Automatic alignment method of photovoltaic glass assembly and related device

Abstract

The invention provides an automatic alignment method of a photovoltaic glass assembly and a related device. The method comprises the following steps: collecting respective visual image information of a plurality of glass backboards in the production process of the photovoltaic glass assembly; collecting production environment information and equipment information in a scene where photovoltaic glass assembly production equipment is located; based on the visual image information, the production environment information and the equipment information, the photovoltaic glass assembly production equipment is automatically calibrated through an equipment calibration model; generating a position deviation correction parameter based on the visual image information; and driving the calibrated photovoltaic glass assembly production equipment to adjust the position relationship among the plurality of glass backboards according to the position deviation correction parameters so as to realize automatic alignment processing of the plurality of glass backboards. According to the embodiment of the invention, accurate alignment of the bus bar, the hole site and the glass boundary can be realized in the photovoltaic glass production process, and the equipment is dynamically calibrated according to the environmental change to maintain the high-precision alignment effect, so that the lamination efficiency and precision are improved.

Description

Technical Field

[0001] The present application relates to the technical field of curtain wall equipment, and in particular to an automatic alignment method for photovoltaic glass components and related devices. Background Art

[0002] With the rapid development of building-integrated photovoltaic technology, photovoltaic curtain walls, as an innovative technology that combines power generation functions with building exterior walls, have been widely used in the construction industry. Photovoltaic curtain walls not only have good power generation efficiency, but also improve the energy-saving effect and aesthetics of buildings. Therefore, they are gradually becoming the mainstream choice in modern green buildings.

[0003] In the manufacturing process of photovoltaic glass components, the precise docking of various components is crucial. These components include glass panels, adhesive films, solar cells and back glass, and their stacking accuracy directly affects the power generation efficiency, stability and overall service life of the photovoltaic curtain wall. However, in the current production process, especially the assembly and stacking links, most of them still rely on manual operation. This method is not only inefficient, but also easily affected by human errors, environmental factors and equipment status, resulting in positioning deviations and unstable quality.

[0004] Therefore, it is urgent to design a new technical solution to solve at least one of the above technical problems. Summary of the invention

[0005] In order to solve or improve the above-mentioned problems, the embodiments of the present application provide an automatic alignment method and related devices for photovoltaic glass assemblies, so as to achieve precise sheet stacking of photovoltaic glass assemblies and improve the production efficiency of photovoltaic glass assemblies.

[0006] In one embodiment of the present application, a method for automatically aligning a photovoltaic glass assembly is provided, wherein the photovoltaic glass assembly includes a plurality of glass back panels, and the method includes: In the photovoltaic glass assembly production process, visual image information of each of the plurality of glass back panels is collected; the visual image information at least includes: bus bar information and edge area information of the plurality of glass back panels; Collecting production environment information and equipment information in the scene where the photovoltaic glass assembly production equipment is located; the production environment information at least includes: environmental compensation information, temperature information, humidity information, vibration information; the equipment information at least includes: optical sensor position, robot arm path, lens focus calibration information; Based on the visual image information, the production environment information and the equipment information, automatically calibrate the photovoltaic glass assembly production equipment through an equipment calibration model; generating a position deviation correction parameter based on the visual image information; The calibrated photovoltaic glass assembly production equipment is driven to adjust the positional relationship between the multiple glass back panels according to the position deviation correction parameters to achieve automatic alignment of the multiple glass back panels.

[0007] In another embodiment of the present application, an automatic alignment device for a photovoltaic glass assembly is provided, wherein the photovoltaic glass assembly includes a plurality of glass back panels and at least includes the following units: A collection unit is used to collect visual image information of each of the plurality of glass back panels during the production process of the photovoltaic glass assembly; the visual image information at least includes: bus bar information and edge area information of the plurality of glass back panels; The acquisition unit is also used to collect production environment information and equipment information in the scene where the photovoltaic glass assembly production equipment is located; the production environment information at least includes: environmental compensation information, temperature information, humidity information, vibration information; the equipment information at least includes: optical sensor position, robot arm path, lens focus calibration information; A calibration unit, configured to automatically calibrate photovoltaic glass assembly production equipment through an equipment calibration model based on the visual image information, the production environment information, and the equipment information; A generating unit, configured to generate a position deviation correction parameter based on the visual image information; The adjustment unit is used to drive the calibrated photovoltaic glass assembly production equipment to adjust the position relationship between the multiple glass back panels according to the position deviation correction parameters to achieve automatic alignment of the multiple glass back panels.

[0008] In one embodiment of the present application, an electronic device includes: at least one processor, memory, and input-output unit; The memory is used to store a computer program, and the processor is used to call the computer program stored in the memory to execute the automatic alignment method of the photovoltaic glass assembly in the first embodiment.

[0009] In one embodiment of the present application, a computer-readable storage medium is provided, which includes instructions. When the instructions are executed on a computer, the computer executes the automatic alignment method of the photovoltaic glass assembly in the first embodiment described above.

[0010] In the technical solution provided by the embodiment of the present application, in the production process of photovoltaic glass components, the visual image information of each of the multiple glass backplanes is collected; the visual image information at least includes: the busbar information and edge area information of the multiple glass backplanes; the production environment information and equipment information in the scene where the photovoltaic glass component production equipment is located are collected; the production environment information at least includes: environmental compensation information, temperature information, humidity information, vibration information; the equipment information at least includes: optical sensor position, robot arm path, lens focus calibration information; based on the visual image information, the production environment information and the equipment information, the photovoltaic glass component production equipment is automatically calibrated through the equipment calibration model; based on the visual image information, the production environment information and the equipment information, the position deviation correction parameter is generated; the calibrated photovoltaic glass component production equipment is driven to adjust the position relationship between the multiple glass backplanes according to the position deviation correction parameter to realize the automatic alignment of the multiple glass backplanes. The embodiment of the present application can realize the precise alignment of the busbars, hole positions and glass boundaries in the photovoltaic glass production process, and dynamically calibrate the equipment according to environmental changes to maintain a high-precision alignment effect, thereby improving the efficiency and accuracy of the lamination. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0012] Figure 1 A schematic diagram of the structure of a photovoltaic glass assembly provided in one embodiment of the present application; Figure 2 A schematic diagram of the structure of a photovoltaic glass assembly production device provided in one embodiment of the present application; Figure 3 A schematic diagram of a process flow of an automatic alignment method for a photovoltaic glass assembly provided in one embodiment of the present application; Figure 4 A schematic diagram of the structure of an electronic device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0013] The present application provides the following embodiments to solve or improve at least some of the above problems. In order to enable people in the technical field to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0014] In some features described in the specification, claims and the above-mentioned drawings of this application, descriptions such as "first" and "second" are used to distinguish different components, parts, modules, devices, etc., and do not represent the order of precedence, nor do they limit "first" and "second" to be different types. In addition, the embodiments described below are only some embodiments of the present application, not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of this application.

[0015] In order to solve the above technical problems, an automatic alignment device for a photovoltaic glass assembly is provided in one embodiment of the present application. The automatic alignment device for a photovoltaic glass assembly includes the following units: A collection unit is used to collect visual image information of each of the plurality of glass back panels during the production process of the photovoltaic glass assembly; the visual image information at least includes: bus bar information and edge area information of the plurality of glass back panels; The acquisition unit is also used to collect production environment information and equipment information in the scene where the photovoltaic glass assembly production equipment is located; the production environment information at least includes: environmental compensation information, temperature information, humidity information, vibration information; the equipment information at least includes: optical sensor position, robot arm path, lens focus calibration information; A calibration unit, configured to automatically calibrate photovoltaic glass assembly production equipment through an equipment calibration model based on the visual image information, the production environment information, and the equipment information; A generating unit, configured to generate a position deviation correction parameter based on the visual image information; The adjustment unit is used to drive the calibrated photovoltaic glass assembly production equipment to adjust the position relationship between the multiple glass back panels according to the position deviation correction parameters to achieve automatic alignment of the multiple glass back panels.

[0016] In the embodiments of the present application, accurate alignment of bus bars, hole positions and glass boundaries can be achieved during the photovoltaic glass production process, and the equipment can be dynamically calibrated according to environmental changes to maintain a high-precision alignment effect, thereby improving the efficiency and accuracy of the lamination.

