Battery welding defect detection system, method, device, and storage medium

Abstract

The present disclosure provides a battery welding defect detection system, method, device and storage medium, which can solve the problems of low efficiency, low accuracy and high cost in the existing battery welding defect detection.SOLUTION: This application provides a battery welding defect detection system and method, a device, and a storage medium. An acquisition processing module, configured to acquire a soldering image of a battery to be soldered, and pre-process the soldering image to obtain a processed soldering image; a defect statistics module, configured to perform classification statistics on the processed soldering image based on a target detection model to obtain a plurality of candidate battery boxes; and a detection analysis module, configured to obtain a corresponding battery detection result based on a preset threshold and the candidate battery boxes through the target detection model.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present application is in the field of battery technology, and specifically relates to a battery weld defect detection system, method, apparatus, and storage medium. [Background technology]

[0002] In the battery manufacturing industry, detecting welding defects in large prismatic batteries plays a vital role in connecting the manufacturing process of raw materials with the manufacturing process of complete batteries. The quality of welding in large prismatic batteries affects the safety and performance of the finished battery. However, traditional welding defect detection is primarily performed by manual inspection using a microscope. Manual inspection is time-consuming and requires inspectors to meticulously inspect each welding point, significantly limiting production efficiency. Prolonged observation can easily cause fatigue, affecting inspection accuracy and making it difficult to ensure the consistency and reliability of inspection results. Furthermore, the human eye's ability to distinguish between welding points with microcracks and those subject to light interference is limited, making it difficult to meet industrial inspection requirements.

[0003] As a result, conventional weld defect detection techniques suffer from significant deficiencies in efficiency, accuracy, and cost. Summary of the Invention [Problem to be solved by the invention]

[0004] An object of the embodiments of the present application is to provide a battery welding defect detection system, method, device, and storage medium that can solve the problems of low efficiency, low accuracy, and high cost in conventional battery welding defect detection. [Means for solving the problem]

[0005] In a first aspect, the present invention provides a battery welding defect detection system, comprising: an acquisition processing module configured to acquire a welding image of a battery welding object and perform pre-processing on the welding image to obtain a processed welding image; a defect statistics module configured to perform classification statistics on the processed welding image based on a target detection model to obtain a plurality of battery candidate frames; a detection analysis module configured to obtain a corresponding battery detection result based on a predetermined threshold and the battery candidate window using the target detection model.

[0006] In an alternative embodiment, the battery weld defect detection system comprises: an acquisition module configured to acquire a battery welding image training set, a battery welding image validation set, and a battery welding image test set for use in training; a construction module configured to construct an initial target detection model using the welding image training set based on a predetermined deep learning model; and a training module configured to train the initial target detection model with the welding image validation set and the welding image test set to obtain the target detection model.

[0007] In an alternative embodiment, the acquisition processing module includes a processing unit configured to perform cropping, noise removal, enhancement, and standardization processes on the welding image through a predetermined image processing algorithm to obtain the processed welding image.

[0008] In an alternative embodiment, the defect statistics module: a candidate unit configured to generate a plurality of candidate frames of corresponding number and size based on the processed welding image and a predetermined size according to a sliding window policy; a classification unit configured to classify each of the candidate frames according to the target detection model to obtain a plurality of types of candidate frames. In an alternative embodiment, the detection and analysis module: a detection unit configured to perform detection on each of the candidate frames using the target detection model to obtain corresponding image features, and to perform analysis on the image features to obtain corresponding defect types and defect locations; a regression unit configured to perform a regression analysis based on the defect types and the defect locations with a predetermined function to obtain a corresponding defect probability distribution; an analysis unit configured to analyze all candidate frames based on a predetermined threshold and the defect probability distribution to obtain the battery detection result.

[0009] In an optional embodiment, the battery welding defect detection system further comprises a result output module configured to display corresponding defect information according to the defect type, the defect location, and the battery detection result.

[0010] In an optional embodiment, the battery welding defect detection system further includes a data update module configured to associate the battery detection result with cell data of the corresponding battery and update the result in a predetermined database.

[0011] In a second aspect, the present invention provides a battery welding defect detection method, the battery welding defect detection method comprising: acquiring a welding image of a battery welding object, and performing preprocessing on the welding image to obtain a processed welding image; performing classification statistics on the processed welding image based on a target detection model to obtain a plurality of battery candidate frames; and obtaining a corresponding battery detection result based on the predetermined threshold and the battery candidate window using the target detection model.

