Battery assembly yield improving method, storage medium and electronic equipment

Abstract

The invention relates to the technical field of photovoltaic module batteries, and provides a method for improving the yield of a battery module, a storage medium and electronic equipment, and the method comprises the steps: determining the defect type of a welded battery module and the current operation parameters of welding equipment; the defect type of the welded battery assembly and the current operation parameters of the welding equipment are input into a parameter generation model, and preparation operation parameters of the welding equipment are obtained; judging whether the preparatory operation parameters of the welding equipment meet the operation conditions of the welding equipment or not; and under the condition that the preparatory operation parameters of the welding equipment meet the operation conditions of the welding equipment, the preparatory operation parameters are adopted to control the welding equipment to weld the unwelded battery assemblies, so that the yield of the battery assemblies is increased. According to the method, when the battery assembly has defects, defect analysis is carried out, the equipment optimization process parameters are generated, the operation state of the equipment is automatically adjusted, and the production capacity loss caused by batch defects and shutdown test equipment adjustment is effectively avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic module batteries, and in particular, to a method for improving the yield of battery modules, a computer-readable storage medium, and an electronic device. Background Art

[0002] With the advancement of the energy revolution, solar photovoltaic power generation, as a green energy source, has received increasing attention. The welding quality of battery modules (such as solar cells) directly affects the conversion efficiency of photovoltaic cells. Poor welding of battery modules will seriously affect the performance of solar cells and reduce the lifespan of battery modules. Therefore, the quality control of battery module welding, which integrates resources in multiple aspects such as equipment, materials, and processes, has become the focus of the industry.

[0003] In the prior art, in order to control the quality of welded battery modules, some methods are adopted to improve the yield of welded battery modules. However, the methods for improving the yield of welded battery modules in the prior art are inefficient and require a long downtime of welding equipment, which affects the production output of battery modules. Summary of the Invention

[0004] The main object of the present invention is to provide a method for improving the yield of battery modules, a computer-readable storage medium, and an electronic device, so as to solve the problems that the methods for improving the yield of welded battery modules in the prior art are inefficient and affect the production output of battery modules.

[0005] To achieve the above object, according to one aspect of the present invention, a method for improving the yield of battery modules is provided, including: determining the defect type of a welded battery module and the current operating parameters of a welding device; inputting the defect type of the welded battery module and the current operating parameters of the welding device into a parameter generation model to obtain the preliminary operating parameters of the welding device; determining whether the preliminary operating parameters of the welding device meet the operating conditions of the welding device, where the operating conditions are preset conditions for the welding device to operate without faults; and when the preliminary operating parameters of the welding device meet the operating conditions of the welding device, using the preliminary operating parameters to control the welding device to weld an un-welded battery module, so as to improve the yield of the battery module.

[0006] Further, the preliminary operating parameters of the welding equipment at least include the preliminary welding temperature and the preliminary welding time of the welding equipment. Judging whether the preliminary operating parameters of the welding equipment meet the operating conditions of the welding equipment includes: determining the calculated power of the welding equipment according to the preliminary welding temperature and the preliminary welding time of the welding equipment; when the calculated power of the welding equipment is less than or equal to the preset power, determining that the preliminary operating parameters of the welding equipment meet the operating conditions of the welding equipment; when the calculated power of the welding equipment is greater than the preset power, determining that the preliminary operating parameters of the welding equipment do not meet the operating conditions of the welding equipment.

[0007] Further, determining the defect type of the welded battery assembly and the current operating parameters of the welding equipment includes: obtaining the defect types of a plurality of the welded battery assemblies and determining the defect positions of each of the welded battery assemblies; when the defect types of a preset number of the welded battery assemblies are all the target type within a preset time range and the defect positions of the preset number of the welded battery assemblies are the same, determining the defect type of the welded battery assembly as the target type and determining the current operating parameters of the welding equipment.

[0008] Further, the method further includes: when the preliminary operating parameters of the welding equipment do not meet the operating conditions of the welding equipment, performing a secondary test on the preliminary operating parameters of the welding equipment to obtain a test result, where the test result characterizes whether the welding equipment is free of faults when operating with the preliminary operating parameters; using the defect type of the welded battery assembly, the current operating parameters of the welding equipment, the preliminary operating parameters and the test result to correct the parameter generation model to obtain a corrected generation model, and when the test result is that the welding equipment is free of faults when operating with the preliminary operating parameters, using the preliminary operating parameters to control the welding equipment to weld the un-welded battery assemblies.

[0009] Further, before determining the defect type of the welded battery assembly and the current operating parameters of the welding equipment, the method further includes: obtaining the electroluminescence image of the welded battery assembly; inputting the electroluminescence image of the welded battery assembly into a defect recognition model to obtain the defect parameters of the welded battery assembly, where the defect parameters are defect-free or the defect type of the welded battery assembly, and the defect recognition model includes a plurality of sub-models, and there are differences in the defect types recognized by different sub-models.

[0010] Further, the multiple sub-models at least include a deep learning algorithm, a random forest model, and an autoregressive time series model. The electroluminescence image of the welded battery assembly is input into the defect recognition model to obtain the defect parameters of the welded battery assembly, including: inputting the electroluminescence image of the welded battery assembly into the first recognition sub-model; in the case where the first recognition sub-model cannot identify whether the battery assembly has defects, inputting the electroluminescence image of the welded battery assembly into the second recognition sub-model; in the case where the second recognition sub-model cannot identify whether the battery assembly has defects, inputting the electroluminescence image of the welded battery assembly into the third recognition sub-model to obtain the defect parameters of the welded battery assembly; wherein, the first recognition sub-model, the second recognition sub-model, and the third recognition sub-model are respectively one of the deep learning algorithm, the random forest model, and the autoregressive time series model, and the first recognition sub-model, the second recognition sub-model, and the third recognition sub-model are not the same.

[0011] Further, after inputting the electroluminescence image of the welded battery assembly into the defect recognition model to obtain the defect parameters of the welded battery assembly, the method further includes: performing a secondary detection on the battery assembly to obtain a detection result, where the detection result indicates whether the battery assembly has defects and the defect type when there are defects; in the case where the defect parameters of the welded battery assembly are defect-free, and the detection result indicates that the battery assembly has defects and the defect type of the welded battery assembly, using the defect parameters of the welded battery assembly and the detection result to correct the defect recognition model to obtain a corrected recognition model.

[0012] Further, the current operating parameters of the welding device include the current welding temperature of the welding device, the current welding time of the welding device, and the current lamp tube power of the welding device. The preparatory operating parameters of the welding device include the preparatory welding temperature of the welding device, the preparatory welding time of the welding device, and the preparatory lamp tube power of the welding device. Wherein, in the case where the defect type of the welded battery assembly is a cold weld, the magnitude relationship between the current operating parameters of the welding device and the preparatory operating parameters of the welding device satisfies at least one of the following: the preparatory lamp tube power of the welding device is greater than the current lamp tube power of the welding device, the preparatory welding temperature of the welding device is greater than the current welding temperature of the welding device, and the preparatory welding time of the welding device is greater than the current welding time of the welding device.