[0017] Figure 1 A schematic structural diagram of a photovoltaic glass assembly provided in one embodiment of the present application is shown. Figure 1 In the figure, the bus bar extends out of the hole 1, the back panel glass 2, the adhesive film 3, the battery panel 4, the bus bar 5, and the panel glass 6.

[0018] The panel glass 6 is located at the outermost layer and plays a role in protecting the internal structure. The panel glass usually has a high light transmittance, which allows as much sunlight as possible to pass through to ensure that the solar panel can receive enough light for photoelectric conversion. The busbar 5 is a conductive component used to connect the solar panel, which can collect the current generated by multiple solar cells to form a larger current output. The busbar is usually made of metal material and has good conductivity. The solar panel 4 is the core part of the photovoltaic glass assembly, which is composed of multiple solar cells. The solar cell can convert the energy of sunlight into electrical energy and generate direct current through the photoelectric effect. The performance of the solar panel directly determines the power generation efficiency of the photovoltaic module. The adhesive film 3 is located between the solar panel and the back panel glass, which plays a role in encapsulating and protecting the solar panel. The adhesive film usually has good adhesion, weather resistance and electrical insulation, which can prevent moisture, oxygen, etc. from entering the interior of the solar panel and extend the service life of the solar panel. The back panel glass 2 is located on the back of the photovoltaic module, opposite to the panel glass. The back panel glass also plays a role in protecting the internal structure and preventing the external environment from damaging the solar panel and the busbar. The busbar extension hole 1 is to allow the busbar to extend from the inside of the photovoltaic module to facilitate connection with the external circuit to achieve the output and transmission of electric energy. This structural design enables the photovoltaic glass module to effectively convert solar energy into electric energy, while having good durability and reliability, and is suitable for various solar power generation systems.

[0019] Figure 2 FIG. 1 is a schematic diagram showing the structure of a photovoltaic glass assembly production device provided by an embodiment of the present application. The automatic alignment device for photovoltaic glass assemblies described above can be installed in the photovoltaic glass assembly production device. Figure 2 As shown, an embodiment of the present application proposes a photovoltaic glass component production device, which aims to achieve precise alignment of bus bars, hole positions and glass boundaries, and dynamically calibrate the equipment according to environmental changes to maintain high-precision alignment effects, thereby improving the efficiency and accuracy of lamination. Figure 2 In the figure, a component 21 consisting of a high-definition camera, an infrared scanner, and a sensor, a bus bar extending hole 22, a back panel glass 23, an adhesive film 24, a battery panel 25, a panel glass 26, a bus bar 27, and a robot 28.

[0020] Specifically, Figure 2The structure of a photovoltaic glass assembly production equipment is shown. Its design purpose is to achieve precise alignment of various components in the photovoltaic glass assembly and to be able to perform dynamic calibration according to environmental changes to improve production efficiency and accuracy. The assembly 21 composed of a high-definition camera, an infrared scanner, and a sensor is used to collect visual image information and environmental information during the production process of the photovoltaic glass assembly. The high-definition camera and the infrared scanner can capture the bus bar information, hole position information, and edge area information of the glass backplane, while the sensor can monitor the temperature, humidity, vibration and other information in the production environment to provide data support for equipment calibration and position deviation correction. The bus bar extension hole 22 is a part of the photovoltaic glass assembly, which is used to allow the bus bar to extend from the inside of the assembly for external circuit connection. The backplane glass 23 is located on the back of the photovoltaic assembly, which plays a role in protecting the internal battery panel and bus bar and preventing them from being damaged by the external environment. The adhesive film 24 is located between the battery panel and the backplane glass, which plays a role in encapsulating and protecting the battery panel. It has good adhesion, weather resistance and electrical insulation, and can prevent moisture, oxygen, etc. from entering the battery panel, thereby extending the service life of the battery panel. The solar panel 25 is the core part of the photovoltaic glass assembly and is composed of multiple solar cells. Solar cells can convert the energy of sunlight into electrical energy and generate direct current through the photoelectric effect. The panel glass 26 is located in the outermost layer and plays a role in protecting the internal structure. The panel glass usually has a high light transmittance, which allows as much sunlight as possible to pass through to ensure that the solar panel can receive enough light for photoelectric conversion. The busbar 27 is a conductive component used to connect the solar panel. It can collect the current generated by multiple cells to form a larger current output. The busbar is usually made of metal material and has good conductivity. The manipulator 28 plays the role of operation and assembly in the production equipment. It can accurately adjust the position of each component according to the collected information and the calculated position deviation correction parameters to achieve precise alignment of the busbar, hole position and glass boundary.

[0021] Through the coordinated work of these components, the production equipment can achieve efficient and high-precision production of photovoltaic glass components, and can adapt to environmental changes and maintain the stability of production quality.

[0022] Exemplarily, a high-resolution camera and a laser scanning device are used to collect multi-angle images of photovoltaic glass. The annotated data include the location of the busbar, the hole position on the backplane glass, and the borders around the glass. When collecting data, ensure that it is carried out under different lighting, glass models and temperatures to ensure the diversity of training data and the robustness of the model. Use a deep learning framework (such as TensorFlow or PyTorch) to train CNN and R-CNN models. The model training process requires a large amount of annotated data to adjust the hyperparameters and improve the accuracy of the model. During the training process, data enhancement techniques such as image rotation, flipping, and cropping are used to improve the model's adaptability to different environmental changes. The recognition results are post-processed, and the coordinates of the busbar and the hole position are adjusted through an optimized edge detection algorithm to ensure that the deviation is less than the preset tolerance range (for example, within 1mm). Real-time optimization is performed through the prediction results of the model to ensure that the alignment accuracy of each piece of glass meets the requirements during the production process.

[0023] For example, in the photovoltaic glass production process, the equipment may be affected by changes in the external environment, hardware wear and error accumulation caused by long-term operation, resulting in a decrease in the alignment accuracy of the equipment. This will directly affect the positioning accuracy of the busbars and holes, and thus affect the quality of the photovoltaic glass. Therefore, an adaptive calibration system is needed to monitor the error in real time and automatically calibrate to ensure that the equipment can continue to maintain high-precision alignment.

[0024] In the embodiment of the present application, the goal of the adaptive calibration system is to optimize the accuracy of the equipment by continuously collecting data, comparing historical and real-time data, and performing automatic calibration. The image data of photovoltaic glass is collected in real time using a high-definition camera, and the image should contain features such as bus bars, hole positions, and boundaries of the glass. These image data are the basis for calibration, and the system calculates errors by analyzing feature points in the image (such as bus bars and hole positions). The system collects real-time data of the production environment through sensors (such as vibration sensors). Environmental factors may affect the accuracy of the equipment, so the collection of environmental data helps to optimize the calibration process. External environmental factors such as temperature, humidity, and mechanical vibration can affect equipment performance, so the system collects these data in real time to assist in error detection and calibration.

[0025] A method for historical data comparison and error analysis is also proposed in the embodiment of the present application. Specifically, the image data, environmental data, equipment parameters and other information collected each time will be stored in the database and managed through timestamps. During each calibration, the system will extract historical data from the database and compare it with the currently collected data. The system determines whether the device has shifted based on data comparison, and can detect any decrease in accuracy that may occur after the device has been running for a long time. The feature matching algorithm in computer vision is used to compare the matching degree of bus bars, hole positions, glass boundaries and other features in the historical data and the current image data. The core goal of error analysis is to calculate the coordinate deviation of feature points (such as bus bar endpoints and hole center points) in the image. Assuming that the coordinate difference of the same feature point in the historical data and the current image data exceeds a preset threshold (such as ±0.1mm), the calibration procedure is triggered. Through error analysis, the system can identify the type of equipment deviation. For example, optical sensor offset, robotic arm error, environmental factors, etc. In the embodiment of the present application, an automatic calibration execution method is also proposed. Specifically, according to the results of the error analysis, if the error is caused by the position of the optical sensor, the system will automatically adjust the position of the sensor. The adjustment method can be to accurately move the installation position of the camera through an electric control device. During the calibration process, the system calculates the adjustment angle or offset (usually less than 0.5mm) and drives the camera to accurately adjust to the correct position through the robotic arm. If the error is caused by the position deviation of the robotic arm, the system will use the inverse kinematics algorithm to calculate the adjustment path of the robotic arm. Assuming that there is a deviation in the position of the bus bar of the glass in the image, the system automatically adjusts the path of the robotic arm by calculating the direction and magnitude of the error. The adjustment path can be finely controlled by a PID controller to ensure the precise positioning of the robotic arm. If the error is related to image clarity or focus, the system will re-autofocus the lens and optimize the image quality. This is done by calculating the sharpness of the image or based on the depth information of the image. The calibration process may include adjusting parameters such as the focal length, exposure time, and white balance of the camera to ensure the accuracy of the image data. Based on the collected environmental data, the system can dynamically adjust the working parameters to adapt to different environmental changes. When temperature changes affect the size of the glass, the system will adjust the alignment parameters of the glass according to the temperature compensation model; when vibration causes a slight shift in the sensor position, the system reduces the impact of vibration by adjusting the image acquisition frequency and sensor sensitivity in real time.

