition platform through the image acquisition software;

[0066] S2, preprocessing and data enhancement of the collected image information of the vacuum tube 6, and using Label Img to mark the cracks and their position information in the image of the vacuum tube 6 to obtain a data set;

[0067] S3, training the data set based on the YOLOv8 model, and identifying the cracks of the vacuum tube 6 and detecting the position information;

[0068] S4, saving the trained YOLOv8 model in torch script format, and using the trained model to detect cracks in the image of the vacuum tube 6 to be identified;

[0069] S5, integrates the trained model into the software system developed based on Qt and deploys the software to the Raspberry Pi 15;

[0070] S6, the Raspberry Pi 15 sends instructions to each actuator to identify the crack of the vacuum tube 6 under test and display its position information.

[0071] To further optimize the solution, in S1, developing the image acquisition software using Qt specifically includes the following steps:

[0072] S11, using the C++ secondary development API of the camera of the image acquisition device 3 to complete the camera call in the Qt development platform;

[0073] S12, using QLabel class of Qt to display the crack image of vacuum tube 6 and the detection result on display screen 2 provided by Raspberry Pi 15;

[0074] S13, controlling the movement of the vertical displacement component and the rotation component, using software to control the output of the GPIO port of the Raspberry Pi 15, when starting to detect the vacuum tube 6, the Raspberry Pi 15 controls the rotation component to drive the vacuum tube 6 to rotate, and then comprehensively detects the vacuum tube 6;

[0075] S14, configure the C++ version of Libtorch in the Qt development environment, and use Libtorch to call the trained model;

[0076] S15, build the software system under Release, package it and release it, and deploy it to Raspberry Pi 15.

[0077] Further optimizing the solution, in S2, preprocessing the collected crack image information of the vacuum tube 6 specifically includes the following steps:

[0078] S21, remove irrelevant content according to the coordinate range of the image, and Resize it to obtain 224×224 pixels compatible with the YOLOv8 network;

[0079] S22, using Label Img to manually annotate the vacuum tube 6 crack image, and convert the vacuum tube 6 crack image data into the YOLO data set format, and divide the data set into three groups: training set, verification set and test set;

[0080] S23, performing data enhancement on the training set data, wherein the enhancement methods include random rotation and cropping to improve the target detection performance.

[0081] To further optimize the solution, in S3, the specific methods for training the data set based on the YOLOv8 model include:

[0082] S31, uses the Backbone module for feature extraction and adopts a series of convolutional and deconvolutional layers, while using residual connections and bottleneck structures to reduce the size of the network and improve performance;

[0083] S32, uses the Neck module to perform multi-scale feature fusion, fuses feature maps from different stages of Backbone, and enhances feature representation capabilities;

[0084] S33, uses the Head module to perform the final target detection and classification tasks, and fuses the feature maps from different stages of Backbone to enhance the feature representation capability.

