Battery welding defect detection system, method and equipment and storage medium
The welding image is preprocessed and detected and analyzed through deep learning models, and the problems of low efficiency, low accuracy and high cost of battery welding defect detection are solved, achieving efficient and accurate welding defect detection.
Patent Information
- Application Number
- CN202410999966.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-24
- Publication Date
- 2024-11-12
AI Technical Summary
Existing battery welding defect detection relies on manual microscopy to detect solder joints that are inefficient, low in accuracy, high in cost, and difficult to detect fine cracks and light-interferenced welding joints.
A battery welding defect detection system is adopted, including a collection and processing module, defect statistics module and detection and analysis module. The welding images are preprocessed, classified and detected and analyzed through deep learning models to generate welding defect detection results, improve detection efficiency and accuracy sex.
It significantly improves the efficiency and accuracy of welding defect detection, reduces production costs, and can effectively identify fine cracks and photometric three-dimensional technology to meet industrial-grade inspection requirements.
Smart Images

Figure CN118941532A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of battery technology, and in particular to a battery welding defect detection system, method, device and storage medium. Background Art
[0002] In the battery manufacturing industry, welding defect detection of large square batteries plays a role as a link, connecting the production of raw materials and the manufacturing process of complete batteries. The welding quality of large square batteries directly affects the safety and performance of finished batteries. However, the traditional welding defect detection is currently widely used, which mainly relies on manual microscope inspection one by one. Manual inspection not only takes a lot of time, but also requires inspectors to carefully check the welding points one by one, which seriously restricts production efficiency. In addition, long-term manual observation is prone to fatigue, which in turn affects the detection accuracy and cannot ensure the consistency and reliability of the detection results. In addition, for fine cracks and light-interfered welding points, the human eye has limited resolution and it is difficult to meet industrial-level detection requirements.
[0003] Therefore, existing welding defect detection technologies have obvious shortcomings in efficiency, accuracy and cost control. Summary of the invention
[0004] The purpose of the embodiments of the present application is to provide a battery welding defect detection system, method, device and storage medium to solve the current problems of low efficiency, low accuracy and high cost of battery welding defect detection.
[0005] In a first aspect, the present invention provides a battery welding defect detection system, comprising:
[0006] An acquisition and processing module, used for acquiring a welding image of a battery welding object, and preprocessing the welding image to obtain a processed welding image;
[0007] A defect statistics module, used for classifying and counting the processed welding images based on a target detection model to obtain a plurality of battery candidate frames;
[0008] The detection and analysis module is used to obtain the corresponding battery detection result according to the preset threshold and the battery candidate box through the target detection model.
[0009] In an optional embodiment, it also includes:
[0010] An acquisition module, used to acquire a battery welding image training set, a battery welding image verification set, and a battery welding image test set for training;
[0011] A construction module, used to construct an initial target detection model according to the welding image training set based on a preset deep learning model;
[0012] A training module is used to train the initial target detection model according to the welding image verification set and the welding image test set to obtain the target detection model.
[0013] In an optional implementation, the acquisition and processing module includes:
[0014] The processing unit is used to perform cropping, denoising, enhancement and standardization processing on the welding image through a preset image processing algorithm to obtain the processed welding image.
[0015] In an optional implementation manner, the defect statistics module includes:
[0016] A candidate unit, used to generate a plurality of candidate frames of corresponding quantity and size according to the processed welding image and a preset size by a sliding window strategy;
[0017] The classification unit is used to classify each of the candidate boxes through the target detection model to obtain multiple categories of candidate boxes.
[0018] In an optional embodiment, the detection and analysis module includes:
[0019] A detection unit, configured to detect each of the candidate frames using the target detection model to obtain corresponding image features, and analyze the image features to obtain corresponding defect types and defect locations;
[0020] A regression unit, used to perform regression analysis according to the defect type and the defect position through a preset function to obtain a corresponding defect probability distribution;
[0021] The analyzing unit is used to analyze all candidate frames based on a preset threshold and the defect probability distribution to obtain the battery detection result.
[0022] In an optional embodiment, it also includes:
[0023] The result output module is used to display corresponding defect information according to the defect type, the defect location and the battery detection result.