[0013] According to another aspect of the present invention, there is provided a computer-readable storage medium including a stored program, wherein when the program runs, it controls the device where the computer-readable storage medium is located to execute any one of the methods for improving the yield rate of battery components.

[0014] According to another aspect of the present invention, there is provided an electronic device, including: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include those for executing any one of the methods for improving the yield rate of battery components.

[0015] The beneficial effects of the present application are as follows: For the above method for improving the yield rate of battery components, first, determine the defect type of the welded battery components and the current operating parameters of the welding equipment; then input the defect type of the welded battery components and the current operating parameters of the welding equipment into the parameter generation model to obtain the preliminary operating parameters of the welding equipment; then determine whether the preliminary operating parameters of the welding equipment meet the operating conditions of the welding equipment, where the operating conditions are the preset conditions for the welding equipment to operate without faults; finally, when the preliminary operating parameters of the welding equipment meet the operating conditions of the welding equipment, use the preliminary operating parameters to control the welding equipment to weld the un-welded battery components to improve the yield rate of the battery components. This method cancels manual re-judgment. In the case of defective battery components, it conducts real-time root cause analysis of the defects and generates real-time equipment optimization process parameters, sends the production equipment optimization process parameters to the machine, and automatically adjusts the equipment operating state in real time, effectively avoiding batch defects and production capacity losses caused by downtime for testing and adjusting the equipment, and solving the problems of low efficiency and impact on the production volume of battery components in the prior art for improving the yield rate of welded battery components. Description of the Drawings

[0016] The specification drawings forming a part of the present application are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention. In the drawings:

[0017] Figure 1 It shows a schematic flowchart of a method for improving the yield rate of battery components according to an embodiment of the present invention;

[0018] Figure 2 It shows a schematic flowchart of another method for improving the yield rate of battery components according to an embodiment of the present invention. Detailed Embodiments

[0019] It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0020] In order to enable those skilled in the art to better understand the solutions of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present application.

[0021] It should be noted that the terms "first", "second", etc. in the specification, claims and the above-mentioned drawings of the present application are used to distinguish similar objects, and do not have to be used to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so as to implement the embodiments of the present application described herein. In addition, the terms "include" and "have" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0022] As introduced in the background art, in the prior art, a 4K line scan camera and an industrial control computer are installed on each production device in the welding process of the component workshop. The 4K line scan camera is used to collect pictures and monitor the quality of battery cells in real time online. If the production line detects a defective product, the industrial control computer software will determine the corresponding picture as a defective picture and upload it to the component workshop for manual rejudgment. The production line automation equipment will release the products manually rejudged as good products, and automatically reject the products manually rejudged as defective. If the personnel in the centralized rejudgment room continuously find the same type of defective products three times, they need to call the process personnel of the workshop production line and require forced line stop to optimize the equipment operation parameters, which seriously affects the production capacity. Moreover, calling to stop the production line after the rejudgment room personnel continuously find the same type of defective products three times will cause a time lag, which is likely to lead to the outflow of batch defective products.

[0023] To solve the problems in the prior art that the method for improving the yield rate of welded battery components is inefficient and affects the output of battery components, the embodiments of the present application provide a method for improving the yield rate of battery components, a computer-readable storage medium and an electronic device.

[0024] The following further describes the present application in detail with specific embodiments, and these embodiments should not be construed as limiting the scope claimed by the present application.

[0025] In this embodiment, a method for improving the yield rate of battery components running on a mobile terminal, a computer terminal or a similar computing device is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0026] Figure 1 1 is a flow chart of a method for improving the yield rate of a battery assembly according to an embodiment of the present application. Figure 1 As shown, the method comprises the following steps:

[0027] Step S101, determining the defect type of the welded battery assembly and the current operating parameters of the welding equipment;

[0028] Specifically, the welded battery assembly is a battery cell welded by a welding device. The current operating parameters include but are not limited to key parameters such as welding temperature, welding time, and flux concentration of the welding device.

[0029] During the welding process of photovoltaic cell modules, various types of defects may occur, including but not limited to cold solder joints, cracks, short circuits, foreign matter contamination, welding offset, etc. These defects will affect the electrical performance and mechanical integrity of the cell modules, thereby reducing the overall efficiency and life of the modules. By determining the defect type of the welded cell modules through image acquisition and other technologies, and recording the current operating parameters of the welding equipment at this time, the relationship between the defects of the welded cell modules and the current operating parameters of the welding equipment can be analyzed, so that the operating parameters of the welding equipment can be adjusted later to eliminate the defects of the subsequent welded cell modules. Among them, image acquisition can use a high-precision camera (such as a 4K line scan camera) to continuously scan and shoot the components to obtain images of the welding parts, and can also obtain electroluminescent images for defect analysis. The current operating parameters of the welding equipment can be monitored and recorded in real time using sensors on the equipment.

[0030] Step S102, inputting the defect type of the welded battery assembly and the current operating parameters of the welding equipment into a parameter generation model to obtain the preliminary operating parameters of the welding equipment;

[0031] Specifically, the parameter generation model is obtained based on random forest model training, and the preparatory operating parameters include but are not limited to key parameters such as welding temperature, welding time, flux concentration, etc. of the welding equipment.

[0032] In some embodiments, the parameter generation model may also be obtained by training models such as deep learning models, autoregressive time series models, support vector machines, gradient boosting trees, and neural networks.

[0033] Before inputting the defect types of the above-welded battery modules and the current operating parameters of the above welding equipment into the parameter generation model, the collected data can be preprocessed to extract features that are helpful for analysis. For example, for the identification of defect types, features such as the grayscale, texture, and shape of the image may need to be extracted; for operating parameters, historical data may need to be standardized to eliminate the influence of dimensions.

[0034] In addition, in the historical process, machine learning or deep learning algorithms are used to train the parameter generation model. The training dataset of this model consists of historical defect data and corresponding operating parameters. The goal is to learn the correlation between defect types and operating parameters, that is, under which combinations of operating parameters, which types of defects are more likely to occur. When the system detects a new defect type, it will input the defect type and the current operating parameters into the trained parameter generation model in real time. Based on the learned patterns, the model predicts which adjustments of operating parameters are most likely to reduce or eliminate this defect. The model outputs the preliminary operating parameters for a specific defect, that is, the recommended equipment adjustment plan. These parameters may include adjusting the welding temperature, welding pressure, welding time, solder tape speed, etc. to adapt to different defect types and reduce the defect rate.