[0026] Furthermore, after calibration, the accuracy of the adjusted equipment is verified by re-collecting image data. If the error still exists, the system will calibrate again. The calibration results will be fed back to the production monitoring system in real time, and the operator can view the adjusted equipment status and accuracy to ensure that the equipment is always in an ideal working state. According to the historical calibration data and error trend of the equipment, an adaptive optimization algorithm (such as gradient descent method, genetic algorithm) is used to dynamically adjust the calibration strategy. As the equipment operation time increases, the system will optimize the calibration process based on historical data to reduce unnecessary calibration operations.

[0027] In addition, automatic calibration can be performed regularly according to the equipment's operating cycle and environmental changes. For example, the calibration procedure can be automatically triggered after a certain amount of glass is produced to ensure that the equipment is always in optimal working condition. The calibration cycle can be dynamically adjusted according to production volume, equipment wear and environmental changes. The equipment accuracy is continuously monitored and real-time feedback is provided to the operator. If a drop in accuracy or abnormal equipment status is detected, the system will automatically remind the operator to perform maintenance or trigger calibration.

[0028] Optionally, image acquisition and environmental monitoring are performed through hardware devices such as high-resolution cameras and laser scanning devices. All hardware devices are connected to the control system through a data acquisition module to ensure the stability and accuracy of data transmission.

[0029] Compared with the prior art, the embodiment of the present application realizes the fully automatic high-precision lamination of photovoltaic glass components by introducing an intelligent alignment algorithm and an adaptive calibration system based on machine learning. The deep learning algorithm is used to identify and locate the features of the busbar and the extension hole, and combined with the glass boundary detection, the alignment accuracy of the multi-layer component is effectively improved, ensuring that the matching error between the busbar and the extension hole is controlled at the micron level, and significantly reducing the misalignment problem caused by manual operation. The introduction of an adaptive calibration mechanism can dynamically adjust the equipment parameters according to the changes in external factors such as ambient light, avoiding positioning deviations caused by long-term operation of the equipment or environmental fluctuations, thereby ensuring the consistency and stability of production quality. By replacing traditional manual operations with fully automated processes, production efficiency is greatly improved, production cycles are shortened, and the needs of large-scale and rapid production in the photovoltaic curtain wall industry are met. The new steps of the embodiment of the present application, including real-time feature capture, deep learning optimization and adaptive parameter adjustment, ensure the standardization of the production of each piece of photovoltaic glass, greatly reduce the defective rate, and improve product performance and market competitiveness. By replacing manual lamination operations with automation, not only the dependence on skilled workers is reduced, but also labor costs are reduced and economic benefits are improved.

[0030] In the embodiment of the present application, the automatic alignment device of the photovoltaic glass assembly can accurately determine the position of each component by collecting visual image information such as the busbar and edge area of ​​the glass backplane. The position deviation correction parameters generated on this basis can realize the precise adjustment of the positions of multiple glass backplanes, ensure the accurate alignment of key parts such as the busbar, thereby improving the electrical performance and overall quality of the photovoltaic glass assembly. Collect production environment information and equipment information, and perform automatic equipment calibration. Whether it is the temperature, humidity, vibration changes in the environment, or the optical sensor position, robot arm path and lens focus changes of the equipment itself, they can be effectively compensated to ensure that the production equipment can maintain high-precision operation under various conditions. From information collection, equipment calibration to position adjustment, the entire process is automated. Compared with the traditional manual alignment method, there is no need for manual repeated measurement and adjustment, which greatly shortens the production time of a single photovoltaic glass assembly and improves the overall production efficiency of the production line. Accurate automatic alignment can effectively avoid errors that may be caused by manual operation, reduce the number of defective products caused by inaccurate backplane position, improve the qualified rate of finished products, reduce production material waste, and thus reduce production costs.

[0031] It should be added here that the control logic of the control device will be introduced in detail in the method embodiment below.

[0032] An embodiment of the present application provides an automatic alignment method for photovoltaic glass assemblies. The method of this embodiment can be performed by the device in the above embodiment, which is implemented on the basis of the structure of the automatic alignment device embodiment of the photovoltaic glass assembly. Figure 3 As shown, the method may include the following steps: 101. In the production process of photovoltaic glass components, visual image information of each of a plurality of glass back panels is collected; the visual image information at least includes: bus bar information and edge area information of the plurality of glass back panels; 102. Collect production environment information and equipment information in the scene where the photovoltaic glass assembly production equipment is located; the production environment information at least includes: environmental compensation information, temperature information, humidity information, and vibration information; the equipment information at least includes: optical sensor position, robot arm path, and lens focus calibration information; 103. Automatically calibrate photovoltaic glass assembly production equipment through an equipment calibration model based on the visual image information, the production environment information, and the equipment information; 104. Generate a position deviation correction parameter based on the visual image information; 105. Drive the calibrated photovoltaic glass assembly production equipment to adjust the positional relationship between the multiple glass back panels according to the position deviation correction parameters to achieve automatic alignment of the multiple glass back panels.

[0033] Through the above steps 101 to 105, firstly, the visual image information of the glass backplane is collected, including the busbar and edge area information, so that the backplane state can be accurately grasped, which is helpful for the subsequent precise control of the position of each component, ensuring the structural integrity of the photovoltaic glass assembly and the reliability of the electrical connection, thereby improving the product quality. Secondly, the collection of production environment information and equipment information covers various key data related to environmental compensation, temperature and humidity, vibration, and equipment. Through these data, the production equipment can better adapt to different working environments, ensure that the equipment can still operate stably under complex and changing conditions, and reduce production errors and equipment failures caused by environmental factors. Automatic equipment calibration based on the comprehensive information collected above can effectively correct possible deviations in the equipment and ensure the high-precision operation of the production equipment. Furthermore, position deviation correction parameters are generated according to the visual image information, and the equipment is driven to perform automatic alignment processing, so as to achieve precise alignment of multiple glass backplanes. This automated alignment processing not only reduces the uncertainty and errors caused by manual operation, improves the accuracy of production and the product qualification rate, but also greatly improves production efficiency, shortens the production cycle, and helps enterprises gain an advantage in market competition.

[0034] As an optional embodiment, after the calibrated photovoltaic glass module production equipment is driven in 105 to adjust the positional relationship between the multiple glass back panels according to the position deviation correction parameters, the visual image information of each of the multiple glass back panels can be re-collected; the error value of the historical calibration parameter is obtained in combination with the re-collected visual image information; the error value at least includes the coordinate deviation value of the bus bar endpoint, the coordinate deviation value of the hole center point, and the coordinate deviation value of the edge area; based on the error value, the equipment calibration model is adaptively optimized to dynamically adjust the calibration strategy in the equipment calibration model.

[0035] In the production process of photovoltaic glass modules, after the positional relationship of the glass backplane is adjusted in step 105, the visual image information is re-collected. By comparing the re-collected image information with the image information based on the previous calibration, the error values ​​of the historical calibration parameters can be obtained. These error values ​​include the coordinate deviations of the busbar endpoints, the hole center points, and the edge areas. The principle of this step is to use the feedback mechanism to continuously obtain new data during the production process to evaluate the accuracy of the previous calibration operation. Adaptive optimization of the equipment calibration model based on these error values ​​means that the equipment calibration model can dynamically adjust the calibration strategy according to the deviations in actual production.

[0036] In this way, the calibration accuracy can be continuously improved through the above embodiments. Each time the calibration strategy is adjusted according to the newly appeared error, the production equipment can align the glass back plate more accurately in subsequent operations, reduce product defects caused by inaccurate calibration, improve product quality and production efficiency, and be able to adapt to various changes that may occur in the production process to ensure the stability and reliability of production.