[0024] In an optional embodiment, it also includes: a data updating module, which is used to bind the battery detection result with the battery cell data of the corresponding battery and update it to a preset database.
[0025] In a second aspect, the present invention provides a battery welding defect detection method, comprising:
[0026] Collecting a welding image of a battery welding object, and preprocessing the welding image to obtain a processed welding image;
[0027] Based on the target detection model, the processed welding image is classified and counted to obtain multiple battery candidate frames;
[0028] The target detection model is used to obtain a corresponding battery detection result according to a preset threshold and the battery candidate frame.
[0029] In a third aspect, the present invention provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the battery welding defect detection method as described in the aforementioned embodiment is implemented.
[0030] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the battery welding defect detection method as described in the aforementioned embodiment is executed.
[0031] The above technical solution adopted in this application has the following effects:
[0032] The present application provides a battery welding defect detection system, method, device and storage medium, the system comprising: an acquisition and processing module, used to acquire the welding image of the battery welding object, pre-process the welding image, and obtain the processed welding image; a defect statistics module, used to classify and count the processed welding image based on the target detection model, and obtain multiple battery candidate frames; a detection and analysis module, used to obtain the corresponding battery detection result according to the preset threshold and the battery candidate frame through the target detection model. The present application performs defect detection on the processed battery welding image based on the target detection model, thereby significantly improving the detection efficiency and the accuracy of the results, and can also effectively reduce the production cost. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.
[0034] Figure 1 A schematic diagram of the structure of a battery welding defect detection system provided in an embodiment of the present application;
[0035] Figure 2 Another structural schematic diagram of a battery welding defect detection system provided in an embodiment of the present application;
[0036] Figure 3 Another structural schematic diagram of a battery welding defect detection system provided in an embodiment of the present application;
[0037] Figure 4 A further structural schematic diagram of a battery welding defect detection system provided in an embodiment of the present application;
[0038] Figure 5 A further structural schematic diagram of a battery welding defect detection system provided in an embodiment of the present application;
[0039] Figure 6 A schematic diagram of the process flow of a battery welding defect detection method provided in an embodiment of the present application.
[0040] Icon: 10-battery welding defect detection system; 100-acquisition and processing module; 200-defect statistics module; 300-detection and analysis module; 101-processing unit; 201-candidate unit; 202-classification unit; 301-detection unit; 302-regression unit; 303-analysis unit; 501-acquisition module; 502-construction module; 503-training module. DETAILED DESCRIPTION
[0041] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application.
[0042] The components of the embodiments of the present invention generally described and shown in the drawings herein may be arranged and designed in a variety of different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.
[0043] Hereinafter, the terms "including", "having" and their cognates, which may be used in various embodiments of the present invention, are intended only to indicate specific features, numbers, steps, operations, elements, components or combinations of the foregoing items, and should not be understood as first excluding the existence of one or more other features, numbers, steps, operations, elements, components or combinations of the foregoing items or adding the possibility of one or more features, numbers, steps, operations, elements, components or combinations of the foregoing items.
[0044] Furthermore, the terms “first”, “second”, “third”, etc. are merely used for distinguishing descriptions and are not to be understood as indicating or implying relative importance.
[0045] Unless otherwise defined, all terms (including technical terms and scientific terms) used herein have the same meanings as those generally understood by those skilled in the art to which the various embodiments of the present invention belong. The terms (such as those defined in generally used dictionaries) will be interpreted as having the same meanings as the contextual meanings in the relevant technical field and will not be interpreted as having idealized meanings or overly formal meanings unless clearly defined in the various embodiments of the present invention.
[0046] Example 1
[0047] In current industrial production, welding defect detection mainly relies on manual microscope inspection one by one, which not only limits the improvement of production efficiency, but also cannot guarantee the accuracy and consistency of the detection results.
[0048] In response to the above technical problems, an embodiment of the present application provides a battery welding defect detection system, which aims to solve the current technical problems of low efficiency, low accuracy and high cost of battery welding defect detection.
[0049] Reference Figure 1 , Figure 1 It is a structural schematic diagram of the battery welding defect detection system provided by the present application. The battery welding defect detection system 10 provided in the embodiment of the present application includes: an acquisition and processing module 100, a defect statistics module 200 and a detection and analysis module 300. Each module is described below.