[0035] Suppose during the welding process of photovoltaic battery modules, the system continuously identifies that there is a "loose weld" defect in the welded battery modules, and the current operating parameters of the welding equipment are a welding temperature of 200°C, a welding pressure of 150 N, a welding time of 2 s, etc. The defect type of "loose weld" and the current operating parameters (for example, a welding temperature of 200°C, a welding pressure of 150 N, etc.) are input into the parameter generation model. The model analyzes the historical data and finds that there is a strong correlation between a lower welding temperature and the loose weld defect. Based on this, the model predicts that increasing the welding temperature can reduce the occurrence of loose welds. The model outputs the preliminary operating parameters and recommends adjusting the welding temperature from 200°C to 210°C, that is, a welding temperature of 210°C is the preliminary operating parameter for the welding equipment.

[0036] In addition, in addition to the defect types of the welded battery modules and the current operating parameters of the above welding equipment, the influence of external conditions such as ambient temperature and humidity on the welding effect can also be considered, that is, both the ambient temperature and ambient humidity are input into the parameter generation model, so that the parameter generation model takes into account the influence of ambient temperature and ambient humidity, generates more suitable preliminary operating parameters, and adjusts the welding parameters in real time to ensure the stability of welding quality under different environments.

[0037] Step S103, determine whether the preliminary operating parameters of the above welding equipment meet the operating conditions of the above welding equipment, and the operating conditions are the preset conditions for the above welding equipment to operate without faults;

[0038] Specifically, each welding device has its specific physical limitations, including but not limited to the parameter ranges of the highest and lowest temperatures, pressures, speeds, etc. The preliminary operating parameters must fall within these limitations; otherwise, the device cannot physically execute these parameters, or the device may be damaged during execution. A series of safety standards must be observed during the welding process, such as preventing overheating, preventing the generation of electric arcs, and ensuring the safety of operators. The preliminary operating parameters need to be compared with these safety specifications to ensure that the operation of the device does not violate the safety regulations under the adjusted parameters. In addition to physical limitations and safety specifications, the impact of parameter adjustment on production efficiency also needs to be considered. For example, if the preliminary operating parameters suggest a significant reduction in the welding speed, although it may improve the welding quality, it will significantly reduce the production rate and may not be adopted. Therefore, it is necessary to ensure that the impact of the parameter adjustment plan on production efficiency is acceptable under the premise of meeting the quality requirements.

[0039] In some embodiments, to determine whether the preliminary operating parameters of the above welding device meet the operating conditions of the above welding device, a manual re-verification method is generally adopted. The parameter generation model is trained based on a random forest model and adopts a manual semi-supervised learning training mechanism, that is, first, the parameter generation model makes a prediction, and then it is manually determined whether the preliminary operating parameters obtained by the model prediction can be actually applied, so as to perform real-time training on the parameter generation model to ensure the real-time accuracy of the parameter generation model.

[0040] Step S104, when the preliminary operating parameters of the above welding device meet the operating conditions of the above welding device, use the above preliminary operating parameters to control the above welding device to weld the un-welded battery components, so as to improve the yield rate of the battery components.

[0041] Specifically, based on the above analysis, the established parameter generation model (such as a deep learning model, random forest, autoregressive time series model, etc.) is used to predict a new set of operating parameters (preliminary operating parameters), which theoretically can reduce or eliminate the defects encountered before. Before applying the preliminary operating parameters to the device, it is necessary to ensure that these parameters do not exceed the physical limits, safety specifications, or production efficiency requirements of the device. This is achieved by comparing the device specifications, safety operation guidelines, and production efficiency goals. Once it is confirmed that the preliminary operating parameters meet the device operating conditions, these parameters become the actual operating parameters and are used to control the welding device to weld the subsequent un-welded battery components. After implementing the new parameters, the system will continuously monitor the welding quality and production efficiency to evaluate the effect of parameter adjustment. If the yield rate increases and the production efficiency remains stable or increases, it indicates that the parameter adjustment is successful.

[0042] In the prior art, in the case of a defective component cell, the defective picture is uploaded to the component workshop for manual rejudgment. If the inspector detects the same type of defect three times in a row, the inspector will call the on-site process personnel to stop the machine and adjust the equipment operation parameters. This inspection method has relatively high requirements for the experience and professionalism of the inspector, and a huge number of cells need to be inspected manually. Since it is manual inspection, it will be affected by physiological and psychological factors, and factors such as human emotions / attention will interfere with the judgment, resulting in problems such as missed judgment / passed judgment. Moreover, different judgment personnel have differences in the boundary between qualified and unqualified, and the same product may have different judgment results in front of different inspectors, increasing the possibility of missed judgment. In addition, a large number of professionals need to be hired for manual inspection, and the personnel flow is frequent, requiring continuous training and maintenance, resulting in high labor input costs. Furthermore, the above traditional inspection method is post-inspection. When defects occur, the process line analyzes the reasons and manually adjusts the machine parameters. This kind of line stop to optimize the equipment operation parameters will seriously affect the system production capacity, and the time lag caused by calling to stop the production line after the rejudgment room personnel continuously find the same type of defective products three times is likely to lead to the outflow of batch defective products.

[0043] The above method for improving the yield of the battery module of the present application first determines the defect type of the welded battery module and the current operation parameters of the welding equipment; then inputs the defect type of the welded battery module and the current operation parameters of the welding equipment into the parameter generation model to obtain the preliminary operation parameters of the welding equipment; then judges whether the preliminary operation parameters of the welding equipment meet the operation conditions of the welding equipment, and the operation conditions are the preset conditions for the welding equipment to operate without faults; finally, when the preliminary operation parameters of the welding equipment meet the operation conditions of the welding equipment, the preliminary operation parameters are used to control the welding equipment to weld the un-welded battery module, so as to improve the yield of the battery module. This method cancels manual rejudgment. In the case of defective battery modules, it conducts real-time root cause analysis of the defects and generates equipment optimization process parameters in real time, sends the production equipment optimization process parameters to the machine tool, and automatically adjusts the equipment operation state in real time, effectively avoiding the production loss caused by batch defects and downtime for testing and adjusting the equipment, and solving the problem that the method for improving the yield of the welded battery module in the prior art is inefficient and affects the output of the battery module.

[0044] Before determining the defect type of the welded battery module, it is necessary to first detect the defects of the welded battery module, that is, to detect whether there are defects in the welded battery module. That is, before determining the defect type of the welded battery module and the current operation parameters of the welding equipment, the above method further includes the following steps:

[0045] Step S201, obtaining the electroluminescence image of the above-mentioned welded battery module;

[0046] Among them, after the battery module is welded, it is necessary to apply an appropriate voltage or current to it through a dedicated electroluminescence (EL) detection device to prompt the semiconductor material inside the battery module to emit visible light, and then capture the EL image. These images contain the internal structure information of the battery module, especially the state of the welding points, such as whether there are poor welds, cracks or other defects.