[0012] In a third aspect, the present invention provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable by the processor, the computer program being executed by the processor to realize the battery welding defect detection method according to the above embodiment.

[0013] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored therein, the computer program being executed by a processor to perform the battery welding defect detection method according to the above embodiment. [Effects of the Invention]

[0014] The above technical solution of the present application has the following advantages:

[0015] The present application provides a battery welding defect detection system, method, device, and storage medium. The system includes an acquisition processing module configured to acquire a welding image of a battery welding object and perform preprocessing on the welding image to obtain a processed welding image, a defect statistics module configured to perform classification statistics on the processed welding image based on a target detection model to obtain a plurality of battery candidate frames, and a detection analysis module configured to obtain corresponding battery detection results using the target detection model based on a predetermined threshold and the battery candidate frames. By performing defect detection on the processed battery welding image based on the target detection model, the present application can significantly improve detection efficiency and result accuracy and effectively reduce production costs. [Brief explanation of the drawings]

[0016] In order to more clearly explain the technical solutions of the embodiments of the present application, the drawings used in the embodiments of the present application will be briefly described below. The drawings described are intended to illustrate some embodiments of the present application and are not intended to limit the scope. Those skilled in the art can obtain other related drawings based on these drawings without using inventive abilities.

[0017] [Figure 1] 1 is a first schematic diagram of a battery welding defect detection system according to an embodiment of the present application. [Figure 2] FIG. 2 is a second schematic diagram of a battery welding defect detection system according to an embodiment of the present application. [Figure 3] FIG. 3 is a third schematic diagram of a battery welding defect detection system according to an embodiment of the present application. [Figure 4] FIG. 4 is a fourth schematic diagram of a battery welding defect detection system according to an embodiment of the present application. [Figure 5] FIG. 5 is a fifth schematic diagram of a battery welding defect detection system according to an embodiment of the present application. [Figure 6] 1 is a schematic flow chart of a battery welding defect detection method according to an embodiment of the present application. DETAILED DESCRIPTION OF THE INVENTION

[0018] The technical solutions in the embodiments of the present application will be described with reference to the drawings in the embodiments of the present application.

[0019] The components in the embodiments of the present invention shown in the drawings herein can be arranged and designed in various ways. Therefore, the following detailed description of the embodiments of the present invention shown in the drawings merely illustrates selected embodiments of the present invention and does not limit the scope of the invention to be protected. All other embodiments that can be obtained by a person skilled in the art based on the embodiments of the present invention without using inventive ability also fall within the scope of protection of the present invention.

[0020] In the following, the terms "comprise", "have" and the like used in the various embodiments of the present invention refer to certain features, numbers, steps, operations, elements, components or combinations of the above elements and should not be understood to exclude the possibility that one or more features, numbers, steps, operations, elements, components or combinations of the above elements may be present or increased.

[0021] Additionally, terms such as "first," "second," and "third" are for descriptive purposes only and do not express or imply any relative importance.

[0022] Unless otherwise specified, all terms (including technical and scientific terms) used herein have the meanings commonly recognized by those skilled in the art to which the various embodiments of the present invention pertain. Unless expressly limited in the various embodiments of the present invention, the above terms (e.g., terms defined in commonly used dictionaries) have the meanings in the relevant technical field and are not to be construed as ideal or formal meanings.

[0023] Example 1 In current industrial production, welding defect detection is mainly performed by humans using microscopes one by one, which limits the improvement of production efficiency and makes it impossible to guarantee the accuracy and consistency of the detection results.

[0024] In response to the above technical problems, the embodiments of the present application provide a battery welding defect detection system that can solve the technical problems of low efficiency, low accuracy, and high cost in conventional battery welding defect detection.

[0025] Referring to Fig. 1, Fig. 1 is a first schematic diagram of a battery welding defect detection system according to the present application. The battery welding defect detection system 10 according to the embodiment of the present application includes an acquisition processing module 100, a defect statistics module 200, and a detection analysis module 300. Each module will be described below.

[0026] The acquisition processing module 100 is configured to acquire a welding image of a battery welding object, and perform pre-processing on the welding image to obtain a processed welding image.