[0050] The acquisition and processing module 100 is used to acquire a welding image of a battery welding object, and pre-process the welding image to obtain a processed welding image.
[0051] Specifically, the welding images of the battery welding object include but are not limited to the welding images of the battery welding by the sealing nails. In this embodiment, a high-resolution 2.5D camera device is preferably used to collect images of the battery welding part. The 2.5D camera device can capture rich depth information of the battery, which is suitable for welding defect detection (such as cracks and explosion points). In addition, in order to obtain high-quality welding images, the welding image acquisition device is equipped with the following functions, including a motor to control the camera lens, an automatic focus function to adjust the focal length in real time to ensure a clear welding image, an automatic exposure function to automatically adjust the camera exposure time according to the light intensity of the shooting environment to ensure that the welding image is not too bright or too dark, and a photometric stereo technology that further improves the accuracy of obtaining depth information by shooting at multiple lighting angles, thereby more accurately detecting surface defects.
[0052] In this embodiment, since the welding defect detection of large square batteries plays a linking role, connecting the production of raw materials and the manufacturing process of complete batteries, it is mainly aimed at large square batteries. The preprocessing includes image cropping operation, image denoising operation, image enhancement operation and image standardization operation, specifically referring to:
[0053] Image cropping operation: By using a sliding window cropping algorithm, the images are cropped to the same size and the dataset is expanded to ensure that each image focuses on the battery welding part.
[0054] Image denoising operation: Use median filtering and Gaussian filtering algorithms to remove noise and interference in the image, such as granular noise and high-frequency interference.
[0055] Image enhancement operation: By adjusting the brightness, contrast and color balance of the image, the visualization effect and edge information of the image are enhanced to ensure that the details of the defects in the image can be better captured.
[0056] Image normalization: This operation ensures the consistency of data input to the deep learning model by adjusting the images to a uniform size and orientation, thereby eliminating differences in brightness and color between different images.
[0057] In this embodiment, an industrial camera or other imaging device is used to photograph the battery welding part to obtain a welding image of the battery welding object with sufficiently high resolution and clear details, and then the acquired welding image is processed. For example, when the original welding image is in color, the original welding image is converted into a grayscale image to simplify subsequent processing steps; median filtering, Gaussian filtering and other methods are used to remove noise in the welding image; the brightness, contrast and color balance are adjusted, and the details of the welding image are enhanced using techniques such as histogram equalization; the welding image is cropped using a sliding window algorithm to focus on the battery welding area, and the data set is expanded at the same time. Edge detection, contour extraction and other processing methods can also be used to make the welding image data conducive to subsequent analysis or training of learning models.
[0058] Specifically, refer to Figure 2 In one implementation of this embodiment, the acquisition and processing module 100 includes: a processing unit 101, which is used to crop, denoise, enhance and standardize the welding image through a preset image processing algorithm to obtain the processed welding image.
[0059] In this embodiment, the collected welding images are preprocessed, including but not limited to image cropping, image denoising, image enhancement, and image standardization operations, so as to ensure that each welding image is focused on the battery welding area, remove noise and interference in the welding image, ensure that the details of the defects in the welding image can be better captured, and eliminate the brightness and color differences between different welding images.
[0060] Please refer again Figure 1 The battery welding defect detection system 10 also includes: a defect statistics module 200, which is used to classify and count the processed welding images based on the target detection model to obtain multiple battery candidate frames.
[0061] Specifically, in the field of industrial automation and quality control, by building a target detection model, it is possible to automatically learn to extract features from welding images, and then identify and classify various types of defects. It is an effective tool for detecting product surface defects.
[0062] In this embodiment, the battery parts in the welding image are located and counted through a pre-built target detection model. First, a large number of candidate boxes are generated in the welding image to cover all areas that may contain battery welds. Then, a sliding window strategy is used to generate a reasonable number and size of candidate boxes according to the size of the welding image and the typical size of the battery, thereby effectively performing defect detection on the battery welding part and avoiding abnormal judgment errors.
[0063] Specifically, refer to Figure 3 In one implementation of this embodiment, the defect statistics module 200 includes: a candidate unit 201, which is used to generate a plurality of candidate boxes of corresponding quantity and size according to the processed welding image and the preset size through a sliding window strategy.