[0047] The specific steps to obtain the electroluminescence image of the above-mentioned welded battery module are as follows:

[0048] 1. According to the characteristics of the battery module, set an appropriate current or voltage (usually the current is between several hundred milliamperes and several tens of amperes). Electroluminescence detection relies on the light radiation generated by the internal electron transition of the photovoltaic cell under the action of current or voltage. Apply the current or voltage to the battery module to stimulate the internal electron transition and generate the electroluminescence phenomenon.

[0049] 2. The camera should be correctly positioned to ensure that the entire battery module or the area to be detected can be captured. Adjust parameters such as the exposure time and gain of the camera to capture a clear and high-contrast EL image. The exposure time is generally long to fully capture the weak electroluminescence. During the application of current or voltage, use the camera to capture the EL image of the battery module. To improve the detection efficiency, a high-speed scanning camera may be used, which can continuously capture multiple images in a short time.

[0050] 3. It may be necessary to correct the captured original image to eliminate the influence of lens distortion, uneven illumination, etc. Save the processed image to the computer system for the defect recognition model to analyze. Transmit the captured and processed EL image to the server or workstation of the defect recognition system. Ensure that the image is stored in an appropriate format for subsequent data management and analysis.

[0051] In step S202, input the electroluminescence image of the above-mentioned welded battery module into the defect recognition model to obtain the defect parameters of the above-mentioned welded battery module. The above-mentioned defect parameters are either defect-free or the defect types of the above-mentioned welded battery module. The above-mentioned defect recognition model includes multiple sub-models. Among them, the defect types identified by applying different above-mentioned sub-models are different.

[0052] Among them, an electroluminescence image (EL image) is input into a defect recognition model composed of multiple sub-models. These sub-models may include, but are not limited to, deep learning models, support vector machines (SVMs), decision trees, etc. Each type of sub-model has its unique advantages and the ability to identify specific types of defects. Each sub-model focuses on identifying a specific type of defect, and through the joint work of the models, comprehensive defect detection results can be obtained. For example, a certain sub-model may be specifically used to identify false soldering, while another may be better at detecting cracks. The defect recognition model analyzes the features of the EL image, such as brightness distribution, texture changes, color differences, etc., to determine whether there are defects in the battery module and the specific type of defect.

[0053] In the prior art, a 4K line scan camera is generally used to obtain an image of the welded battery module and input the image into an existing defect recognition model for defect recognition. A 4K line scan camera is a high-resolution line array camera, mainly used for high-precision detection and imaging applications in industrial automation. The working principle of a line scan camera is to scan and image the surface of a continuously moving object through one or more linear image sensors. Different from a area array camera that captures the entire area at once, a line scan camera captures image information row by row (or point by point) during the movement of the object. 4K resolution refers to the number of pixel points per row reaching 4096 or 3840 pixels, and the specific value depends on different standards. For example, digital movies usually use a resolution of 4096×2160, while consumer-grade ultra-high-definition TVs use a resolution of 3840×2160.

[0054] The electroluminescence (EL) image adopted in the embodiments of this application has multiple advantages in the detection of photovoltaic battery modules compared with the image obtained by a 4K line scan camera. These advantages are mainly reflected in the sensitivity of defect detection, the interpretability of the image, and the ability to reveal internal defects, as follows:

[0055] 1. High-sensitivity identification of internal defects: The EL image can reveal subtle defects inside the battery module, such as microcracks, hidden cracks, or local battery performance degradation, which are difficult to detect in surface images (such as images taken by a 4K line scan camera). Under the electroluminescence effect, the defect area will show a different brightness level from the normal area, which provides a direct and obvious signal for defect identification.

[0056] 2. Intuitive display of electrical performance: The EL image not only reflects physical defects but also can intuitively display the electrical performance of the battery module. In the EL image, the brightness of the battery cell is closely related to its electrical performance. Therefore, it can be used to quickly evaluate the working state of the battery cell, such as short circuit or power drop, which is crucial for ensuring the overall efficiency of the battery module.

[0057] 3. Adapt to complex environments: EL imaging is carried out in a darkroom environment, which can effectively shield the interference of external ambient light and ensure that the image quality is not affected by external light. In contrast, the imaging quality of a 4K line-scan camera may be affected by factors such as ambient light conditions and camera settings. Especially when the light conditions are unstable, it may affect the clarity and contrast of the image.

[0058] 4. Detailed defect type identification: EL images can more accurately distinguish different types of defects, such as poor soldering, cracks, short circuits, and power attenuation. This is crucial for targeted adjustment of production parameters and optimization of the process flow. The images of a 4K line-scan camera are mainly used to detect surface defects, and its ability to distinguish internal or electrical property defects is relatively weak.

[0059] In summary, EL images demonstrate significant advantages in revealing internal defects, displaying electrical performance, and adapting to complex detection environments. They are particularly suitable for the production and quality control of photovoltaic modules, providing strong support for precise detection and process optimization.

[0060] Therefore, the automatic detection and analysis of EL images replace manual visual inspection, greatly improving the detection efficiency, reducing labor costs, and avoiding human errors. By continuously collecting and analyzing EL images and their corresponding defect parameters, a large amount of valuable data can be accumulated for training models, enabling them to perform better when dealing with more diverse and complex defects. In summary, the above steps S201 - S202, through automated and intelligent means, not only improve the detection efficiency and accuracy of the welding quality of battery modules but also provide important data support for dynamic parameter optimization on the production line, contributing to the continuous improvement of the production process and the enhancement of product quality.

[0061] In addition, the existing defect recognition models are only one type of model with a relatively low recognition accuracy, generally below 95%. In contrast, the defect recognition model of this embodiment includes multiple sub-models, each focusing on the recognition of different types of defects. Therefore, it can provide more accurate defect detection results, reduce the occurrence probability of false positives and false negatives, and the recognition accuracy can reach over 99%, greatly improving the defect recognition rate of solar cells. Moreover, the existing defect recognition models do not have an autonomous learning function.

[0062] The defect recognition and detection models of the prior art generally only include one type of model, resulting in relatively low accuracy of the detected results. This leads to 90% of the workload of manual rechecking in the prior art being ineffective work due to the low recognition accuracy of the existing defect recognition models. In this embodiment, at least the above-mentioned multiple sub-models include deep learning algorithms, random forest models, and autoregressive time series models. Inputting the electroluminescence image of the welded battery module into the defect recognition model to obtain the defect parameters of the welded battery module includes the following steps:

[0063] Step S2021: Input the electroluminescence image of the above-mentioned welded battery assembly into the first recognition sub-model;

[0064] Step S2022: In the case where the first recognition sub-model cannot recognize whether there are defects in the battery assembly, input the electroluminescence image of the above-mentioned welded battery assembly into the second recognition sub-model;

[0065] Step S2023: In the case where the second recognition sub-model cannot recognize whether there are defects in the battery assembly, input the electroluminescence image of the above-mentioned welded battery assembly into the third recognition sub-model to obtain the defect parameters of the above-mentioned welded battery assembly; wherein, the first recognition sub-model, the second recognition sub-model, and the third recognition sub-model are respectively one of the above-mentioned deep learning algorithm, the above-mentioned random forest model, and the above-mentioned autoregressive time series model, and the first recognition sub-model, the second recognition sub-model, and the third recognition sub-model are different from each other.

