Visual detection method and device for lithium battery pole piece

Abstract

The invention provides a lithium battery pole piece visual detection method and device, and the method comprises the steps: employing an industrial camera to collect a multi-angle image of a lithium battery pole piece, and employing a temperature and humidity sensor to monitor environment parameters in real time; the multi-angle image is preprocessed; performing feature extraction on the preprocessed multi-angle image by adopting a deep learning model to obtain multi-angle features; after defect identification is conducted on the multi-angle features through a trained neural network model, a defect identification result is obtained, and the trained neural network model adopts a self-adaptive threshold strategy considering environment parameters. According to the technical scheme provided by the invention, the visual detection precision of the lithium battery pole piece can be improved by considering environmental factors and adopting the adaptive threshold strategy.

Description

Technical Field

[0001] This document relates to the field of visual inspection technology, and in particular to a method and device for visual inspection of lithium battery pole pieces. Background Art

[0002] Lithium battery pole piece is one of the core components of lithium-ion battery. It is composed of positive electrode material, negative electrode material, current collector and adhesive. Adhesive is a material that bonds positive and negative electrode materials with current collector. It can improve the mechanical strength and stability of the electrode and help the transmission of lithium ions. The separator is located between the positive and negative electrode materials and is made of polymer material. Its function is to isolate the positive and negative electrodes, prevent short circuit, and allow lithium ions to pass through.

[0003] As the core part of the battery, the quality of the pole piece directly affects the overall performance of the battery. The traditional lithium battery pole piece detection method mainly relies on manual inspection, which is inefficient, costly, and easily affected by human factors, resulting in low detection accuracy. Manual inspection cannot meet the needs of large-scale production and cannot monitor quality changes in the production process in real time. Machine vision technology, as an automated detection method, has the advantages of non-contact, high precision, fast speed, and good repeatability. With the development of computer vision and image processing technology, machine vision is used more and more widely in the field of industrial inspection. The defects that may occur in the production process of lithium battery pole pieces include holes, cracks, impurities, dimensional deviations, etc. These defects may affect the performance and safety of the battery. Traditional manual detection methods are difficult to accurately identify these tiny defects, while machine vision technology can provide higher detection accuracy.

[0004] However, in the prior art, traditional visual inspection systems often rely on fixed thresholds and algorithms, which are easily affected by environmental changes, resulting in reduced detection accuracy; traditional visual inspection methods usually require manual intervention, are inefficient, and are slow when processing large amounts of data; complex system design and maintenance costs are high, and are prone to failure in harsh environments, requiring frequent maintenance and replacement of components. Therefore, a lithium battery pole piece visual inspection method and device are needed to solve the above problems. Summary of the invention

[0005] According to an embodiment of the present invention, a method for visually inspecting a lithium battery electrode is provided, comprising:

[0006] Industrial cameras are used to collect multi-angle images of lithium battery pole pieces, and temperature and humidity sensors are used to monitor environmental parameters in real time;

[0007] Preprocess multi-angle images;

[0008] A deep learning model is used to extract features from the preprocessed multi-angle images to obtain multi-angle features;

[0009] After using a trained neural network model to perform defect recognition on the multi-angle features, a defect recognition result is obtained, wherein the trained neural network model uses an adaptive threshold strategy that takes environmental parameters into consideration;

[0010] The defect identification result is sent to a monitoring center.

[0011] According to an embodiment of the present invention, a lithium battery pole piece visual inspection device is provided, comprising:

[0012] Multi-angle detection module, image preprocessing module, deep learning algorithm module, adaptive threshold setting module, result transmission module and monitoring center;

[0013] The multi-angle detection module is used to collect multi-angle images of lithium battery pole pieces using an industrial camera;

[0014] The image preprocessing module is used to preprocess the multi-angle image;

[0015] The adaptive threshold setting module sets the threshold according to the environmental parameters;

[0016] The deep learning algorithm module uses a deep learning model to extract features from the preprocessed multi-angle images to obtain multi-angle features; uses a trained neural network model to perform defect recognition on the multi-angle features to obtain defect recognition results, and the trained neural network model uses an adaptive threshold strategy that takes environmental parameters into consideration;

[0017] The result transmission module sends the defect identification results to the monitoring center.

[0018] By adopting the embodiments of the present invention, through adaptive image preprocessing, deep learning feature extraction and adaptive threshold setting, high-precision detection can be maintained in different environments, reducing the impact of environmental factors on the detection results. For example, through denoising and contrast enhancement, defects on the pole piece can be clearly identified even in insufficient light or the presence of noise.