[0027] Specifically, welding images of battery welding targets include, but are not limited to, welding images of battery welding using sealed nails. In this embodiment, a high-resolution 2.5D camera is preferably used to capture images of battery welds. The 2.5D camera can capture rich deep-level information about the battery and is suitable for detecting welding defects (e.g., cracks and explosions). To obtain high-quality welding images, the welding image capture device has the following features: a motor to control the camera lens and an autofocus function to adjust the focus in real time to ensure clear welding images; an autoexposure function to automatically adjust the camera exposure time according to the light intensity of the shooting environment to prevent welding images from becoming overly bright or dark; and photometric stereo technology is used to capture images from multiple lighting angles, further improving the accuracy of capturing deep-level information and more accurately detecting surface defects.

[0028] In this embodiment, since the detection of welding defects in large prismatic batteries plays a role of connecting the manufacturing process of raw materials with the manufacturing process of complete batteries, the preprocessing mainly for large prismatic batteries includes image cropping, image noise removal, image enhancement, and image normalization. Specifically, the operations are as follows:

[0029] Image cropping operation: A sliding window cropping algorithm is used to crop the images to the same size, expanding the data set and ensuring that each image is focused on the battery weld.

[0030] Image denoising operation: Median filter and Gaussian filter algorithms are used to remove noise and interference in the image, for example, remove grain noise and high frequency interference.

[0031] 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 defect details in the image can be better captured.

[0032] Image standardization operation: By standardizing the size and orientation of images, we ensure the consistency of the data input to the deep learning model and eliminate brightness and color differences between different images.

[0033] In this embodiment, an industrial camera or other imaging equipment is used to capture images of the battery-welded area, obtaining welding images of the battery-welded object with sufficiently high resolution and clear details. The obtained welding images are then processed. For example, if the original welding images are color images, they are converted to grayscale images to simplify subsequent processing. Noise in the welding images is removed using methods such as a median filter or Gaussian filter. Brightness, contrast, and color balance are adjusted, and details in the welding images are enhanced using techniques such as histogram equalization. The welding images are cropped using a sliding window algorithm to focus on the battery-welded area and expand the data set. Furthermore, processing methods such as edge detection and contour extraction are used to prepare the welding image data for subsequent analysis or training of a learning model.

[0034] Specifically, as shown in FIG. 2 , in one embodiment of this embodiment, the acquisition processing module 100 includes a processing unit 101 configured to perform cropping, noise removal, enhancement and standardization processes on the welding image through a predetermined image processing algorithm to obtain the processed welding image.

[0035] In this embodiment, the acquired welding images are pre-processed to ensure that each welding image is focused on the battery welding area, remove noise and interference in the welding images, better capture defect details in the welding images, and eliminate brightness and color differences between different welding images, where the pre-processing includes, but is not limited to, image cropping, image noise removal, image enhancement, and image standardization.

[0036] Referring to FIG. 1, the battery welding defect detection system 10 further includes a defect statistics module 200 configured to perform classification statistics on the processed welding image based on a target detection model to obtain a plurality of battery candidate frames.

[0037] Specifically, in the fields of industrial automation and quality control, by constructing a target detection model, it can automatically learn to extract features from welding images and identify and classify various defects, which is an effective tool for detecting defects on product surfaces.

[0038] In this embodiment, a pre-constructed target detection model is used to locate and count the battery parts in the welding image. First, a large number of candidate frames are generated in the welding image to cover all areas including the battery welds. Then, a sliding window policy is used to generate a reasonable number and size of candidate frames based on the size of the welding image and the typical size of the battery, which can effectively detect defects in the battery welds and prevent abnormal judgment errors.

[0039] Specifically, as shown in FIG. 3 , in one embodiment of this embodiment, the defect statistics module 200 includes a candidate unit 201 configured to generate a plurality of candidate frames of corresponding number and size based on the processed welding image and a predetermined size according to a sliding window policy.

[0040] Specifically, the sliding window policy is a target detection technique that searches for potential target areas by moving windows of different sizes across a welding image. First, the size of the window to be used is determined based on a predetermined size. To detect targets of different scales, the window size may be a fixed size or a group of different sizes. Then, the first window is placed in one corner (usually the upper left corner) of the welding image to identify the initial window position. The window is then moved horizontally or vertically across the welding image at a predetermined stride to generate multiple candidate frames with corresponding numbers and sizes, which can be used to specifically analyze the defect detection situation using a subsequent model.