[0064] Specifically, the sliding window strategy is a target detection technique that searches for potential target areas by moving windows of different sizes on the welding image. First, the window size to be used is determined according to the preset size. The window size can be a fixed size or a series of different sizes to detect targets of different scales. Then, the first window is placed from a corner of the welding image (usually the upper left corner) to confirm the initial window position, and then the window is moved horizontally or vertically on the welding image according to the preset stride to generate multiple candidate boxes of corresponding numbers and sizes, which is conducive to the subsequent model to specifically analyze the defect detection situation.
[0065] In this embodiment, through the sliding window strategy, according to the processed welding image and the preset size, targets of different scales can be detected, thereby improving the comprehensiveness of the detection, and then multiple candidate boxes of corresponding numbers and sizes are generated to realize intelligent screening of candidate areas to reduce the computational complexity and optimize the computational efficiency.
[0066] The classification unit 202 is used to classify each of the candidate boxes through the object detection model to obtain multiple categories of candidate boxes.
[0067] In this embodiment, all candidate boxes are input into the target detection model. For each candidate box, the corresponding features are extracted. The features may be color histograms, gradient direction histograms, deep learning features, etc. All candidate boxes are summarized and sorted according to different features, and candidate boxes that meet a preset threshold are selected by non-maximum suppression (NMS). For example, if the same target is detected in multiple adjacent candidate box windows, the non-maximum suppression algorithm is used to remove duplicate detection results and retain the most likely candidate box, thereby obtaining multiple types of candidate boxes.
[0068] Please refer again Figure 1 The battery welding defect detection system 10 also includes: a detection analysis module 300, which is used to obtain a corresponding battery detection result according to a preset threshold and the battery candidate frame through the target detection model.
[0069] In this embodiment, a trained target detection model is used to identify whether there are defects in the welding image, the type of defects, and the location information of the defects, specifically including: using the trained target detection model to perform defect detection on the collected welding image to obtain corresponding image features, and then inputting the image features into the model for analysis, and obtaining the battery detection results for detection based on the preset function through forward propagation, including the defect type, defect location and the probability distribution of each defect, thereby being able to simultaneously solve the defect type detection problem and the defect location estimation problem.
[0070] Specifically, refer to Figure 4 In one implementation of this embodiment, the detection and analysis module 300 includes: a detection unit 301, which is used to detect each candidate box through the target detection model to obtain corresponding image features, analyze the image features to obtain corresponding defect types and defect locations.
[0071] Specifically, image features refer to the features of welding images extracted and processed using a deep learning model, including but not limited to edge, texture, shape and other information.
[0072] In this embodiment, by training the target detection model and using the intermediate layer of the target detection model, feature extraction is performed on the welding image area within each candidate frame, and the extracted image features are input into the fully connected layer for feature pooling, normalization, nonlinear transformation and other operations to enhance the distinctiveness and robustness of the features. The processed features are then identified and classified to identify specific types of defects (e.g., cracks, scratches, stains, etc.), and the exact location of the defect in each candidate frame is predicted. The positioning is refined by adjusting the coordinates of the bounding box, and finally the detected defect type and defect location are obtained, thereby reducing manual intervention and improving detection speed and accuracy.
[0073] A regression unit 302 is used to perform regression analysis according to the defect type and the defect position through a preset function to obtain a corresponding defect probability distribution;
[0074] Specifically, defect probability distribution refers to the probability distribution used to describe the number of defects that may occur in the inspected batteries, so as to predict and understand the defect rate in the production process for improvement and optimization.
[0075] In this embodiment, a regression analysis is performed according to the defect type and the defect location through a preset function to obtain the corresponding defect probability distribution. Preferably, the output layer of the deep learning model is connected to the softmax function, and the model output is converted into the probability distribution of various defects through a normalized model and output, so that the output of the model is more intuitive and easy to understand and explain.
[0076] The analyzing unit 303 is used to analyze all candidate frames based on a preset threshold and the defect probability distribution to obtain the battery detection result.
[0077] Specifically, the preset threshold is used to judge the credibility of the target detection model results. For example, for each candidate box, the model will give a confidence score, indicating the possibility that the corresponding candidate box belongs to a certain type.