[0019] The lightweight algorithm Mob ileNet-V2 is used to optimize the detection model, which significantly improves the detection speed while maintaining high accuracy. For example, by using Mob il eNet-V2, the detection time can be shortened from several minutes to a few seconds, greatly improving the efficiency of the production line.

[0020] When the ambient temperature rises, the material may expand, causing the defect size to increase, so the threshold needs to be increased to avoid false detection. Adjustment strategy: For every 1°C increase in temperature, the high threshold increases by 0.5% and the low threshold increases by 0.25%. Humidity impact: High humidity may cause water vapor to adhere to the surface of the material, affecting the image contrast, and the threshold may need to be lowered to enhance defect recognition. Adjustment strategy: For every 10% increase in humidity, the high threshold decreases by 1% and the low threshold decreases by 0.5%. Before each inspection, a new threshold is calculated based on the current temperature and humidity and applied to the defect detection process. The updated threshold should take effect immediately to ensure the real-time and accuracy of the detection process. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate one or more embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0022] Figure 1 It is a flow chart of a lithium battery pole piece visual inspection method according to an embodiment of the present invention;

[0023] Figure 2 It is a structural diagram of a lithium battery pole piece visual inspection device according to an embodiment of the present invention;

[0024] Figure 3 A schematic diagram of a lithium battery pole piece visual inspection device according to an embodiment of the present invention;

[0025] Description of the drawings: 1- conveyor belt, 2- detection box, 3- top industrial camera, 4- bottom industrial camera, 5- side industrial camera, 6- display screen. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the following will be combined with the drawings in one or more embodiments of this specification to clearly and completely describe the technical solutions in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this document.

[0027] Method Embodiment

[0028] According to an embodiment of the present invention, a method for visually inspecting a lithium battery electrode is provided. Figure 1Flow chart of the lithium battery pole piece visual inspection method according to an embodiment of the present invention. Figure 1 As shown, the lithium battery pole piece visual inspection method of the embodiment of the present invention specifically includes:

[0029] S1. Use industrial cameras to collect multi-angle images of lithium battery pole pieces, and use temperature and humidity sensors to monitor environmental parameters in real time;

[0030] Multi-angle inspection means observing the target object from different perspectives during the inspection process to obtain more comprehensive information. According to the characteristics of the target object, the appropriate inspection angle is selected, which usually includes the front view, side view and top view. For lithium battery electrodes, oblique views or special angles may also need to be considered to detect defects that are difficult to observe at conventional angles. Use multi-angle industrial cameras or camera arrays to ensure that each angle can cover the entire surface of the target object and that each camera can adjust the focal length and angle to meet different inspection needs.

[0031] S2, preprocessing the multi-angle images;

[0032] For each detection angle, image acquisition is performed according to preset parameters, including resolution, frame rate, etc., to ensure that image acquisition at each angle is performed synchronously to maintain image consistency. The image at each angle is preprocessed, including denoising, contrast enhancement, and white balance correction. The preprocessed image is input into the same feature extraction and defect recognition model. The detection results of different angles are fused to obtain more accurate defect location and size assessment. Weighted averaging or other fusion algorithms can be used to weight different angles according to their reliability and importance.

[0033] S3, using a deep learning model to extract features from the preprocessed multi-angle images to obtain multi-angle features;

[0034] S4, after using the trained neural network model to perform defect recognition on the multi-angle features, a defect recognition result is obtained, wherein the trained neural network model adopts an adaptive threshold strategy that takes environmental parameters into consideration;

[0035] A trained convolutional neural network (CNN) model is used to perform defect recognition on features. Defect recognition includes defect classification and positioning. When performing defect recognition, the embodiment of the present invention adjusts the detection threshold according to the real-time environmental parameters to ensure accurate detection in different environments. The pole piece is detected from different angles to increase the comprehensiveness and accuracy of the detection. A lightweight algorithm is used to optimize the detection process, increase the detection speed and reduce the consumption of computing resources. The detected defect information, including the defect type, location and size, is output, and the detection results are transmitted to the monitoring center for further analysis and processing.