[0066] Generally, the first recognition sub-model is a deep learning algorithm, such as a convolutional neural network (CNN). Deep learning models perform well in the field of image recognition and can capture complex features in EL images, such as poor soldering and cracks. Due to its high accuracy, it is first analyzed by the CNN to quickly filter out obvious defect situations. If the first sub-model cannot clearly identify whether there are defects in the battery assembly or there is uncertainty in the defect recognition, the second recognition sub-model, such as a random forest model, will be activated. The random forest model can handle multi-dimensional data, has a good tolerance for data imbalance and noise, and is suitable for providing supplementary judgment when the first model is unable to determine. When the second sub-model also fails to give a clear judgment, the third recognition sub-model, such as an autoregressive time series model, will be used for further analysis. The autoregressive time series model is good at processing time series data, can analyze the changing trends of device operating parameters over time, and how these trends affect the welding quality of battery assemblies. In a continuous production environment, this model can capture potential quality problems related to time.

[0067] Through a multi-stage recognition model, it is possible to gradually refine defect recognition, improve recognition accuracy, especially when dealing with complex or borderline cases. The complementarity of different types of models in recognition can enhance the robustness of the overall recognition process. Even when a single model performs poorly under certain conditions, the entire system can still give accurate judgments through the supplementation of other models. Adopting a multi-model recognition strategy can analyze images from different perspectives, significantly reduce the situations of misjudgment and missed judgment, and improve production efficiency and product quality. In the recognition process, a model with lighter computational resources is first used for rapid screening, and a model with higher complexity is only called when further analysis is needed. This can optimize the allocation of model training and running resources and reduce the overall computational cost.

[0068] In addition, the above first recognition sub-model, second recognition sub-model, and third recognition sub-model include but are not limited to the above three models, and the order of the three models can also be adjusted according to the actual situation. The above steps only show an optimal embodiment.

[0069] Wherein, after inputting the electroluminescence image of the above-mentioned welded battery assembly into the defect recognition model to obtain the defect parameters of the above-mentioned welded battery assembly, the above method further includes the following steps:

[0070] Step S301, perform a secondary detection on the above battery assembly to obtain a detection result, and the detection result characterizes whether there are defects in the above battery assembly and the defect type when there are defects.

[0071] Among them, a secondary detection is performed on the battery assembly after preliminary defect recognition. This step usually involves manual re-inspection or using higher-precision detection equipment to verify the results of the first defect recognition. The purpose of the secondary detection is to confirm the existence or non-existence of defects and more precisely define the defect type, preventing the defect recognition model from misinterpreting whether the battery assembly has defects or misidentifying the defect type, so as to provide a benchmark or "true value" for the subsequent model correction process.

[0072] Step S302, in the case that the defect parameter of the above-mentioned welded battery assembly is defect-free, and the detection result characterizes that there are defects in the above battery assembly and the defect type of the above-mentioned welded battery assembly, use the defect parameter of the above-mentioned welded battery assembly and the detection result to correct the above defect recognition model to obtain a corrected recognition model.

[0073] Specifically, if the preliminary defect recognition model determines that a battery module has no defects, but the secondary detection results show that there are indeed defects, then this instance of "false alarm" or "missed detection" is used for model correction. By comparing the model predictions with the actual detection results, the machine learning algorithm can identify the deficiencies in the model and adjust the model parameters accordingly. This correction is usually achieved by training the model, that is, feeding the real data containing the defect types to the model, enabling the model to learn from the mistakes and improve its recognition ability in similar situations.

[0074] That is, steps S301 and S302 can actually be regarded as a process of semi-supervised learning of a machine model. In the early stage, multiple manual secondary detections may be required to continuously correct the defect recognition model. With continuous correction over a certain period of time, the recognition accuracy of the defect recognition model can be continuously improved, and the number of manual secondary detections can be gradually reduced until no manual secondary detection is required at the end.

[0075] Through the feedback correction mechanism in the above steps, the defect recognition model can learn more comprehensive defect features, gradually reduce the probability of misjudgment and missed detection, and improve the overall detection accuracy. As the model is corrected by more false alarm instances, it can better understand the defect patterns of the battery modules and make more accurate judgments even when faced with defect types or slight variations that have not been seen before. As the accuracy of the model improves, the number of battery modules that need to be manually reinspected will decrease, reducing the labor cost and improving the production efficiency.

[0076] Among them, determining the defect types of the welded battery modules and the current operating parameters of the welding equipment includes the following steps:

[0077] Step S1011, obtaining the defect types of multiple above-mentioned welded battery modules and determining the defect positions of each of the above-mentioned welded battery modules;

[0078] The system first obtains the defect types of multiple welded battery modules from electroluminescence (EL) images or other types of defect detections and determines the specific positions of these defects on the modules. This process relies on advanced image processing techniques and pattern recognition algorithms to accurately identify the defect areas, such as poor welding, cracks, etc., and determine whether these defects occur at the same positions on the battery modules.

[0079] Step S1012, when the defect types of a preset number of welded battery modules are all the target type within a preset time range and the defect positions of the above-mentioned preset number of welded battery modules are the same, determining the defect type of the above-mentioned welded battery module as the above-mentioned target type and determining the current operating parameters of the above-mentioned welding equipment.

[0080] Among them, the preset time range can be set to 5 minutes, and the preset number is generally three. The above step S1012 can also be equivalent to determining the target type according to the frequency of occurrence of the defect types of the welded battery components.

[0081] Within a certain preset time range, if the system detects that a preset number of welded battery components have the same type of defect and the positions of these defects are the same, then the system will determine the currently occurring defect type as the target type and record the current operating parameters of the welding equipment when these defects occur, such as welding temperature, pressure, speed, etc.

[0082] Suppose the system detects that 5 battery components have defects within 5 consecutive minutes. The defect type of the first battery component is poor welding, the defect type of the second battery component is poor welding, the defect type of the third battery component is cracking, the defect type of the fourth battery component is short circuit, and the defect type of the fifth battery component is poor welding, and the poor welding positions of the first battery component, the second battery component, and the fifth battery component are the same. Then the target type is poor welding. That is, when the occurrence frequency of the same defect reaches the preset frequency, the parameters of the welding equipment need to be adjusted.

[0083] In some other embodiments, the parameters of the welding equipment are also adjusted when the same defect appears three times continuously. That is, the defect types of the first battery component, the second battery component, and the third battery component are all poor welding and the poor welding positions are the same.