[0078] In this embodiment, for the defect probability distribution of each candidate frame, the probability scores corresponding to different types are calculated according to the defect probability distribution, and the probability score of each candidate frame is compared with the preset threshold. When the score of the candidate frame is higher than the preset threshold, the corresponding candidate frame is considered to be a valid detection result; otherwise, the currently calculated candidate frame is ignored. For the valid candidate frames of the current battery welding image, the type with the highest score is selected as the detection state of the battery welding image. For example, if all the valid candidate frames in the battery welding image have the highest score in the "damaged sealing nail" category, then the detection state of the corresponding battery welding image is a damaged sealing nail, and the battery welding part is also in a damaged state, thereby forming a battery detection result based on the preset threshold and defect probability distribution, including the detection state of the welding part in the battery welding image, the defect type, the defect location and the defect probability distribution.
[0079] Reference Figure 5 The battery welding defect detection system 10 provided in the embodiment of the present application also includes: an acquisition module 501, a construction module 502 and a training module 503. Each module is described below.
[0080] The acquisition module 501 is used to acquire a battery welding image training set, a battery welding image verification set and a battery welding image test set for training.
[0081] In this embodiment, a large amount of welding image data of battery welding objects is collected, and the type, location and bounding box of each welding defect are pre-labeled to form a battery welding image training set, a battery welding image verification set and a battery welding image test set, which are used for model training, verification and testing, respectively.
[0082] The construction module 502 is used to construct an initial target detection model according to the welding image training set based on a preset deep learning model.
[0083] In this embodiment, a model architecture suitable for welding image classification and defect detection is first selected to build a target detection model, preferably a YOLOv8 (Ultralytics YOLOv8) model is selected as the basic model, and an initial target detection model is constructed to receive an input welding image and have multiple output layers. One output layer is used to identify different defect types, and the other output layer is used to predict the specific location of the defect. Then, specific improvements are made to the initial target detection model and parameters, especially adding an adaptive enhancement module for reflective point and crack recognition, adding a medium target output layer, and optimizing the feature extraction and prediction modules of the network to improve detection efficiency and accuracy, including: removing reflective points, for example, by using adaptive filtering technology to effectively remove high light reflection areas in welding images; detecting fine cracks requires high-resolution features and fine texture information. For example, through a high-resolution network structure and Attention mechanism, detail information is captured to improve the recognition accuracy of small objects, especially in the detection of small objects and fine cracks; adding a medium target output layer, and optimizing the feature extraction and prediction modules of the network. For example, we introduced the high-resolution feature extraction module of HRNet to capture more detailed information; added a medium-sized target output layer; added an output layer in the network structure to enhance the detection capability of medium-sized targets; introduced the Attention mechanism; embedded SE (Squeeze-and-Excitation) blocks in the feature map of YOLOv8 to enhance the feature weights of specific areas; set hyperparameters such as learning rate, batch size, and number of network layers. After multiple experiments and verifications, we optimized these parameters to improve the performance and generalization ability of the model.
[0084] The training module 503 is used to train the initial target detection model according to the welding image verification set and the welding image test set to obtain the target detection model.
[0085] In this embodiment, the model is trained using a welding image validation set and a welding image test set, and back propagation and stochastic gradient descent (SGD) are used to optimize parameters to minimize the loss function. The specific steps of parameter optimization are: through a large number of experiments on the data set, the results are compared, the results of different image enhancement methods, learning rates, optimizers, etc. are observed, and the optimal model is obtained by migrating to a large data set. The loss function is specifically: 1. Bounding Box Regression Loss: The difference between the predicted bounding box and the true bounding box is calculated by using GIoU (Generalized IoU). 2. Classification Loss: The difference between the predicted category and the true category is calculated using cross-entropy loss. 3. Confidence Loss: The confidence of the predicted object and the gap between the actual object are evaluated using binary cross-entropy loss. Finally, data enhancement and cross-validation techniques are applied to further improve the robustness and accuracy of the model.