[0037] As an optional embodiment, in 101, during the production process of the photovoltaic glass assembly, visual image information of each of the plurality of glass back panels is collected. The visual image information at least includes: bus bar information and edge area information of the plurality of glass back panels.

[0038] Here, visual image information of multiple glass backplanes, including busbar information and edge area information, is collected during the production process of photovoltaic glass modules. The glass backplane is photographed by professional image acquisition equipment to obtain its visual image. Busbar information and edge area information are key data in these images. The busbar is an important component for current collection and conduction in photovoltaic glass modules, and its position and state directly affect the electrical performance of the module. Edge area information is crucial for determining the size, shape and splicing relationship of the backplane.

[0039] Thus, the position of the busbar can be accurately obtained to ensure that the busbars on different glass backplanes can be accurately docked during the assembly process, avoiding poor electrical connection or failure due to busbar position deviation. The edge area information can help determine the actual size and shape of each glass backplane, ensuring that the backplanes can be tightly and accurately spliced ​​together during assembly, preventing gaps or overlaps caused by size mismatches, thereby ensuring the structural integrity and sealing of photovoltaic glass modules. These visual image information can serve as an important basis for quality inspection. By analyzing the status of the busbars and edge areas, defects in the backplane during the production process, such as damage to the busbars and breakage of the edges, can be discovered in a timely manner, which helps to improve the overall quality of the product.

[0040] As an optional embodiment, collecting visual image information of each of the plurality of glass back panels in 102 includes: The busbar in the first glass backplane is identified by a first positioning model to obtain first visual feature information of the busbar in the first glass backplane; the first positioning model is trained based on the busbar morphological features in the historical glass backplane image; The protruding hole in the second glass back plate is identified by a second positioning model to obtain second visual feature information of the protruding hole in the second glass back plate; the second positioning model is trained based on the morphological features of the protruding hole in the historical glass back plate image; The edge regions of the first glass back panel and the second glass back panel are detected by an edge detection model to obtain third visual feature information corresponding to the respective edge regions; the edge detection model is trained based on historical glass back panel images.

[0041] It can be understood that the first positioning model is trained based on the busbar morphological features in the historical glass backplane images. During the training process, the model learns the morphological features of the busbars through a large number of annotated historical busbar images, including the shape, width, length, color, texture, and distribution pattern of the busbars on the backplane. These features will be converted into weights and parameters within the model, so that the model can identify the busbars in new glass backplane images.

[0042] Thus, using the learned busbar morphological features, the model performs convolution and other operations on the new first glass backplane image, extracts the feature map in the image, and matches it with the busbar feature pattern stored in the model, thereby determining the position of the busbar and extracting its visual feature information. The busbar in the first glass backplane can be accurately identified, providing accurate busbar position information for subsequent alignment work. This helps to ensure that the busbars on different glass backplanes can be accurately docked, ensure the electrical connection performance of the photovoltaic glass assembly, and improve product quality and power generation efficiency.

[0043] The second positioning model is similar to the first positioning model and is also a deep learning model trained based on historical data. It learns the morphological features of the protruding holes in the historical glass backplane images and grasps the various features of the protruding holes, such as shape (circular, square, etc.), size, edge features, and positional relationship on the backplane. These feature information is encoded into the model parameters for the recognition of new images.

[0044] When faced with the image of the second glass backplane, the model searches for matching areas in the image based on the pre-learned morphological features of the protruding holes. Through a series of feature extraction and matching operations, the location of the protruding holes is found and their visual feature information is obtained.

[0045] Thus, the extension hole in the second glass back plate can be accurately positioned, providing a guarantee for accurate assembly during the production process. This is conducive to ensuring the correct fit of various components with the extension hole during assembly, such as accurate lead-out of the busbar through the extension hole, avoiding assembly difficulties or functional defects caused by inaccurate position of the extension hole.

[0046] The edge detection model is trained based on historical glass back images. It may use a combination of traditional edge detection algorithms and deep learning. For example, during the training process, the model learns the grayscale changes and texture changes of the glass back edge in the historical images, and by constructing a suitable loss function and adjusting the model parameters, the model can accurately detect the edge area.

[0047] Furthermore, when processing the images of the first glass backplane and the second glass backplane, the model identifies the edge area by calculating the difference, gradient and other information between the image pixels based on the learned edge feature pattern. It can distinguish between edge and non-edge areas, thereby extracting the third visual feature information corresponding to each edge area.

[0048] In this way, the edge area of ​​the glass backplane can be accurately detected, providing an important basis for the alignment of the glass backplanes. This helps to ensure that the edges of different backplanes can fit tightly during assembly, improve the structural stability and sealing of photovoltaic glass modules, and also provide accurate data support for operations such as size matching.

[0049] As an optional embodiment, in the above steps, it is assumed that the historical glass backplane image includes at least glass backplane images collected from multiple angles under multiple light source systems; the glass backplane has multiple material types, shape types, and size specifications; the illumination intensity and light source inclination angles of multiple light source systems are different. Based on this, before collecting the visual image information of each of the multiple glass backplanes, random adjustment parameters can also be used to perform corresponding modification operations on the historical glass backplane images to obtain new samples of the historical glass backplane images; wherein the random adjustment parameters are obtained by randomly scaling at least one of the bus bar annotation information, the protruding hole annotation information, and the edge area annotation information of the historical glass backplane images; the new samples are added to the set of historical glass backplane images to achieve data enhancement of the historical glass backplane images.

[0050] In the above steps, there are rich sources of historical glass back panel images, including images collected from multiple angles under a variety of light source systems. In addition, there are many types of glass back panels in terms of material, shape, size, etc., and the light intensity and tilt angle of the light source system also vary.

[0051] In principle, new samples are obtained by modifying the historical glass backplane images using random adjustment parameters. These random adjustment parameters are obtained by randomly scaling the annotation information of the busbars, protruding holes, and edge areas. This is because in the actual production environment and under different visual acquisition conditions, the appearance of these key areas may change due to various factors, such as different lighting, shooting angles, etc., which will cause them to appear differently in the image. By randomly scaling these annotation information, more possible actual situations can be simulated.

[0052] In terms of effect, data enhancement is achieved by adding new samples to the historical glass backplane image collection. Doing so can enrich the training data and make the data used to train the positioning model and edge detection model more diverse. When the model faces the complex and changeable glass backplane images in the real production environment, it can have better generalization ability and more accurately identify bus bars, protruding holes and edge areas. This helps to improve the accuracy and stability of the entire automatic alignment process, reduce recognition errors caused by data deviation or insufficiency, and thus improve the production quality and efficiency of photovoltaic glass components.

[0053] As an optional embodiment, in the above steps, it is assumed that the edge detection model is used to perform boundary reconstruction processing and boundary fitting processing on the edge area of ​​the glass back panel. Based on this, the edge detection model is used to detect the edge areas of the first glass back panel and the second glass back panel to obtain third visual feature information corresponding to the respective edge areas, including: Perform edge detection on the first glass back panel and the second glass back panel to obtain fourth visual image information of their respective edge areas; extract key visual features from the fourth visual image information; the key visual features are pixel feature points with color scale jumps or gray value jumps; the color scale jumps or gray value jumps indicate that the pixel feature points are located on the boundary between the glass back panel and the surrounding environment; the key visual features are fitted into the third visual feature information.

[0054] Specifically, the first glass backplane and the second glass backplane are firstly subjected to edge detection, which is realized based on the color level or grayscale change law of the pixels in the image. Since the glass backplane and the surrounding environment are different in material, color, etc., the pixels at the boundary will have obvious jumps in color level and grayscale, so the corresponding algorithm is used to capture the pixel feature points with color level jumps or grayscale value jumps. These points represent the edge area of ​​the glass backplane, and then the fourth visual image information is obtained. Then, the key visual features are extracted from these image information, that is, the pixel feature points that can accurately reflect the boundary, which carry key information such as the edge position. Finally, these key visual features are fitted to form the third visual feature information. This process is actually the integration and processing of the scattered feature points that can represent the edge, so that they can be used for more accurate backplane alignment and other operations in the future. In essence, it is to convert the boundary-related features intuitively reflected in the image into a more regular and more conducive to analysis and use of feature information.