[0041] In this embodiment, targets of different scales can be detected based on the processed welding image and a predetermined size using a sliding window policy, thereby improving the detection comprehensiveness, and generating multiple candidate frames with corresponding numbers and sizes, thereby intelligently selecting candidate regions to reduce computational complexity and optimize computational efficiency.

[0042] The classification unit 202 is configured to classify each of the candidate frames according to the target detection model to obtain a plurality of types of candidate frames.

[0043] In this example, all candidate frames are input into a target detection model, and corresponding features are extracted for each candidate frame, which may include color histograms, gradient direction histograms, deep learning features, etc. All candidate frames are aggregated and sorted by feature, and non-maximal suppression (NMS) is used to select candidate frames that meet a predetermined threshold. For example, if the same target is detected in multiple adjacent candidate frame windows, the non-maximal suppression algorithm is used to remove duplicate detections, leaving only the candidate frame with the highest target probability, resulting in multiple candidate frames.

[0044] Referring to FIG. 1, the battery welding defect detection system 10 further includes a detection analysis module 300 configured to obtain a corresponding battery detection result according to the target detection model based on a predetermined threshold and the battery candidate window.

[0045] In this embodiment, the trained target detection model is used to identify whether there is a defect in the welding image, and the defect type and defect location information. Specifically, the trained target detection model is used to perform defect detection on the acquired welding image to obtain corresponding image features, and the image features are input into the model for analysis, and then, through forward propagation based on a predetermined function, a battery detection result including the defect type, defect location and probability distribution of each defect for detection is obtained, thereby simultaneously solving the problems of defect type detection and defect location estimation.

[0046] Specifically, as shown in FIG. 4, in one embodiment of this embodiment, the detection and analysis module 300 includes a detection unit 301 configured to perform detection on each of the candidate frames using the target detection model to obtain corresponding image features, and perform analysis on the image features to obtain corresponding defect types and defect locations.

[0047] Specifically, the image features are features of the welding image after extraction processing using a deep learning model, and include, but are not limited to, information such as edges, textures, and shapes.

[0048] In this embodiment, the trained target detection model utilizes the intermediate layer of the target detection model to extract features for the welding image region in each candidate frame, and the extracted image features are input into the fully connected layer, where feature pooling, normalization, nonlinear transformation and other processes are performed 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.), predict the exact location of the defects in each candidate frame, and refine the location by adjusting the coordinates of the bounding box to obtain the detected defect type and defect location, thereby reducing human intervention and improving the detection speed and accuracy.

[0049] The regression unit 302 is configured to perform a regression analysis based on the defect types and the defect locations with a predetermined function to obtain a corresponding defect probability distribution.

[0050] Specifically, the defect probability distribution is intended to represent the probability distribution of the number of defects that may occur in detected batteries, thereby enabling prediction and understanding of the defect rate in the production process, and improvement and optimization can be carried out.

[0051] In this embodiment, a regression analysis is performed based on the defect type and the defect location using a predetermined function to obtain the corresponding defect probability distribution. Preferably, the output layer of the deep learning model is connected to a softmax function, and the model output is converted into the probability distribution of each type of defect using a normalization model and output, making the model output more intuitive and easier to understand and interpret.

[0052] The analysis unit 303 is configured to analyze all candidate frames based on a predetermined threshold and the defect probability distribution to obtain the battery detection result.

[0053] Specifically, the predetermined threshold is used to determine the reliability of the target detection model result, for example, by assigning a confidence score to each candidate frame, which indicates the likelihood that the corresponding candidate frame belongs to a certain type according to the model.

[0054] In this embodiment, for each candidate frame's defect probability distribution, a probability score corresponding to a different type is calculated based on the defect probability distribution. The probability score of each candidate frame is compared with a predetermined threshold. If the score of the candidate frame is higher than the predetermined threshold, the candidate frame is recognized as a valid detection result; otherwise, the currently calculated candidate frame is rejected. For the valid candidate frame currently inspecting the battery welding image, the type with the highest score is determined as the detection state of the battery welding image. For example, if all valid candidate frames in the battery welding image have the highest score for the type "damaged sealed nail," the detection state of the corresponding battery welding image is damaged sealed nail, and the battery weld is also damaged. Therefore, a battery detection result is formed based on the predetermined threshold and the defect probability distribution. The battery detection result includes the detection state of the weld in the battery welding image, the defect type, the defect location, and the defect probability distribution.