[0036] The adaptive threshold strategy considering environmental parameters in the embodiment of the present invention specifically includes: when the ambient temperature rises, the material may expand, resulting in an increase in the size of the defect, so the threshold needs to be increased to avoid false detection. The setting of the threshold range includes: setting a reasonable threshold range, for example, the high threshold is set to the high percentile of the image gradient intensity (such as 90% or 95%), and the low threshold is set to a certain proportion of the high threshold (such as 1:2 to 1:3), and by real-time monitoring of the ambient temperature and humidity, it is ensured that the detection equipment can obtain environmental change information in a timely manner. The adjustment strategy of the embodiment of the present invention is: for every 1°C increase in temperature, the high threshold is increased by 0.5%, and the low threshold is increased by 0.25%. Humidity influence: high humidity may cause water vapor to adhere to the surface of the material, affecting the image contrast, and the threshold may need to be lowered to enhance defect recognition. Adjustment strategy: for every 10% increase in humidity, the high threshold is reduced by 1%, and the low threshold is reduced by 0.5%. Before each detection, a new threshold is calculated based on the current temperature and humidity, and applied to the defect detection process. The updated threshold should take effect immediately to ensure the real-time and accuracy of the detection process.

[0037] The trained neural network model outputs the detection results for each angle, as well as the final detection results after fusion. The results should include the type, location, and size of the defect.

[0038] An industrial camera is used to capture images of lithium battery pole pieces with an image resolution of 1920×1080 pixels. The captured images are preprocessed, including denoising, contrast enhancement, and color correction operations. A deep learning algorithm is used to extract features from the preprocessed images, including edge detection and region segmentation operations.

[0039] S5. Send the defect identification result to a monitoring center.

[0040] The following is a specific example of the lithium battery electrode visual inspection method provided by the embodiment of the present invention:

[0041] During the production process of lithium battery pole pieces, their quality is greatly affected by factors such as ambient temperature. To ensure the quality of pole pieces, the embodiment of the present invention simulates a lithium battery pole piece visual inspection process and adjusts the inspection parameters according to conditions such as ambient temperature.

[0042] Install temperature and humidity sensors in the detection area to monitor the ambient temperature and humidity in real time. Set the temperature monitoring range: 30℃-40℃, and the humidity monitoring range: 60%-80%.

[0043] Image acquisition: Place the lithium battery electrode to be tested on the testing platform and use an industrial camera to view it from the front view, side view and top view. Figure 3The pole piece is photographed at different angles to obtain multi-angle image data. The image acquisition equipment parameters are: resolution 1920x1080, frame rate 15fps, to adapt to the degradation of image quality in high temperature and high humidity environment.

[0044] Image preprocessing: De-noise the collected images to remove random noise in the images. Since high temperature and high humidity may reduce the image contrast, contrast enhancement is performed to improve the image details. The color deviation caused by light changes is corrected by the white balance algorithm, the dark channel first leveling method is used to remove the fogging phenomenon in the image, and the exposure compensation is increased to cope with the high temperature environment.

[0045] Feature extraction: The preprocessed image is input into the pre-trained ResNet-50 model to extract image features. The feature extraction device parameters are: CPU Intel Xeon E5-2680 v4, GPU NVIDIA Quadro P6000, to meet the computing requirements in high temperature environments. The extracted features are input into the trained CNN model to identify the defect types on the electrode, such as holes, cracks, impurities, etc.

[0046] Defect recognition model parameters: The CNN model uses the CIFAR-10 pre-trained model, adjusts the parameters of the convolution layer and the fully connected layer to adapt to the high temperature and high humidity environment, dynamically adjusts the detection threshold according to the real-time monitored ambient temperature and humidity, and sets the mapping relationship between temperature and threshold: for every 1°C increase in temperature, the threshold increases by 0.5%; for every 10% increase in humidity, the threshold increases by 1%. The adjusted detection threshold is 105% of the initial threshold + 1% * humidity percentage.

[0047] Multi-angle detection: Preprocess, extract features and identify defects for images at each angle to ensure comprehensive detection of the pole piece from different angles. The detected defect information (including type, location and size) is output to the monitoring center, which analyzes the defect information and generates a test report.

[0048] Detection efficiency: At 35°C and 70% relative humidity, the detection efficiency reaches 80 electrodes per minute, adapting to the reduced efficiency in high temperature and high humidity environments.

[0049] Detection accuracy: At 35°C and 70% relative humidity, the detection accuracy reaches 98%. The accuracy of detection is ensured by adjusting the threshold and preprocessing methods.

[0050] The entire testing process is automated with high real-time performance, meeting the needs of the production line. Although reduced in high temperature and high humidity environments, at 35°C and 70% relative humidity, the testing device operated stably without performance fluctuations, proving the reliability of the equipment in harsh environments.