[0055] Through such processing, the edge area of ​​the glass backplane can be determined more accurately. Compared with directly using the original image or rough edge detection results, the third visual feature information obtained can provide a very accurate basis for subsequent judgment of the positional relationship between the backplanes and accurate alignment. Moreover, this method based on pixel-level feature extraction and fitting enables better recognition of the edges of glass backplanes of different materials, shapes, and lighting conditions, improving the adaptability and accuracy of the entire automatic alignment method in complex actual production environments, helping to improve the quality and efficiency of photovoltaic glass module production and reduce assembly deviations caused by inaccurate positioning of edge areas.

[0056] For example, a rectangular glass back panel is placed in a production environment with certain lighting and a lighter background color. The glass back panel itself is darker in color. During edge detection, along the four sides of the glass back panel, the corresponding pixels on the image gradually transition from the dark pixels representing the glass back panel to the light pixels representing the surrounding environment. There is an obvious grayscale value jump in this transition area. The algorithm will detect these jumping pixel feature points. Obvious grayscale change points such as the four corners of the rectangle and the middle parts of the four sides will be extracted as key visual features. These scattered feature points are then fitted according to a certain mathematical model or algorithm, ultimately forming the third visual feature information that can accurately describe the edge area of ​​the glass back panel so that it can be aligned and assembled with other glass back panels in the future. The positional relationship can be accurately adjusted based on this feature information.

[0057] As an optional embodiment, in the above steps, generating position deviation correction parameters based on the visual image information includes: obtaining multidimensional deviation trend information between the first glass back panel and the second glass back panel through a deviation prediction model based on the first visual feature information, the second visual feature information, and the third visual feature information; obtaining the cumulative deviation value between the first glass back panel and the second glass back panel based on the multidimensional deviation trend information; and generating the position deviation correction parameters based on the cumulative deviation value.

[0058] In an embodiment of the present application, the multidimensional deviation trend information includes multiple types of visual deviation trends. Multidimensional deviation trend information is a kind of data information that comprehensively describes the deviations and their changing trends in multiple dimensions. In scenarios such as photovoltaic glass component production, it can be used to indicate the deviations of multiple key factors (such as position, size, angle, etc.) from the standard or expected state, as well as the changing trends of these deviations over time or production batches. Further optionally, the multiple types of visual deviation trends include: at least one of position deviation, size deviation, shape deviation, surface quality deviation, edge deviation, hole center point deviation, and busbar endpoint deviation.

[0059] Taking the glass backplane of a photovoltaic glass module as an example, during the automatic alignment process, there may be positional deviations between multiple glass backplanes in the plane. For example, in the x-axis and y-axis directions, the center position of the glass backplane deviates from the ideal alignment position, and these deviations may show a certain trend of change as the production process proceeds. If the production equipment gradually wears out or environmental factors have an impact, this positional deviation may become larger or show periodic changes, and the multi-dimensional deviation trend information will record the size and changes of these positional deviations.

[0060] The size of the glass back panel may also have deviations. For example, its length and width may be different from the designed size due to the manufacturing process. At the same time, this size deviation may also change during the production process. The multi-dimensional deviation trend information will record the dynamic changes of these size deviations, such as whether the size deviation is gradually stabilizing or whether it suddenly increases due to changes in raw material batches.

[0061] In photovoltaic glass modules, the angle alignment between the glass backplanes is also critical. For example, the two backplanes may need to maintain a strict parallel relationship, but angle deviations may occur in actual production. Moreover, with changes in the production process, such as the decrease in the operating accuracy of the robot arm or the vibration interference of the production environment, the angle deviation may show a trend of increase or fluctuation. The multi-dimensional deviation trend information will record these angle deviations and their changing trends.

[0062] By analyzing multidimensional deviation trend information, quality problems in the production process can be discovered in advance. For example, if it is found that the deviation trend of a certain dimension gradually exceeds the acceptable range, the production equipment or process parameters can be adjusted in time to reduce the production of defective products and improve product quality. Multidimensional deviation trend information can reflect the operating status of production equipment. If the change in deviation trend is caused by equipment failure or wear, then this information can help determine the best time and specific maintenance content for equipment maintenance and extend the service life of the equipment. It provides data support for process optimization. By observing the relationship between deviation trends in different dimensions and production process parameters, the direction of optimizing the process can be found, such as adjusting parameters such as temperature and humidity in the production environment, or improving the calibration model of the equipment to improve production stability and product consistency.

[0063] In the embodiment of the present application, the position deviation correction parameters include at least: deviation adjustment parameters of bus bar endpoints, deviation adjustment parameters of hole center points, and deviation adjustment parameters of edge areas.

[0064] The busbar plays a key role in collecting and transmitting current in photovoltaic glass modules. The position accuracy of the busbar endpoints directly affects the connection quality and current transmission efficiency of the internal circuit of the module. The deviation adjustment parameters of the busbar endpoints are used to measure and correct the deviation between the busbar endpoint position and the ideal position. During the production process of photovoltaic glass modules, when the collected visual image information shows that the busbar endpoint position is offset, the production equipment can accurately move the busbar endpoint to the correct position through the deviation adjustment parameters, ensuring that the busbar endpoints on adjacent glass backplanes can be accurately docked, thereby ensuring that the current can be smoothly transmitted between the glass backplanes, avoiding power loss or circuit failure caused by poor connection.

[0065] Holes are usually used for various connection or fixing purposes in photovoltaic glass modules, such as busbar extension holes. The positional accuracy of the center point of the hole is crucial to the correct installation and functional realization of the components. The deviation adjustment parameter of the center point of the hole is a parameter used to quantify and adjust the position deviation of the center point of the hole. If the busbar needs to extend through the hole to connect to the external circuit, the position of the center point of the hole must be accurate. When a position deviation of the center point of the hole is detected, the production equipment can use this deviation adjustment parameter to accurately adjust the position of the hole, so that the busbar can smoothly extend through the hole, and ensure that the position of the extended part meets the design requirements, ensuring the accuracy and stability of the subsequent connection with external equipment.

[0066] The edge area of ​​the glass backplane determines the splicing accuracy between the backplanes and the overall dimensional accuracy of the entire photovoltaic glass module. The deviation adjustment parameters of the edge area are used to correct the deviation of the edge position of the glass backplane to ensure that the backplanes can be tightly spliced ​​to form a photovoltaic glass module with complete structure and good sealing. During the assembly process, when it is found that the edge areas of the two glass backplanes cannot be well aligned, and there are gaps or overlaps, the production equipment can accurately adjust the position of the glass backplanes through the deviation adjustment parameters of the edge area so that their edges can fit perfectly. This can not only improve the appearance quality of the module, but also enhance the structural stability of the module, prevent impurities such as moisture and dust from entering the module, and extend the service life of the module.

[0067] In the above steps, the deviation prediction model is first used to analyze the first visual feature information (from the first glass backplane bus bar), the second visual feature information (from the second glass backplane extension hole) and the third visual feature information (from the edge area of ​​the two backplanes). Because these visual feature information of different dimensions reflects the state of the two backplanes at key positions, the deviation prediction model can obtain the multi-dimensional deviation trend information between the first glass backplane and the second glass backplane by learning and analyzing the association between these features and the difference from the standard state. This information covers the change trend of deviations in multiple key positions in multiple dimensions such as different directions. Then, based on this multi-dimensional deviation trend information, the cumulative deviation value is further calculated, that is, the total deviation accumulated by multi-dimensional deviations over time, production batches and other factors is comprehensively considered, and it quantifies and summarizes the multi-dimensional deviations. Finally, the position deviation correction parameters are generated based on the cumulative deviation values, in order to use these parameters to reversely adjust the position of the backplane to return it to an accurate state that meets the requirements.

[0068] Through such a set of processes, the deviation between the two glass backplanes can be grasped more accurately and comprehensively. It is no longer limited to the deviation consideration of a single dimension or a certain moment, but from the perspective of multiple dimensions and deviation change trends, so that the generated position deviation correction parameters are more scientific and accurate. When driving the production equipment to adjust the backplane position, it can effectively improve the alignment accuracy between the backplanes and reduce the problem of poor alignment caused by incomplete consideration of local deviations, thereby improving the overall production quality of photovoltaic glass components, ensuring its stable performance, and in the long-term production process, it can better adapt to various complex production situations, make product quality more stable, reduce defective rates, and improve production efficiency.