[0085] The dark box 1 is evenly divided into three dark rooms by two partition plates 12. Each dark room is provided with a light source 7 and a linear motor 4. A movable slide 5 is provided on the linear motor 4 for placing the image acquisition device 3. The movable slide 5 and the image acquisition device 3 fixed thereon can be moved up and down along the movable slide 5 by the linear motor 4. The vacuum tube 6 is manually inserted vertically through the tube inlet 14 opened above the dark room until the bottom of the vacuum tube 6 contacts the vacuum tube support port 11. The vacuum tube support port 11 is an elastic material with a high friction coefficient and can clamp the vertically inserted vacuum tube 6. The rotary drive motor 10 transmits its power to the worm 17 connected to the bottom of the vacuum tube support port 11 through the worm gear 16, thereby controlling the rotation of the vacuum tube support port 11. When the vacuum tube 6 is inserted into the vacuum tube support port 11, image acquisition begins, and since the vacuum tube 6 is long, its image is acquired in three sections. At this time, the Raspberry Pi 15 sends a command to the linear motor 4 to move the movable slide 5 to the designated first position. At the same time, during this process, the position sensor 13 of the movable slide 5 continuously feeds back the position information of the movable slide 5 to the Raspberry Pi 15, so that the Raspberry Pi 15 can adjust and control the position of the movable slide 5. When the Raspberry Pi 15 receives a signal from the photoelectric sensor 8, it will send a command to stop the movable slide 5 from sliding, and at the same time send a command to the image acquisition device 3 to start collecting the first image of the vacuum tube 6. The collected image will be transmitted to the Raspberry Pi 15 via the USB data cable, and the display screen 2 connected to the Raspberry Pi 15 will display its detection result. Then the Raspberry Pi 15 sends a command to the linear motor 4 again to move the movable slide 5 to the designated second position, and repeat the above process until the third image of the vacuum tube 6 is collected. After that, the Raspberry Pi 15 will send a command to the linear motor 4 to move the movable slide 5 to the designated first position. At this time, the image of the side of the vacuum tube 6 facing the image acquisition device 3 has been collected. Then, the Raspberry Pi 15 will send a command to the rotary drive motor 10 to control the vacuum tube support 11 to rotate 180°, so that the back of the vacuum tube 6 faces the image acquisition device 3, thereby acquiring the image of the back of the vacuum tube 6. The specific steps are consistent with the image acquisition process of the front of the vacuum tube 6. The control interface and encoding interface of the linear motor 4 and its position sensor 13, the photoelectric sensor 8, the image acquisition device 3, and the rotary drive motor 10 are connected to the Raspberry Pi 15 with an integrated display screen 2, and the setting parameters can be controlled by the Raspberry Pi 15 integrated display screen 2; the Raspberry Pi 15 display screen 2 is placed on the dark box 1; the power supply of the image acquisition device 3, the power supply of the linear motor 4 and its sensor, the power supply of the rotary drive motor 10, the power supply of the light source 7, the power supply of the photoelectric sensor 8 and the power supply of the Raspberry Pi 15 are all connected in series, and then the device is powered by connecting an external power supply.

[0086] In the description of the present invention, it should be understood that the terms "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside" and "outside" etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0087] The above embodiments are only descriptions of the preferred modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A solar collector vacuum tube crack detection device, characterized in that: include: A dark box (1), wherein a plurality of groups of partition plates (12) are fixedly connected at equal intervals in the dark box (1), and the partition plates (12) divide the dark box (1) into a plurality of dark chambers; A vacuum tube (6) support mechanism, the vacuum tube (6) support mechanism comprises a plurality of support components, the plurality of support components are respectively arranged in a plurality of darkrooms, the support components comprise a bearing platform (9), three groups of rotating components are installed on the bearing platform (9), a vacuum tube support port (11) is installed at the top of the rotating component, a pipe inlet (14) is provided on the top wall of the darkroom (1), the pipe inlet (14) is coaxially arranged with the vacuum tube support port (11), the vacuum tube (6) penetrates into the darkroom (1) through the pipe inlet (14) and is limitedly matched with the vacuum tube support port (11); An image acquisition platform, the image acquisition platform comprising a plurality of acquisition units, the acquisition units comprising a vertical displacement assembly installed in the darkroom, an image acquisition device (3) being installed on the vertical displacement assembly, the image acquisition device (3) being arranged corresponding to the vacuum tube (6); Positioning modules, the positioning modules are provided in a plurality of groups, the plurality of groups of positioning modules are respectively arranged on a plurality of groups of spacer plates (12), and the positioning modules are used to position the image acquisition device (3); A terminal system, the terminal system comprising a Raspberry Pi (15), a display fixedly connected to the top surface of the dark box (1), the display, the positioning module, the image acquisition device (3), and the rotating assembly are all electrically connected to the terminal system; Wherein, a position sensor (13) is installed on the vertical displacement component.

2. A solar collector vacuum tube crack detection device according to claim 1, characterized in that: The rotating assembly comprises a rotating drive motor (10) fixedly connected to the supporting platform (9), the output shaft of the rotating drive motor (10) is fixedly connected to a worm gear (16), a worm (17) is vertically rotatably connected to the supporting platform (9), the worm gear (16) and the worm gear (17) are in transmission cooperation, and the top end of the worm gear (17) is fixedly connected to the vacuum tube support (11).