[0084] This step of determining whether to update the operating parameters of the welding equipment based on the occurrence frequency and occurrence times of the defect types can effectively prevent the problem of overly frequent parameter changes caused by the contingency of battery component defects leading to changes in the operating parameters of the welding equipment. For example, if there are no requirements for frequency and number, then the operating parameters of the welding equipment will be changed every time a battery component has a defect, which will also cause great damage to the welding equipment and easily reduce the life of the welding equipment.

[0085] In some embodiments, the preliminary operating parameters of the above welding equipment at least include the preliminary welding temperature of the above welding equipment and the preliminary welding time of the above welding equipment. Judging whether the preliminary operating parameters of the above welding equipment meet the operating conditions of the above welding equipment includes the following steps:

[0086] Step S1031, determine the calculated power of the above welding equipment according to the preliminary welding temperature of the above welding equipment and the preliminary welding time of the above welding equipment;

[0087] Among them, the energy requirement during the welding process mainly comes from converting electrical energy into heat energy to heat the solder to make it melt and form good contact with the battery chip.

[0088] In some embodiments, the energy required for the welding process can be calculated according to the energy calculation formula Q = m×c×ΔT, where Q is the energy required for the welding process, m is the mass of the solder, c is the specific heat capacity of the solder, and ΔT is the temperature change, that is, the temperature change value from room temperature to the pre-welding temperature.

[0089] After calculating the energy required for the welding process according to the energy calculation formula, according to the power calculation formula P = Q / t, calculate the power of the welding equipment required to provide the energy required for the welding process, where P is the required power and t is the pre-welding time.

[0090] Calculate the power required by the welding equipment when welding at the pre-welding temperature and pre-welding time according to the above energy calculation formula and power calculation formula, and determine whether this power exceeds the maximum power that the welding equipment can provide, so as to prevent the situation of exceeding the maximum load of the welding equipment during the welding process using the pre-welding temperature and pre-welding time, resulting in damage to the welding equipment.

[0091] Step S1032, in the case where the calculated power of the above welding equipment is less than or equal to the preset power, determine that the pre-operation parameters of the above welding equipment meet the operating conditions of the above welding equipment;

[0092] Among them, the calculated theoretical power demand is compared with the preset safety power threshold. If the calculated power is less than or equal to the preset threshold, it means that the equipment can operate safely at the current set welding temperature and time, there will be no overload, and there is enough energy to complete the welding, meeting the operating conditions of the equipment.

[0093] Step S1033, in the case where the calculated power of the above welding equipment is greater than the above preset power, determine that the pre-operation parameters of the above welding equipment do not meet the operating conditions of the above welding equipment.

[0094] The above steps are judged by a judgment model such as machine learning. If the calculated power demand is greater than the preset power threshold, then the current welding temperature and time settings will be judged not to meet the operating conditions of the equipment. In this case, the system will prevent the equipment from operating under unsafe parameters to avoid equipment damage or a decrease in welding quality.

[0095] Specifically, the above steps can ensure that the device operates within a safe power range, prevent the device from being overloaded or damaged, and ensure production safety. By verifying the power requirements of the device under given parameters, it is possible to avoid poor welding due to insufficient energy or component damage caused by excessive welding, thereby improving welding quality and production efficiency. Avoiding the device from operating at high power for a long time reduces the wear on the device and extends its service life. This process supports automated parameter verification and adjustment, reduces the dependence on manual intervention, improves the level of production automation, and reduces the risk of human error.

[0096] Generally, the preset power is set relatively small, usually about 2w. This can ensure that the preliminary operating parameters automatically selected by the system are basically within the safe range and will not exceed the maximum power that the welding device can withstand.

[0097] For cases where the preset power is exceeded, it is necessary to conduct a secondary manual test to determine whether the preliminary welding temperature and preliminary welding time can be used for welding. That is, the above method further includes the following steps:

[0098] Step S401, in the case where the preliminary operating parameters of the above welding device do not meet the operating conditions of the above welding device, conduct a secondary test on the preliminary operating parameters of the above welding device to obtain a test result, where the test result characterizes whether the above welding device has no faults when operating with the above preliminary operating parameters;

[0099] Among them, when the preliminary operating parameters do not meet the theoretically operating conditions, these parameters will be subject to a secondary test, and the secondary test is generally a manual recheck. The purpose of the manual recheck is to actually verify the performance and stability of the welding device when using these preliminary operating parameters to ensure that the device can operate without faults.

[0100] Step S402, use the defect type of the above welded battery components, the current operating parameters of the above welding device, the above preliminary operating parameters, and the above test result to correct the above parameter generation model to obtain a corrected generation model, and in the case where the above test result is that the above welding device has no faults when operating with the above preliminary operating parameters, use the above preliminary operating parameters to control the above welding device to weld the un-welded battery components.

[0101] Specifically, if the result of the secondary test shows that the welding device can weld without faults when operating with the preliminary operating parameters and the welding quality meets the requirements, then the system will use the defect type of the welded battery components, the current operating parameters of the welding device, the preliminary operating parameters, and the test result to correct the parameter generation model. The purpose is to make the model more accurately reflect the relationship between the actual performance of the device and the welding quality, thereby improving the prediction accuracy and parameter optimization ability of the model.

[0102] The above steps S401 and S402 can also be regarded as a process of semi-supervised learning of a machine model. Record the test results of the secondary test and the corresponding preliminary operating parameters, and continuously train the model based on these data. For example, a set of preliminary operating parameters is obtained, and the calculated power based on it is greater than the preset power. However, the test result obtained by manual secondary test is that this set of preliminary operating parameters can be used for welding. Then, record this set of preliminary operating parameters and the test result obtained by manual secondary test into the model training. The next time this set of preliminary operating parameters is encountered, the model can directly determine that welding can be performed, and there is no need to perform manual testing again. Similarly, a set of preliminary operating parameters is obtained, and the calculated power based on it is greater than the preset power. However, the test result obtained by manual secondary test is that this set of preliminary operating parameters cannot be used for welding. Then, record this set of preliminary operating parameters and the test result obtained by manual secondary test into the model training. The next time this set of preliminary operating parameters is encountered, the model can directly determine that welding cannot be performed, and there is also no need to perform manual testing again.

[0103] Multiple manual secondary tests may be required in the early stage to continuously correct the judgment model. With continuous correction over a certain period of time, the recognition accuracy of the judgment model is continuously improved, and the number of manual secondary tests can be gradually reduced until no manual secondary test is required at the end.

[0104] In some embodiments, the current operating parameters of the above welding device include the current welding temperature of the above welding device, the current welding time of the above welding device, and the current lamp power of the above welding device. The preliminary operating parameters of the above welding device include the preliminary welding temperature of the above welding device, the preliminary welding time of the above welding device, and the preliminary lamp power of the above welding device. Among them, when the defect type of the above welded battery assembly is virtual welding, the magnitude relationship between the current operating parameters of the above welding device and the preliminary operating parameters of the above welding device satisfies at least one of the following: the preliminary lamp power of the above welding device is greater than the current lamp power of the above welding device, the preliminary welding temperature of the above welding device is greater than the current welding temperature of the above welding device, and the preliminary welding time of the above welding device is greater than the current welding time of the above welding device.