[0069] For example, when producing a batch of photovoltaic glass modules, the busbar of the first glass backplane is found to have its endpoint gradually shifted to the right in the x-axis direction in multiple image acquisition comparisons, and the center position of the extension hole also fluctuates up and down to a certain extent. The edge area is slightly tilted at a certain angle and this situation changes with production. The deviation prediction model analyzes these visual feature information from the busbar, extension hole, and edge area, and obtains multi-dimensional deviation trend information, such as the busbar endpoint may continue to shift to the right at a certain rate in the subsequent batches. Then, based on this trend, the cumulative deviation value is calculated. Assuming that the busbar endpoint has accumulated a rightward deviation of 5 mm, the center of the extension hole has accumulated a vertical deviation of 3 mm, etc., the corresponding position deviation correction parameters are generated based on these cumulative deviation values. The production equipment adjusts the position of the first glass backplane and the second glass backplane according to these parameters, so that the busbar endpoint, extension hole, and edge area can be accurately aligned to ensure the quality and function of the module.

[0070] Further optionally, the process of obtaining the position deviation correction parameter is expressed as the following formula:

[0071] in, is the deviation adjustment parameter used to calculate the busbar end points, Represents the batch index, Indicates busbar terminals, represents the busbar endpoint deviation feature dimension index, is a weight matrix, which is used for calculation in the linear mapping layer. Used for linear mapping layer to offset. Used to calculate the cumulative deviation value, It means The feature index corresponding to the busbar endpoint in the cumulative deviation value, Indicates the index value related to the cumulative deviation value. Indicates the range of the busbar endpoint deviation feature dimension index, from 1 to a constant D, where the constant D is the total number of busbar endpoint deviation feature dimensions. The value range of is the index range of the bus endpoints, from 1 to the total number of bus endpoints. The value range of represents the range of batch indexes, from 1 to the total number of batches.

[0072] This formula is used to calculate the deviation adjustment parameter of the bus bar endpoint. Based on an assumed linear mapping layer, the deviation adjustment parameter of the bus bar endpoint is generated by extracting characteristic information related to the bus bar endpoint from the accumulated deviation value.

[0073] Further optionally, the process of obtaining the position deviation correction parameter of the center of the extension hole is expressed as the following formula:

[0074] in, Used to calculate the deviation adjustment parameters of the center of the extended hole, represents the batch index, m represents the center of the mth protruding hole, and n represents the deviation feature dimension index of the protruding hole center. is a weight matrix, which is used for calculation in the linear mapping layer. Used for linear mapping layer to offset. Used to calculate the cumulative deviation value, It represents the feature index corresponding to the center of the m-th protruding hole in the cumulative deviation value, and p represents the index value related to the cumulative deviation value.

[0075] in, The range of the deviation feature dimension index of the protruding hole center is from 1 to the constant D, which is the total number of deviation feature dimensions of the protruding hole center. The value range of m, that is, the index range of the protruding hole center, is from 1 to the total number of protruding hole centers. The value range of n, that is, the range of batch index, is from 1 to the total number of batches.

[0076] This formula is used to calculate the deviation adjustment parameters of the hole center point. This formula is based on the linear mapping layer and generates the deviation adjustment parameters of the hole center point by accumulating the feature information related to the hole center point in the deviation value.

[0077] Further optionally, the process of obtaining the position deviation correction parameter of the edge area is expressed as the following formula:

[0078] in, The deviation adjustment parameter used to calculate the edge area, Represents the batch index, Indicates Edge area attention units, Indicates the deviation feature dimension index of the edge area, is a weight matrix, which is used for calculation in the linear mapping layer. Used for linear mapping layer to offset. Used to calculate the cumulative deviation value, It means The feature index corresponding to the edge region attention unit in the cumulative deviation value, Indicates the index value related to the cumulative deviation value.

[0079] in, The range of the deviation feature dimension index of the edge region is from 1 to a constant D, which is the total number of deviation feature dimensions of the edge region. The value range of m, that is, the index range of the edge region, is from 1 to the total number of attention units in the edge region. The value range of n, that is, the range of the batch index, is from 1 to the total number of batches.

[0080] This formula is used to calculate the deviation adjustment parameter of the edge area. This formula is based on the linear mapping layer and generates the deviation adjustment parameter of the edge area by accumulating the feature information related to the edge area in the deviation value.

[0081] As an optional embodiment, in 103, based on the visual image information, the production environment information and the equipment information, the photovoltaic glass assembly production equipment is automatically calibrated by using an equipment calibration model, including: Through the error analysis layer in the equipment calibration model, error analysis is performed on the visual image information, the production environment information and the equipment information to identify the equipment error type of photovoltaic glass component production equipment; each equipment error type is associated with a visual deviation trend in the visual image information; an error correction strategy is determined based on the identified equipment error type; the equipment error type includes at least one of the following: optical sensor error, robot arm path error, lens focus calibration error, environmental compensation; the error correction strategy is used to generate error correction parameters for automatically correcting photovoltaic glass component production equipment.

[0082] It is worth noting that the error analysis layer in the equipment calibration model will comprehensively process the visual image information, production environment information and equipment information. For the visual image information, it can analyze the possible visual deviation trends, such as busbar position deviation, hole position deviation, edge area deviation and other trends. Factors such as temperature, humidity, vibration in the production environment information, as well as the optical sensor position, robot arm path, lens focus calibration and other conditions in the equipment information will affect the final production results. The error analysis layer classifies the possible errors by analyzing these information and identifies the equipment error types of photovoltaic glass module production equipment. Because each equipment error type is associated with a visual deviation trend in the visual image, for example, the optical sensor error may cause inaccurate image acquisition, which is manifested as a certain deviation trend in the visual image. Once the equipment error type is determined, the error correction strategy can be determined based on the preset corresponding relationship, and then the error correction parameters can be generated to automatically correct the production equipment.

[0083] This automatic calibration method can effectively improve the accuracy and stability of production equipment. Through comprehensive analysis of multi-source information, it can comprehensively consider various factors affecting production and accurately identify problems with the equipment. Correction for different types of equipment errors can reduce production defects caused by equipment failure or deviation and improve the production quality of photovoltaic glass components. At the same time, the automated calibration process reduces manual intervention, improves production efficiency, and can adapt to changes in the production environment and equipment status in real time to ensure the continuity and stability of the production process.

[0084] Assume that during the production process, the collected visual image information shows that the busbar has a visual deviation trend of gradually shifting to the right during multiple acquisitions. The error analysis layer combines the production environment information (for example, the ambient temperature is gradually rising) and the equipment information (such as the optical sensor position data) for analysis. After analysis, it is found that the optical sensor error causes this visual deviation trend, which is the type of equipment error identified. Based on this optical sensor error type, the corresponding error correction strategy is determined, such as adjusting the calibration parameters of the optical sensor. Finally, the error correction parameters are generated and applied to the production equipment to automatically correct the optical sensor, thereby eliminating the deviation trend of the busbar shifting to the right in the visual image and ensuring the accuracy of photovoltaic glass module production.

[0085] As an optional embodiment, in 104, determining an error correction strategy based on the identified device error type includes: If the device error type is an optical sensor error, calculating an installation position offset and / or an installation angle offset of the optical sensor based on the visual image information, and configuring a first compensation parameter corresponding to the optical sensor based on the calculated installation position offset and / or installation angle offset; If the device error type is a robot arm path error, inverse kinematics calculation is performed on the visual image information, and the robot arm is driven to restore to the correct posture according to the calculated adjustment path; the adjustment path includes the movement direction, amplitude, and movement path required for the robot arm to return to the correct posture from the current posture; If the device error type is a lens focus calibration error, obtaining a lens adjustment parameter of the device according to the clarity or focus quality of the visual image information; the lens adjustment parameter includes at least focal length, exposure time, and / or white balance; If the equipment error type is environmental compensation, the environmental compensation model is used to predict the production environment change trend according to the production environment information, and the second compensation parameters of the photovoltaic glass component production equipment are generated based on the production environment change trend; the second compensation parameters include at least: temperature compensation parameters, humidity compensation, lighting condition compensation, and vibration compensation.

[0086] Specifically, when the device error type is identified as an optical sensor error, visual image information is used to determine whether there is an offset in the installation position and / or installation angle of the optical sensor. By analyzing the captured image, for example, observing the position and shape changes of a specific reference object in the image, the deviation of the optical sensor in the installation position or angle can be calculated. Based on these offsets, corresponding first compensation parameters can be further configured to correct the acquisition effect of the optical sensor.

[0087] Therefore, the position and angle deviation of the optical sensor can be accurately corrected to ensure that the collected visual image information is accurate and reliable. This helps to improve the accuracy of glass back panel feature recognition in the subsequent production process and avoid image distortion caused by optical sensor errors that affect production quality.