[0055] 5, the battery welding defect detection system 10 according to the embodiment of the present application further includes an acquisition module 501, a construction module 502, and a training module 503. Each module will be described below.

[0056] The acquisition module 501 is configured to acquire a battery welding image training set, a battery welding image validation set, and a battery welding image test set for use in training.

[0057] 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-marked to form a battery welding image training set, a battery welding image validation set and a battery welding image test set, which are used to train, validate and test the model, respectively.

[0058] The construction module 502 is configured to construct an initial target detection model using the welding image training set based on a predetermined deep learning model.

[0059] In this example, a model architecture suitable for welding image classification and defect detection is first selected for constructing the target detection model. Preferably, the YOLOv8 (Ultralytics YOLOv8) model is used as the base model to construct the initial target detection model. The initial target detection model is configured to receive input welding images and has 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, particularly adding an adaptive enhancement module to identify reflection points and cracks, adding a mid-level target output layer, and optimizing the network's feature extraction and prediction modules to improve detection efficiency and accuracy. This includes removing reflection points, for example, by using adaptive filtering techniques to effectively remove high-light reflection areas in welding images. Detecting fine cracks requires high-resolution features and fine texture information. For example, a high-resolution network structure and attention mechanism can capture detail information and improve the accuracy of identifying fine objects, especially in detecting small objects and fine cracks. A mid-level target output layer is added, and the network's feature extraction and prediction modules are optimized. For example, to capture more detailed information, the high-resolution feature extraction module of HRNet is introduced. A mid-level target output layer is added. Adding one output layer to the network structure enhances the detection ability for mid-level targets. An attention mechanism is introduced. A squeeze-and-excitation (SE) block is added to the YOLOv8 feature map to strengthen the weight of features in specific regions. Hyperparameters such as the learning rate, batch size, and number of network layers are set, and after multiple experiments and validation, these parameters are optimized to improve the model's performance and generalization ability.

[0060] A training module 503 is configured to train the initial target detection model with the welding image validation set and the welding image test set to obtain the target detection model.

[0061] In this example, a model is trained using a welding image validation set and a welding image test set, and parameters are optimized using backpropagation and stochastic gradient descent (SGD) to minimize the loss function. The specific parameter optimization step involves conducting a large number of experiments on the dataset to compare the results and observe the merits and demerits of different image enhancement methods, learning rates, optimizers, etc., before moving to a larger dataset to obtain an optimized model. The specific loss functions are as follows: 1. Bounding Box Regression Loss: GIoU (Generalized IoU) is used to calculate the difference between the predicted bounding box and the true bounding box. 2. Classification Loss: Cross-Entropy Loss is used to calculate the difference between the predicted class and the true class. 3. Confidence Loss: Binary Cross-Entropy Loss is used to evaluate the difference between the reliability of the predicted object and the actual object. Then, data enhancement and cross-validation techniques are utilized to further improve the robustness and accuracy of the model.

[0062] In this embodiment, a plurality of battery welding image data sets are used to train the model to be constructed later, and model parameters are adjusted to optimize performance. The model's performance on unknown data is then evaluated to ensure its generalization ability. An initial target detection model is constructed based on the YOLOv8 model using the welding image training set, and the initial target detection model is trained using the welding image verification set and the welding image test set. The initial target detection model extracts features from the welding image verification set and the welding image test set that can contribute to distinguishing different defect types. Based on the extracted features, the initial target detection model predicts the presence or absence of defects in the welding images and the specific defect types. The reliability of the model detection results is confirmed using the welding image verification set and the welding image test set to obtain a target detection model for detecting defects in subsequent battery welding images. The target detection model is constructed based on a deep learning model, improving the target detection model's ability to detect reflected explosions and microcracks, thereby automatically and efficiently realizing battery welding defect detection work, improving detection speed and detection accuracy, and significantly improving production efficiency.

[0063] The battery welding defect detection system 10 according to the embodiment of the present application further comprises a result output module configured to display corresponding defect information according to the defect type, the defect location and the battery detection result.

[0064] 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 target, 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 numerical form, and the detection results are shown for direct observation. Image processing software or professional tools are used to superimpose defect information on the battery welding image, and the detected welding defect locations are marked on the welding image, displaying detailed defect type, defect number, and reliability information, which allows workers to quickly perform quality assessment of the battery based on the type and severity of the defect and adjust the manufacturing process and equipment parameters, contributing to predicting potential problems and optimizing production policies.