[0105] Specifically, for the virtual welding defect that appears in the welded battery assembly, the system will automatically adjust the preliminary operating parameters of the welding device in order to improve the welding quality. Virtual welding is usually caused by insufficient heat energy during the welding process, resulting in insufficient fusion of the solder joints. Therefore, by increasing the lamp power, welding temperature, or welding time of the welding device, the heat energy supply can be increased, the fusion degree of the solder joints can be improved, and thus the generation of virtual welding defects can be reduced.

[0106] Suppose that the battery modules produced by the welding equipment under the current welding parameters (welding temperature: 350°C, welding time: 2.5 seconds, lamp power: 800W) have the defect of false soldering. Based on this situation, the system adjusts the preliminary operating parameters to: Preliminary lamp power: increased from 800W to 850W (enhanced heat energy input). Preliminary welding temperature: remain unchanged at 350°C (maintain the stability of heat energy input). Preliminary welding time: increased from 2.5 seconds to 3 seconds (extended heat action time).

[0107] By increasing the heat energy input, such as increasing the lamp power or welding temperature, and extending the heat action time (i.e., welding time), the false soldering phenomenon can be significantly reduced, and the fusion degree of the solder joints and the overall quality of the welded components can be improved. Although increasing the power and temperature may slightly increase the energy consumption, the overall production efficiency is improved by reducing defective products and increasing the one-time welding success rate, reducing the costs of re-welding and material waste, and ultimately achieving effective optimization of energy and costs. The automated parameter adjustment reduces the dependence on manual judgment and manual adjustment, and reduces the product quality problems caused by manual operation errors. The system makes intelligent adjustment decisions based on historical defect data and current operating parameters, and through continuous learning and optimization, can set the preliminary operating parameters more precisely, improving the scientificity and rationality of the decisions.

[0108] To enable those skilled in the art to more clearly understand the technical solution of the present application, the implementation process of the method for improving the yield rate of the battery modules of the present application will be described in detail below in combination with specific embodiments.

[0109] This embodiment relates to a specific method for improving the yield rate of battery modules, such as Figure 2As shown in the figure, first, the solar cells are loaded through a cassette, and the solar cells are placed on the conveyor table by a suction cup for stringing. Then, the solar cells are welded into strings with soldering tapes. The welded cell strings are detected by a defect recognition model (the defect recognition model is a combination of three models: deep learning algorithm, random forest model, and autoregressive time series model). The machine determines according to the set threshold, determines whether there are welding defects in the welded cell strings, and confirms the type of welding defects that occur. When it is determined that there are no welding defects in the welded cell strings, the cell strings are directly stacked and welded and typeset. When it is determined that there are welding defects in the welded cell strings, a manual secondary test is performed to confirm whether the judgment is correct. When the manual secondary test confirms that the judgment is correct (that is, there are indeed real defects), new welding parameters of the welding equipment are generated through a parameter generation model to automatically adjust the parameters of the welding equipment (to achieve closed-loop system control of welding). Then, the defective cell strings are repaired. If the same defects still occur continuously, photos are taken and fed back to determine the subsequent processing strategy. When the manual secondary test confirms that the judgment is incorrect (that is, the model misjudges, and there is no fault in the actual cell string or the fault type is incorrect), a manual misjudgment recognition (that is, a manual secondary test) is performed to determine whether there is actually a fault or the real fault type in the cell string. According to the results of the manual secondary test, the defect recognition model is iteratively optimized to implement a supervised machine learning strategy, and the non-fault cell strings after recognition are stacked and welded and typeset.

[0110] In the prior art, 80% of the defective component solar cells are caused by virtual soldering. Through the defect recognition model, the intelligent root cause analysis of welding defects is carried out, and the process operation parameters of the solar cell welding equipment are automatically adjusted in real time according to the parameter generation model, reducing the proportion of virtual soldering defects and preventing batch defects. The introduction of deep learning algorithms, random forest models, autoregressive time series models, etc. improves the AI recognition rate, and the root cause analysis of defective products is carried out in real time, dynamically adjusting the equipment operation process parameters, reducing the defective rate of series welding of solar cells, and reducing the downtime of production equipment.

[0111] The above embodiments are applicable to the online intelligent detection of defects and the intelligent adjustment of equipment operation parameters in the welding production process of photovoltaic module solar cells. Aiming at the accurate identification of defective conditions by the AI system in the prior art, in-depth analysis of defect problems, and design of correction schemes for virtual soldering defects. The correction process involves automatically sending the optimal parameter configuration obtained based on intelligent analysis to the production machine to achieve immediate improvement of defects. In addition, the system records the results of each correction operation in detail to continuously iterate and optimize the future parameter adjustment strategy.

[0112] In the prior art, the low detection and recognition rate of the component cell welding process is a pain point in the photovoltaic industry. Through deep learning algorithms, random forest models, autoregressive time series models, etc., the above embodiments improve the accuracy of the AI algorithm model, gradually eliminate manual rejudgment, and all are detected in real time by the online AI of the digital system. When the same type of defect appears continuously three times in the AI quality inspection, root cause analysis is carried out through AI, and the optimized process parameters of the production equipment are sent to the machine tool to automatically adjust the operating state of the equipment in real time, solving the production capacity loss caused by stopping the machine for testing and adjusting the equipment. The defective products are removed in real time through an automated mechanism to prevent them from flowing to the next process.

[0113] Compared with the prior art, the above embodiments add a variety of advanced AI algorithms on the basis of the original welding equipment software and hardware. Through deep learning algorithms, random forest models, autoregressive time series models, etc., the accuracy of the AI algorithm model is improved. Manual rejudgment is eliminated. When the AI detects defects, root cause analysis is carried out in real time, and the optimized process parameters of the equipment are generated in real time to adjust the machine tool in real time, effectively avoiding the generation of batch defects.

[0114] Based on the recommendation algorithm and the automatic tuning technology route, with the machine tool equipment parameters and the virtual soldering defect rate as the input and the reduction of the virtual soldering defect rate as the goal, the process principle and the machine tool state are comprehensively considered to realize the construction of the data model. Based on the supervised machine learning algorithm, within the range allowed by the actual process conditions, the reduction of the battery string defects (virtual soldering) is realized, the goal of the virtual soldering defect rate being less than 0.27% is achieved, and the stable growth of the good product rate is realized. According to the AI detection results and the feedback defect rate results of the automatic tuning, the unqualified rate of the battery components caused by virtual soldering is reduced by about 79%, and the income in the covered area is increased by about 64.77 million yuan per year.