[0088] If the busbar image is always biased to one side of the image in the collected glass backplane image, and the problem of the backplane itself is ruled out after analysis, it can be inferred that the installation position of the optical sensor is offset. By calculating the offset, for example, determining that the sensor is offset by 2 mm in the horizontal direction and 1 mm in the vertical direction, and then configuring the corresponding first compensation parameter, the optical sensor can automatically correct this offset when collecting images in the subsequent process, thereby obtaining an accurate visual image.

[0089] For the robot path error, inverse kinematics calculation is performed on the visual image information. That is, based on the deviation between the current position and the ideal position of the glass backplane presented in the image, the motion parameters required for the robot to reach the correct position from the current position are inferred, including the motion direction, amplitude and motion path. Then the robot is driven to move according to the calculated adjustment path to restore it to the correct position.

[0090] Therefore, the path deviation of the robot arm during operation can be effectively corrected to ensure accurate handling and assembly of the glass backplane during production. This improves the accuracy and stability of the robot arm operation and reduces problems such as inaccurate alignment of the glass backplane caused by the robot arm path error.

[0091] For example, suppose that during production, the robot arm needs to place a glass back panel at a specific position, but the visual image shows that the placement of the back panel is deviated. By performing inverse kinematic calculations on the current deviation image, it is concluded that the robot arm needs to move 3 cm to the left horizontally and 1 cm downward vertically, and move along a specific motion path. In this way, the robot arm's adjustment path is configured, and the robot arm is driven to complete the correction so that the glass back panel is accurately placed.

[0092] When a lens focus calibration error occurs, the lens adjustment parameters are determined based on the clarity or focus quality of the visual image information. By analyzing the image blur, contrast and other focus quality indicators, the required lens adjustment parameters such as focal length, exposure time and white balance are calculated. In this way, the lens focus effect can be optimized, making the collected visual image clearer and more accurate. This helps to improve the recognition of subtle features on the glass backplane (such as busbar details, hole positions, etc.) and ensure accuracy during the production process.

[0093] If the image of the glass backplane is blurred, the image clarity and contrast are analyzed to find that the lens focal length is inaccurate. Further calculations show that the lens focal length needs to be increased by 5 mm, while the exposure time and white balance parameters need to be adjusted to obtain a clear and accurate visual image, ensuring that subsequent production operations are based on high-quality image data.

[0094] When the equipment error type is environmental compensation, the environmental compensation model is used to analyze the production environment information and predict the changing trend of the production environment. For example, by analyzing the monitoring data of environmental factors such as temperature, humidity, light and vibration, the change direction and amplitude of these factors are predicted, and then the corresponding second compensation parameters are generated. In this way, the photovoltaic glass component production equipment can adapt to environmental changes in advance, ensuring the stability and production quality of the production equipment under different environmental conditions. Reduce production deviations caused by changes in environmental factors and improve the adaptability of equipment to complex environments.

[0095] If the environmental monitoring data shows that the temperature of the production environment is gradually increasing and the humidity is decreasing, the environmental compensation model predicts that this trend of change will continue. Based on this, the second compensation parameters including temperature compensation parameters (such as adjusting the equipment cooling system), humidity compensation parameters (such as adjusting the film drying process), light condition compensation parameters (such as adjusting the sensitivity of the optical sensor) and vibration compensation parameters (such as adjusting the equipment's shock absorption system) are generated to ensure that the production equipment can still operate normally and accurately during environmental changes.

[0096] In the embodiment of the present application, during the production process of photovoltaic glass components, visual image information of each of the multiple glass backplanes is collected; the visual image information at least includes: busbar information and edge area information of the multiple glass backplanes; the production environment information and equipment information in the scene where the photovoltaic glass component production equipment is located are collected; the production environment information at least includes: environmental compensation information, temperature information, humidity information, vibration information; the equipment information at least includes: optical sensor position, robot arm path, lens focus calibration information; based on the visual image information, the production environment information and the equipment information, the photovoltaic glass component production equipment is automatically calibrated through the equipment calibration model; based on the visual image information, the production environment information and the equipment information, the position deviation correction parameter is generated; the calibrated photovoltaic glass component production equipment is driven to adjust the position relationship between the multiple glass backplanes according to the position deviation correction parameter to realize the automatic alignment of the multiple glass backplanes. In the embodiment of the present application, the busbars, hole positions and glass boundaries can be accurately aligned during the production process of photovoltaic glass, and the equipment can be dynamically calibrated according to environmental changes to maintain a high-precision alignment effect, thereby improving the efficiency and accuracy of lamination.

[0097] Based on the same implementation principle, an embodiment of the present application provides an automatic alignment device for a photovoltaic glass assembly, wherein the photovoltaic glass assembly includes a plurality of glass back panels, and at least includes the following units: A collection unit is used to collect visual image information of each of the plurality of glass back panels during the production process of the photovoltaic glass assembly; the visual image information at least includes: bus bar information and edge area information of the plurality of glass back panels; The acquisition unit is also used to collect production environment information and equipment information in the scene where the photovoltaic glass assembly production equipment is located; the production environment information at least includes: environmental compensation information, temperature information, humidity information, vibration information; the equipment information at least includes: optical sensor position, robot arm path, lens focus calibration information; A calibration unit, configured to automatically calibrate photovoltaic glass assembly production equipment through an equipment calibration model based on the visual image information, the production environment information, and the equipment information; A generating unit, configured to generate a position deviation correction parameter based on the visual image information; The adjustment unit is used to drive the calibrated photovoltaic glass assembly production equipment to adjust the position relationship between the multiple glass back panels according to the position deviation correction parameters to achieve automatic alignment of the multiple glass back panels.

[0098] The specific implementation of each unit in the above device is referred to the above embodiments and will not be expanded here.

[0099] Based on the same implementation principle, the embodiment of the present application also provides an electronic device, Figure 4is a structural block diagram of the electronic device 300, such as Figure 4 As shown, the electronic device 300 includes: a processor 301, a memory 302, a communication interface 303, a communication bus 304 and a controller 305; wherein the processor 301, the memory 302, and the communication interface 303 communicate with each other via the communication bus 304; the memory 302 is used to store computer programs; the processor 301 is used to execute the programs stored in the memory 302 to realize the corresponding processing functions; the communication bus 304 is used for communication between the electronic device 300 and other devices; the controller 305 is used to realize the automatic alignment method of the photovoltaic glass assembly described in the above method embodiment.

[0100] In the embodiment of the present application, the communication bus 304 may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The communication bus 304 may be divided into an address bus, a data bus, a control bus, etc., and the specific form is not limited. For ease of representation, Figure 4 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0101] The memory 302 may include a random access memory (RAM) or a non-volatile memory (non-volatile memory), such as at least one disk memory. Optionally, the memory may also be at least one storage device located far away from the processor 301 .

[0102] The processor 301 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other editable logic devices, discrete gate or transistor logic devices, discrete hardware components, and the specific form can be determined according to actual needs.

[0103] Accordingly, an embodiment of the present application further provides a computer-readable storage medium storing a computer program, which, when executed, can implement each step that can be executed by an electronic device in the above method embodiment.

[0104] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for automatically aligning a photovoltaic glass assembly, characterized in that: The photovoltaic glass assembly includes a plurality of glass back panels, and the method includes: In the photovoltaic glass assembly production process, visual image information of each of the plurality of glass back panels is collected; the visual image information at least includes: bus bar information and edge area information of the plurality of glass back panels; Collecting production environment information and equipment information in the scene where the photovoltaic glass assembly production equipment is located; the production environment information at least includes: environmental compensation information, temperature information, humidity information, vibration information; the equipment information at least includes: optical sensor position, robot arm path, lens focus calibration information; Based on the visual image information, the production environment information and the equipment information, automatically calibrate the photovoltaic glass assembly production equipment through an equipment calibration model; generating a position deviation correction parameter based on the visual image information; The calibrated photovoltaic glass assembly production equipment is driven to adjust the positional relationship between the multiple glass back panels according to the position deviation correction parameters to achieve automatic alignment of the multiple glass back panels.