3. A solar collector vacuum tube crack detection device according to claim 1, characterized in that: The vertical displacement assembly comprises a linear motor (4) vertically arranged in the darkroom, a movable slide (5) is mounted on the linear motor (4), and the image acquisition device (3) is mounted on the movable slide (5).

4. A solar collector vacuum tube crack detection device according to claim 1, characterized in that: The positioning module comprises photoelectric sensors (8), and three groups of the photoelectric sensors (8) are arranged at equal intervals on the spacer plate (12).

5. A solar collector vacuum tube crack detection device according to claim 1, characterized in that: A light source (7) is mounted on the partition plate (12).

6. A method for detecting cracks in a solar collector vacuum tube, based on the solar collector vacuum tube crack detection device according to any one of claims 1 to 5, characterized in that: The steps include: S1, building an image acquisition platform for the solar collector vacuum tube (6), developing image acquisition software through Qt, and acquiring image information of the vacuum tube (6) on the acquisition platform through the image acquisition software; S2, preprocessing and data enhancement of the collected image information of the vacuum tube (6), and labeling the cracks and position information of the image of the vacuum tube (6) using LabelImg to obtain a data set; S3, training the data set based on the YOLOv8 model, and identifying the cracks of the vacuum tube (6) and detecting the position information; S4, saving the trained YOLOv8 model in torch script format, and using the trained model to detect cracks in the image of the vacuum tube (6) to be identified; S5, integrate the trained model into the software system developed based on Qt and deploy the software to Raspberry Pi (15); S6, the Raspberry Pi (15) sends instructions to each actuator to identify the crack of the vacuum tube (6) under test and display its position information.

7. A method for detecting cracks in a solar collector vacuum tube according to claim 6, characterized in that: In S1, developing image acquisition software using Qt specifically includes the following steps: S11, using the C++ secondary development API of the camera of the image acquisition device (3) to complete the camera call in the Qt development platform; S12, using Qt's QLabel class to display the crack image of the vacuum tube (6) and the detection result on the display screen (2) of the Raspberry Pi (15); S13, controlling the movement of the vertical displacement component and the rotation component, using software to control the output of the GPIO port of the Raspberry Pi (15), when starting to detect the vacuum tube (6), the Raspberry Pi (15) controls the rotation component to drive the vacuum tube (6) to rotate, thereby fully detecting the vacuum tube (6); S14, configure the C++ version of Libtorch in the Qt development environment, and use Libtorch to call the trained model; S15, build the software system under Release and package it for release, and deploy it to the Raspberry Pi (15).

8. A method for detecting cracks in a solar collector vacuum tube according to claim 6, characterized in that: In S2, preprocessing the collected crack image information of the vacuum tube (6) specifically includes the following steps: S21, remove irrelevant content according to the coordinate range of the image, and resize it to 224×224 pixels compatible with the YOLOv8 network; S22, using LabelImg to manually annotate the crack image of the vacuum tube (6), and convert the crack image data of the vacuum tube (6) into a YOLO data set format, and divide the data set into three groups: a training set, a validation set, and a test set; S23, performing data enhancement on the training set data, wherein the enhancement methods include random rotation and cropping to improve the target detection performance.

9. A method for detecting cracks in a solar collector vacuum tube according to claim 6, characterized in that: In S3, the specific methods for training the data set based on the YOLOv8 model include: S31, uses the Backbone module for feature extraction and adopts a series of convolutional and deconvolutional layers, while using residual connections and bottleneck structures to reduce the size of the network and improve performance; S32, uses the Neck module to perform multi-scale feature fusion, fuses feature maps from different stages of Backbone, and enhances feature representation capabilities; S33, uses the Head module to perform the final target detection and classification tasks, and fuses the feature maps from different stages of Backbone to enhance the feature representation capability.

Images