[0115] From the above description, it can be seen that the above embodiments of the present invention achieve the following technical effects:

[0116] The method for improving the yield of the battery module in the present application first determines the defect type of the welded battery module and the current operating parameters of the welding equipment; then inputs the defect type of the welded battery module and the current operating parameters of the welding equipment into the parameter generation model to obtain the preliminary operating parameters of the welding equipment; then determines whether the preliminary operating parameters of the welding equipment meet the operating conditions of the welding equipment, and the operating conditions are the preset conditions for the welding equipment to operate without faults; finally, when the preliminary operating parameters of the welding equipment meet the operating conditions of the welding equipment, the preliminary operating parameters are used to control the welding equipment to weld the un-welded battery modules, so as to improve the yield of the battery modules. This method cancels manual re-judgment. In the case of defective battery modules, it conducts real-time root cause analysis of the defects and generates real-time equipment optimization process parameters. It sends the production equipment optimization process parameters to the machine tool and automatically adjusts the equipment operating state in real time, effectively avoiding batch defects and production capacity losses caused by downtime for testing and adjusting the equipment, and solving the problems of low efficiency and affecting the output of battery modules in the prior art methods for improving the yield of welded battery modules.

[0117] The foregoing are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for improving the yield rate of a battery assembly, characterized in that: include: Determine the defect type of the welded battery assembly and the current operating parameters of the welding equipment; Inputting the defect type of the welded battery assembly and the current operating parameters of the welding equipment into a parameter generation model to obtain the preliminary operating parameters of the welding equipment; Determining whether the preparatory operating parameters of the welding equipment meet the operating conditions of the welding equipment, wherein the operating conditions are preset conditions for trouble-free operation of the welding equipment; When the preparatory operating parameters of the welding equipment meet the operating conditions of the welding equipment, the preparatory operating parameters are used to control the welding equipment to weld the unwelded battery assemblies to improve the yield rate of the battery assemblies.

2. The method for improving the yield rate of battery components according to claim 1, characterized in that: The preparatory operation parameters of the welding equipment at least include a preparatory welding temperature of the welding equipment and a preparatory welding time of the welding equipment, and judging whether the preparatory operation parameters of the welding equipment meet the operation conditions of the welding equipment includes: determining a calculated power of the welding device according to a pre-welding temperature of the welding device and a pre-welding time of the welding device; In a case where the calculated power of the welding equipment is less than or equal to the preset power, determining that the preliminary operation parameters of the welding equipment meet the operation conditions of the welding equipment; In a case where the calculated power of the welding device is greater than the preset power, it is determined that the preliminary operation parameters of the welding device do not meet the operation conditions of the welding device.

3. The method for improving the yield rate of battery components according to claim 1, characterized in that: Determine the defect type of the welded battery assembly and the current operating parameters of the welding equipment, including: Obtaining defect types of the plurality of welded battery assemblies, and determining defect locations of the respective welded battery assemblies; If within a preset time range, the defect types of preset welded battery assemblies are all target types, and the defect positions of the preset welded battery assemblies are the same, the defect type of the welded battery assemblies is determined as the target type, and the current operating parameters of the welding equipment are determined.

4. The method for improving the yield rate of battery components according to claim 1, characterized in that: The method further comprises: When the preparatory operation parameters of the welding equipment do not meet the operation conditions of the welding equipment, a secondary test is performed on the preparatory operation parameters of the welding equipment to obtain a test result, wherein the test result indicates whether the welding equipment is fault-free when operating with the preparatory operation parameters; The parameter generation model is corrected using the defect type of the welded battery assembly, the current operating parameters of the welding equipment, the preliminary operating parameters and the test results to obtain a corrected generation model, and when the test result shows that the welding equipment has no faults when operating with the preliminary operating parameters, the preliminary operating parameters are used to control the welding equipment to weld the unwelded battery assemblies.

5. The method for improving the yield rate of battery components according to claim 1, characterized in that: Before determining the defect type of the welded battery assembly and the current operating parameters of the welding equipment, the method further includes: Acquiring an electroluminescent image of the welded battery assembly; The electroluminescent image of the welded battery assembly is input into a defect recognition model to obtain defect parameters of the welded battery assembly, wherein the defect parameters are no defect or the defect type of the welded battery assembly. The defect recognition model includes multiple sub-models, wherein different defect types are identified using different sub-models.

6. The method for improving the yield rate of battery components according to claim 5, characterized in that: The multiple sub-models include at least a deep learning algorithm, a random forest model, and an autoregressive time series model. The electroluminescent image of the welded battery assembly is input into the defect recognition model to obtain the defect parameters of the welded battery assembly, including: inputting the electroluminescent image of the welded battery assembly into a first identification sub-model; In the case where the first identification sub-model cannot identify whether the battery assembly has defects, inputting the electroluminescent image of the welded battery assembly into the second identification sub-model; In the case that the second identification sub-model cannot identify whether the battery assembly has defects, the electroluminescent image of the welded battery assembly is input into the third identification sub-model to obtain defect parameters of the welded battery assembly; Among them, the first identification sub-model, the second identification sub-model, and the third identification sub-model are respectively one of the deep learning algorithm, the random forest model, and the autoregressive time series model, and the first identification sub-model, the second identification sub-model, and the third identification sub-model are not the same.

7. The method for improving the yield rate of battery components according to claim 5, characterized in that: After inputting the electroluminescent image of the welded battery assembly into a defect recognition model to obtain defect parameters of the welded battery assembly, the method further includes: Performing a secondary inspection on the battery assembly to obtain an inspection result, wherein the inspection result indicates whether the battery assembly has a defect and the type of defect when a defect exists; When the defect parameters of the welded battery assembly are defect-free and the detection result indicates the presence of defects in the battery assembly and the defect type of the welded battery assembly, the defect parameters of the welded battery assembly and the detection result are used to correct the defect recognition model to obtain a corrected recognition model.

8. The method for improving the yield rate of battery components according to claim 1, characterized in that: The current operating parameters of the welding equipment include the current welding temperature of the welding equipment, the current welding time of the welding equipment, and the current lamp power of the welding equipment. The preparatory operating parameters of the welding equipment include the preparatory welding temperature of the welding equipment, the preparatory welding time of the welding equipment, and the preparatory lamp power of the welding equipment. Among them, when the defect type of the welded battery assembly is a cold weld, the size relationship between the current operating parameters of the welding equipment and the preparatory operating parameters of the welding equipment satisfies at least one of the following: the preparatory lamp tube power of the welding equipment is greater than the current lamp tube power of the welding equipment, the preparatory welding temperature of the welding equipment is greater than the current welding temperature of the welding equipment, and the preparatory welding time of the welding equipment is greater than the current welding time of the welding equipment.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method for improving the yield rate of a battery assembly as described in any one of claims 1 to 8.

10. An electronic device, characterized in that: include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for improving the yield of a battery component as described in any one of claims 1 to 8.

Images