2. The method according to claim 1, characterized in that: After the calibrated photovoltaic glass assembly production equipment is driven to adjust the positional relationship between the plurality of glass back panels according to the position deviation correction parameter, the method further comprises: Re-collecting the visual image information of the multiple glass back panels; In combination with the re-collected visual image information, the error values ​​of the historical calibration parameters are obtained; the error values ​​at least include the coordinate deviation values ​​of the busbar endpoints, the coordinate deviation values ​​of the hole center points, and the coordinate deviation values ​​of the edge areas; The device calibration model is adaptively optimized based on the error value to dynamically adjust the calibration strategy in the device calibration model.

3. The method according to claim 1, characterized in that The collecting of visual image information of the plurality of glass back panels includes: The busbar in the first glass backplane is identified by a first positioning model to obtain first visual feature information of the busbar in the first glass backplane; the first positioning model is trained based on the busbar morphological features in the historical glass backplane image; The protruding hole in the second glass back plate is identified by a second positioning model to obtain second visual feature information of the protruding hole in the second glass back plate; the second positioning model is trained based on the morphological features of the protruding hole in the historical glass back plate image; The edge regions of the first glass back panel and the second glass back panel are detected by an edge detection model to obtain third visual feature information corresponding to the respective edge regions; the edge detection model is trained based on historical glass back panel images.

4. The method according to claim 3, characterized in that The historical glass back panel images at least include glass back panel images collected from multiple angles under multiple light source systems; the glass back panel has various material types, shapes, and sizes; the illumination intensities and light source inclination angles of multiple light source systems are different; Before collecting the visual image information of the plurality of glass back panels, the method further includes: Using random adjustment parameters, performing corresponding modification operations on the historical glass backplane image to obtain newly added samples of the historical glass backplane image; wherein the random adjustment parameters are obtained by randomly scaling at least one of the bus bar annotation information, the protruding hole annotation information, and the edge area annotation information of the historical glass backplane image; The newly added samples are added to the set of historical glass back images to achieve data enhancement of the historical glass back images.

5. The method according to claim 3, characterized in that: The edge detection model is used to perform boundary reconstruction processing and boundary fitting processing on the edge area of ​​the glass back plate; The detecting edge regions of the first glass back plate and the second glass back plate by using the edge detection model to obtain third visual feature information corresponding to the respective edge regions includes: Performing edge detection on the first glass back plate and the second glass back plate to obtain fourth visual image information of respective edge regions; Extracting key visual features from the fourth visual image information; the key visual features are pixel feature points with color scale jumps or gray scale value jumps; the color scale jumps or gray scale value jumps indicate that the pixel feature point is located at the boundary between the glass back panel and the surrounding environment; The key visual feature is fitted as the third visual feature information.

6. The method according to claim 3, characterized in that The generating of the position deviation correction parameter based on the visual image information comprises: Based on the first visual feature information, the second visual feature information, and the third visual feature information, multi-dimensional deviation trend information between the first glass back plate and the second glass back plate is obtained through a deviation prediction model; the multi-dimensional deviation trend information includes multiple types of visual deviation trends; the multiple types of visual deviation trends include: at least one of position deviation, size deviation, shape deviation, surface quality deviation, edge deviation, hole center point deviation, and busbar endpoint deviation; Based on the multi-dimensional deviation trend information, acquiring a cumulative deviation value between the first glass back plate and the second glass back plate; The position deviation correction parameter is generated based on the accumulated deviation value; the position deviation correction parameter at least includes: a deviation adjustment parameter of a bus bar end point, a deviation adjustment parameter of a hole center point, and a deviation adjustment parameter of an edge area.

7. The method according to claim 6, characterized in that The process of obtaining the deviation adjustment parameter of the busbar endpoint is expressed as the following formula: ； in, is the deviation adjustment parameter used to calculate the busbar end points, Represents the batch index, Indicates busbar terminals, represents the busbar endpoint deviation feature dimension index, is a weight matrix, which is used for calculation in the linear mapping layer. Used for linear mapping layer to offset. Used to calculate the cumulative deviation value, It means The feature index corresponding to the busbar endpoint in the cumulative deviation value, Indicates the index value related to the cumulative deviation value.

8. The method according to claim 6, characterized in that The process of obtaining the position deviation correction parameter of the center of the protruding hole is expressed as the following formula: ； in, Used to calculate the deviation adjustment parameters of the center of the extended hole, represents the batch index, m represents the center of the mth protruding hole, and n represents the deviation feature dimension index of the protruding hole center. is a weight matrix, which is used for calculation in the linear mapping layer. Used for linear mapping layer to offset. Used to calculate the cumulative deviation value, It represents the feature index corresponding to the center of the m-th protruding hole in the cumulative deviation value, and p represents the index related to the cumulative deviation value.

9. The method according to claim 6, characterized in that The process of obtaining the position deviation correction parameter is expressed as the following formula: ； in, The deviation adjustment parameter used to calculate the edge area, Represents the batch index, Indicates Edge area attention units, Indicates the deviation feature dimension index of the edge area, is a weight matrix, which is used for calculation in the linear mapping layer. Used for linear mapping layer to offset. Used to calculate the cumulative deviation value, It means The feature index corresponding to the edge region attention unit in the cumulative deviation value, Indicates the index value related to the cumulative deviation value.

10. The method according to claim 1, characterized in that The method of automatically calibrating photovoltaic glass assembly production equipment through an equipment calibration model based on the visual image information, the production environment information, and the equipment information includes: Through the error analysis layer in the equipment calibration model, the visual image information, the production environment information and the equipment information are subjected to error analysis to identify the equipment error type of the photovoltaic glass assembly production equipment; each equipment error type is associated with a visual deviation trend in the visual image information; Determine an error correction strategy based on the identified device error type; the device error type includes at least one of the following: optical sensor error, robot arm path error, lens focus calibration error, and environmental compensation; The error correction strategy is adopted to generate error correction parameters for automatically correcting photovoltaic glass assembly production equipment.

11. The method according to claim 10, characterized in that The step of determining an error correction strategy based on the identified device error type includes: If the device error type is an optical sensor error, calculating an installation position offset and / or an installation angle offset of the optical sensor based on the visual image information, and configuring a first compensation parameter corresponding to the optical sensor based on the calculated installation position offset and / or installation angle offset; If the device error type is a robot arm path error, inverse kinematics calculation is performed on the visual image information, and the robot arm is driven to restore to the correct posture according to the calculated adjustment path; the adjustment path includes the movement direction, amplitude, and movement path required for the robot arm to return to the correct posture from the current posture; If the device error type is a lens focus calibration error, obtaining a lens adjustment parameter of the device according to the clarity or focus quality of the visual image information; the lens adjustment parameter includes at least focal length, exposure time, and / or white balance; If the equipment error type is environmental compensation, the environmental compensation model is used to predict the production environment change trend according to the production environment information, and the second compensation parameters of the photovoltaic glass component production equipment are generated based on the production environment change trend; the second compensation parameters include at least: temperature compensation parameters, humidity compensation, lighting condition compensation, and vibration compensation.

12. An automatic alignment device for photovoltaic glass components, characterized in that: The photovoltaic glass assembly comprises a plurality of glass back panels, and the automatic alignment device of the photovoltaic glass assembly comprises at least the following units, wherein A collection unit, used to collect visual image information of each of a plurality of glass back panels during the production process of photovoltaic glass modules; The visual image information at least includes: bus bar information and edge area information of a plurality of glass back panels; The acquisition unit is also used to collect production environment information and equipment information in the scene where the photovoltaic glass assembly production equipment is located; the production environment information at least includes: environmental compensation information, temperature information, humidity information, vibration information; the equipment information at least includes: optical sensor position, robot arm path, lens focus calibration information; A calibration unit, configured to automatically calibrate photovoltaic glass assembly production equipment through an equipment calibration model based on the visual image information, the production environment information, and the equipment information; A generating unit, configured to generate a position deviation correction parameter based on the visual image information; The adjustment unit is used to drive the calibrated photovoltaic glass assembly production equipment to adjust the position relationship between the multiple glass back panels according to the position deviation correction parameters to achieve automatic alignment of the multiple glass back panels.

13. An electronic device, characterized in that: include: Memory for storing computer software programs; A processor is used to read and execute the computer software program, thereby implementing the automatic alignment method of the photovoltaic glass assembly as described in any one of claims 1 to 11.

14. A non-transitory computer-readable storage medium, characterized in that: The storage medium stores a computer software program, and when the computer software program is executed by a processor, the automatic alignment method of a photovoltaic glass assembly as described in any one of claims 1 to 11 is implemented.

Images