[0086] In this embodiment, multiple battery welding image data sets are used to train the model constructed subsequently, the model parameters are adjusted to optimize the performance, and the performance of the model on unseen data is evaluated to ensure its generalization ability. Mainly, an initial target detection model is constructed based on the welding image training set based on the YOLOv8 model, and then the initial target detection model is trained based on the welding image verification set and the welding image test set. The initial target detection model automatically extracts features that help distinguish different defect types from the welding image verification set and the welding image test set. Based on the extracted features, the initial target detection model predicts whether there are defects in the welding image and the specific type of the defect, and then confirms the confidence of the model detection result based on the welding image verification set and the welding image test set, and obtains the target detection model for detecting subsequent battery welding image defects. In addition, the target detection model is constructed based on the deep learning model, which improves the detection ability of the target detection model for reflective explosion points and micro cracks, and realizes the automatic and efficient completion of the battery welding defect detection work, which not only improves the detection speed, but also improves the detection accuracy, thereby significantly improving the production efficiency.
[0087] The battery welding defect detection system 10 provided in the embodiment of the present application further includes: a result output module, which is used to display corresponding defect information according to the defect type, the defect location and the battery detection result.
[0088] In this embodiment, the battery detection results, including the detection status, defect type, defect location and defect probability distribution of the welding image of the battery welding object, are output and displayed on the user interface to provide real-time detection information. For example, the number and type of battery defects are clearly displayed in digital form to intuitively present the detection results. The defect information is superimposed on the battery welding image using image processing software or professional tools, the detected welding defect location is marked on the welding image, and detailed defect type, defect number and confidence information are displayed, so that the operator can quickly evaluate the battery quality according to the type and severity of the defect, so as to adjust the production process or equipment parameters, help predict potential problems and optimize production strategies.
[0089] The battery welding defect detection system 10 provided in the embodiment of the present application further includes: a data update module, which is used to bind the battery detection results with the cell data of the corresponding battery and update them to a preset database.
[0090] In this embodiment, the preset database is preferably a local database or a cloud server database. The corresponding data is identified by the QR code on the surface of the battery, and the battery test results are bound to the specific battery cell data and stored in the preset database. At the same time, it is uploaded to the production MES (Manufacturing Execution System) manufacturing execution system to achieve data traceability. Then, an efficient scanning device is used to quickly and accurately identify the QR code on the surface of the battery to obtain the unique identification of the battery. The test results are then bound to the battery identification and saved in the local database for subsequent query and traceability. Finally, the test data is uploaded to the MES system in real time to ensure that the automated processing of subsequent steps of the production line can obtain the latest test results and traceability information in a timely manner.
[0091] This embodiment discloses a battery welding defect detection system, which includes: an acquisition and processing module, which is used to acquire welding images of battery welding objects, pre-process the welding images, and obtain processed welding images; a defect statistics module, which is used to classify and count the processed welding images based on a target detection model to obtain multiple battery candidate frames; a detection and analysis module, which is used to obtain corresponding battery detection results according to a preset threshold and the battery candidate frames through the target detection model. This application performs defect detection on processed battery welding images based on the target detection model, thereby significantly improving detection efficiency and accuracy of results, while also effectively reducing production costs.
[0092] Example 2
[0093] A battery welding defect detection method is provided in an embodiment of the present invention, which is applied to the battery welding defect detection system provided in Example 1.
[0094] like Figure 6 As shown, Figure 6 1 is a flow chart of a battery welding defect detection method, which includes:
[0095] Step S100, collecting a welding image of a battery welding object, and preprocessing the welding image to obtain a processed welding image;
[0096] Step S200, classifying and counting the processed welding image based on the target detection model to obtain a plurality of battery candidate frames;
[0097] Step S300, obtaining a corresponding battery detection result through the target detection model according to a preset threshold and the battery candidate box.
[0098] The battery welding defect detection method provided in this embodiment is applied to the battery welding defect detection system provided in Example 1, and can realize the corresponding functions of the battery welding defect detection system and achieve the corresponding effects. To avoid repetition, it will not be repeated here.
[0099] In summary, the present invention also provides a battery welding defect detection method, which includes: collecting a welding image of a battery welding object, preprocessing the welding image to obtain a processed welding image; classifying and counting the processed welding image based on a target detection model to obtain multiple battery candidate frames; and obtaining corresponding battery detection results based on a preset threshold and the battery candidate frame through the target detection model. The present application performs defect detection on the processed battery welding image based on the target detection model, thereby significantly improving the detection efficiency and the accuracy of the results, while also effectively reducing production costs.