[0065] The battery welding defect detection system 10 according to the embodiment of the present application further includes a data update module configured to associate the battery detection result with cell data of the corresponding battery and update the data in a predetermined database.

[0066] In this embodiment, the predetermined database is preferably a local database or a cloud server database. The corresponding data is identified by the two-dimensional code on the battery surface, and the battery detection results are linked to specific cell data and stored in the predetermined database, which are then uploaded to the production MES (Manufacturing Execution System) to achieve data traceability. A highly efficient code reader is used to quickly and accurately identify the two-dimensional code on the battery surface to obtain the unique battery identifier. The detection results are linked to the battery identifier and stored in the local database for later retrieval and tracing. The detection data is uploaded to the MES system in real time to ensure that the latest detection results and tracing information can be obtained in a timely manner by the automated processing of later steps in the production line.

[0067] This embodiment discloses a battery welding defect detection system. The system includes: an acquisition processing module configured to acquire a welding image of a battery welding object and perform preprocessing on the welding image to obtain a processed welding image; a defect statistics module configured to perform classification statistics on the processed welding image based on a target detection model to obtain a plurality of battery candidate frames; and a detection analysis module configured to obtain corresponding battery detection results based on a predetermined threshold and the battery candidate frames using the target detection model. By performing defect detection on the processed battery welding image based on the target detection model, the present application can significantly improve detection efficiency and result accuracy and effectively reduce production costs.

[0068] Example 2 An embodiment of the present invention provides a battery welding defect detection method, which is used in the battery welding defect detection system according to the first embodiment.

[0069] 6, which is a schematic flow chart of a battery welding defect detection method, which includes the following steps:

[0070] Step S100: A welding image of a battery welding object is acquired, and preprocessing is performed on the welding image to obtain a processed welding image.

[0071] Step S200: Perform classification statistics on the processed welding image based on a target detection model to obtain a plurality of battery candidate frames.

[0072] Step S300: Obtain a corresponding battery detection result according to the target detection model based on a predetermined threshold and the battery candidate frame.

[0073] The battery welding defect detection method according to this embodiment can be used in the battery welding defect detection system according to embodiment 1, realize the functions of the battery welding defect detection system, and obtain corresponding effects, so to avoid duplication, the description will be omitted here.

[0074] As described above, the present invention further provides a battery welding defect detection method, which includes the steps of: acquiring a welding image of a battery welding object, performing preprocessing on the welding image to obtain a processed welding image; performing classification statistics on the processed welding image based on a target detection model to obtain a plurality of battery candidate frames; and obtaining corresponding battery detection results using the target detection model based on a predetermined threshold and the battery candidate frames. By performing defect detection on the processed battery welding image based on the target detection model, the present application can significantly improve detection efficiency and result accuracy and effectively reduce production costs.

[0075] Example 3 The present application also provides an electronic device, which includes a memory and a processor. A computer program is stored in the memory, and when the computer program is executed by the processor, a battery welding defect detection method according to a second embodiment is realized.

[0076] The electronic device according to this embodiment can implement the battery welding defect detection method according to the second embodiment, and therefore, to avoid duplication, the description will be omitted here.

[0077] Example 4 The present application further provides a computer-readable storage medium having a computer program stored therein, the computer program being executed by a processor to realize the battery welding defect detection method according to the second embodiment.

[0078] In this embodiment, the computer-readable storage medium may be a read-only memory (abbreviated as ROM), a random access memory (abbreviated as RAM), a magnetic disk, an optical disk, or the like.

[0079] The computer-readable storage medium according to this embodiment can implement the battery welding defect detection method according to the second embodiment, and therefore, to avoid duplication, the description thereof will be omitted here.

[0080] In the embodiments of the present application, the described devices and methods may be realized in other ways. The above device embodiments are merely illustrative. For example, the division of the units is merely a logical functional division, and may be a different division in actual implementation. For example, multiple units or components may be combined or integrated into another system, or some features may be omitted or not implemented. Furthermore, the shown or discussed mutual couplings or direct couplings or communication connections may be indirect couplings or communication connections via several communication interfaces, devices, or units, and may be electrical, mechanical, or other types of connections.

[0081] In addition, the units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units, i.e., they may be located in the same location or distributed across multiple network units. Some or all of the units can be selected according to actual requirements to achieve the purpose of the proposed embodiment.