[0100] Example 3
[0101] In addition, the present application provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program runs on the processor, the battery welding defect detection method provided in Example 2 is executed.
[0102] The electronic device provided in this embodiment can implement the battery welding defect detection method provided in Example 2, and will not be described again here to avoid repetition.
[0103] Example 4
[0104] The present application also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the battery welding defect detection method provided in Example 2 is implemented.
[0105] In this embodiment, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.
[0106] The computer-readable storage medium provided in this embodiment can implement the battery welding defect detection method provided in Example 2, and will not be described again here to avoid repetition.
[0107] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.
[0108] In addition, the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0109] Furthermore, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist independently, or two or more modules may be integrated to form an independent part.
[0110] It should be noted that if the function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can essentially be embodied in the form of a software product, or the part that contributes to the prior art or the part of the technical solution. The computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM) random access memory (RAM), disk or optical disk, and other media that can store program codes.
[0111] In this document, relational terms such as first and second, etc. are used merely to distinguish one entity or operation from another entity or operation, but do not necessarily require or imply any such actual relationship or order between these entities or operations.
[0112] The above description is only an embodiment of the present application and is not intended to limit the protection scope of the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims
1. A battery welding defect detection system, characterized in that: include: An acquisition and processing module, used for acquiring a welding image of a battery welding object, and preprocessing the welding image to obtain a processed welding image; A defect statistics module, used for classifying and counting the processed welding images based on a target detection model to obtain a plurality of battery candidate frames; The detection and analysis module is used to obtain the corresponding battery detection result according to the preset threshold and the battery candidate box through the target detection model.
2. The battery welding defect detection system according to claim 1, characterized in that: Also includes: An acquisition module, used to acquire a battery welding image training set, a battery welding image verification set, and a battery welding image test set for training; A construction module, used to construct an initial target detection model according to the welding image training set based on a preset deep learning model; A training module is used to train the initial target detection model according to the welding image verification set and the welding image test set to obtain the target detection model.
3. The battery welding defect detection system according to claim 2, characterized in that: The acquisition and processing module comprises: The processing unit is used to perform cropping, denoising, enhancement and standardization processing on the welding image through a preset image processing algorithm to obtain the processed welding image.
4. The battery welding defect detection system according to claim 2, characterized in that: The defect statistics module includes: A candidate unit, used to generate a plurality of candidate frames of corresponding quantity and size according to the processed welding image and a preset size by a sliding window strategy; The classification unit is used to classify each of the candidate boxes through the target detection model to obtain multiple categories of candidate boxes.
5. The battery welding defect detection system according to claim 2, characterized in that: The detection and analysis module comprises: A detection unit, configured to detect each of the candidate frames using the target detection model to obtain corresponding image features, and analyze the image features to obtain corresponding defect types and defect locations; A regression unit, used to perform regression analysis according to the defect type and the defect position through a preset function to obtain a corresponding defect probability distribution; The analyzing unit is used to analyze all candidate frames based on a preset threshold and the defect probability distribution to obtain the battery detection result.
6. The battery welding defect detection system according to claim 5, characterized in that: Also includes: The result output module is used to display corresponding defect information according to the defect type, the defect location and the battery detection result.
7. The battery welding defect detection system according to claim 1, characterized in that: Also includes: The data updating module is used to bind the battery detection result with the cell data of the corresponding battery and update it to a preset database.
8. A battery welding defect detection method, characterized in that: include: Collecting a welding image of a battery welding object, and preprocessing the welding image to obtain a processed welding image; Based on the target detection model, the processed welding image is classified and counted to obtain multiple battery candidate frames; The target detection model is used to obtain a corresponding battery detection result according to a preset threshold and the battery candidate frame.
9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the battery welding defect detection method as claimed in claim 8 is implemented.
10. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by the processor, the battery welding defect detection method as claimed in claim 8 is executed.
Citation Information
Cited By
Deep learning-based detection result display method and device, and storage medium
CN119722849A
Multi-channel data-based labeling method and labeling system
CN120340033A
A labeling method and system based on multi-channel data
CN120340033B
Welded product quality detection method and device and quality detection equipment
CN120598950A