[0082] Furthermore, each functional module in each embodiment of the present application may be integrated to form a single independent part, each module may exist independently, or two or more modules may be integrated to form a single independent part.

[0083] The functions may be realized in the form of software functional modules and stored in a computer-readable storage medium when sold or used as an independent product. From this understanding, the technical solution of the present application itself, or a portion of the technical solution that contributes to the prior art, may be realized in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of commands for causing a computer device (such as a personal computer, a server, or a network device) to execute all or part of the steps of the above-described methods in each embodiment of the present application. The storage medium includes various media capable of storing program code, such as a USB disk, a portable hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0084] As used herein, relational terms such as first and second, etc., are used only to distinguish one entity or action from another entity or action and do not necessarily require or imply any factual relationship or order between such entities or actions.

[0085] The above is only an example of the present application and does not limit the scope of protection of the present application. Those skilled in the art may have various modifications and variations to the present application. As long as they do not deviate from the spirit and principle of the present application, any modifications, equivalent substitutions, improvements, etc., will fall within the scope of protection of the present application. [Explanation of symbols]

[0086] 10 Battery welding defect detection system 100 Acquisition Processing Module 200 Defect Statistics Module 300 Detection Analysis Module 101 Processing Unit 201 candidate units 202 Classification Units 301 Detection Unit 302 Recurrence Unit 303 Analysis Unit 501 Acquisition Module 502 Construction Module 503 Training Module

Claims

1. an acquisition processing module configured to acquire a welding image of a battery welding object and perform pre-processing on the welding image to obtain a processed welding image; a defect statistics module configured to perform classification statistics on the processed welding image based on a target detection model to obtain a plurality of battery candidate frames; a detection analysis module configured to obtain a corresponding battery detection result based on a predetermined threshold and the battery candidate window using the target detection model. A battery welding defect detection system.

2. an acquisition module configured to acquire a battery welding image training set, a battery welding image validation set, and a battery welding image test set for use in training; a construction module configured to construct an initial target detection model using the welding image training set based on a predetermined deep learning model; a training module configured to train the initial target detection model with the welding image validation set and the welding image test set to obtain the target detection model. The battery welding defect detection system according to claim 1 .

3. The acquisition processing module includes a processing unit configured to perform cropping, noise removal, enhancement, and standardization processes on the welding image using a predetermined image processing algorithm to obtain the processed welding image.

3. The battery welding defect detection system according to claim 2.

4. The defect statistics module: a candidate unit configured to generate a plurality of candidate frames of corresponding number and size based on the processed welding image and a predetermined size according to a sliding window policy; a classification unit configured to classify each of the candidate frames according to the target detection model to obtain a plurality of types of candidate frames.

3. The battery welding defect detection system according to claim 2.

5. The detection and analysis module includes: a detection unit configured to perform detection on each of the candidate frames using the target detection model to obtain corresponding image features, and to perform analysis on the image features to obtain corresponding defect types and defect locations; a regression unit configured to perform a regression analysis based on the defect types and the defect locations with a predetermined function to obtain a corresponding defect probability distribution; an analysis unit configured to analyze all candidate frames based on a predetermined threshold and the defect probability distribution to obtain the battery detection result.

3. The battery welding defect detection system according to claim 2.

6. The battery detection device further includes a result output module configured to display corresponding defect information according to the defect type, the defect location, and the battery detection result.

6. The battery welding defect detection system according to claim 5.

7. The battery detection result is linked to cell data of the corresponding battery, and the data update module is configured to update the data in a predetermined database. The battery welding defect detection system according to claim 1 .

8. acquiring a welding image of a battery welding object, and performing preprocessing on the welding image to obtain a processed welding image; performing classification statistics on the processed welding image based on a target detection model to obtain a plurality of battery candidate frames; and obtaining a corresponding battery detection result based on the predetermined threshold and the battery candidate frame using the target detection model. A battery welding defect detection method.

9. The method includes a memory, a processor, and a computer program stored in the memory and executable by the processor, and when the computer program is executed by the processor, the method for detecting welding defects in a battery according to claim 8 is realized. An electronic device characterized by:

10. A computer program is stored in the battery welding defect detection method according to claim 8, and the computer program is executed by a processor to perform the battery welding defect detection method according to claim 8. A computer-readable storage medium comprising:

Images