[0051] Comparative analysis example: Comparison of visual inspection methods and devices for lithium battery pole pieces. The existing technology uses a traditional visual inspection system: it usually relies on fixed image processing algorithms and threshold settings, is sensitive to environmental changes, and has difficulty adapting to harsh conditions of high temperature and high humidity.

[0052] Fixed threshold detection systems: may perform well under normal conditions, but detection accuracy and stability will drop significantly when the environment changes, such as fluctuations in temperature and humidity.

[0053] Non-adaptive detection system: lacks real-time monitoring and adjustment of environmental parameters, so the detection effect is poor in extreme environments of high temperature and high humidity.

[0054] The present invention has adaptive image preprocessing: through denoising, contrast enhancement, color correction and defogging algorithms, the image quality is improved, and the image can be kept clear even in a high temperature and high humidity environment.

[0055] Comparative analysis, parameters of this embodiment (high temperature and high humidity environment) and the prior art (normal temperature and normal humidity environment).

[0056] Temperature monitoring range: 30℃-40℃; 20℃-30℃.

[0057] Humidity monitoring range: 60%-80%; 40%-60%.

[0058] Image acquisition frame rate, 15fps; 30fps.

[0059] Feature extraction CPU: Intel Xeon E5-2680 v4; Intel Core i7-8700K.

[0060] Feature extraction GPU, NVIDIAQuadro P6000 NVIDIA; GeForceRTX2080Ti.

[0061] Detection threshold adjustment, dynamic adjustment, considering temperature and humidity fixed thresholds, regardless of environmental changes.

[0062] Image acquisition, consider the impact of high temperature and humidity on image quality, reduce the frame rate, increase exposure compensation conventional image acquisition parameters, and do not consider environmental factors.

[0063] Image preprocessing, strengthens denoising, contrast enhancement, white balance correction and dark channel leveling methods to cope with environmental changes, conventional image preprocessing, does not consider environmental factors.

[0064] Feature extraction, using more powerful CPUs and GPUs to meet the computing needs in high temperature environments. Using conventional CPUs and GPUs without considering environmental factors.

[0065] Defect identification, dynamically adjust thresholds, and optimize based on temperature and humidity changes. Use fixed thresholds without considering environmental changes.

[0066] The result output outputs defect information, including type, location and size, and generates a report to output defect information, but does not consider environmental changes.

[0067] Detection efficiency: 80 pieces / minute (high temperature and high humidity); 100 pieces / minute (normal temperature and normal humidity).

[0068] Detection accuracy: 98% (high temperature and high humidity); 99.5% (normal temperature and normal humidity).

[0069] Real-time performance is affected by environmental factors, but it can basically meet the needs of the production line and is not affected by environmental factors.

[0070] Stability: stable operation in high temperature and high humidity environment with little performance fluctuation. Stable operation in normal temperature and humidity environment with little performance fluctuation.

[0071] Compared with the prior art, this embodiment makes the following improvements to the lithium battery pole piece visual inspection device under high temperature and high humidity environment:

[0072] Environmental adaptability: By monitoring temperature and humidity and dynamically adjusting detection parameters, the adaptability of the equipment in harsh environments is improved.

[0073] Image processing capability: The use of more powerful CPU and GPU, as well as enhanced image preprocessing methods, ensures image quality and detection accuracy.

[0074] Detection efficiency: Although the detection efficiency has declined, it can still meet the basic needs of the production line.

[0075] Stability: By optimizing parameters and equipment configuration, the stability of the equipment is ensured in high temperature and high humidity environments.

[0076] In summary, this embodiment has significant advantages over the prior art in the visual inspection of lithium battery pole pieces, can better adapt to the actual production environment, and improve inspection efficiency and product quality.

[0077] Working principle: The lithium battery pole pieces to be tested are placed on the conveyor belt 1 in sequence, and the conveyor belt 1 is used to transport the lithium battery pole pieces to the test box, and the lithium battery pole pieces are tested in the test box, and the image of the lithium battery pole pieces is captured by using an industrial camera. The camera converts the light reflected from the surface of the pole piece into an electrical signal through the optical system to form a digital image, removes random noise in the image through a median filter to improve the clarity of the image, and uses a histogram equalization algorithm to adjust the contrast of the image to make the dark and bright details in the image clearer. The white balance algorithm is used to correct the color deviation caused by light changes to ensure the true color of the image. In a high temperature and high humidity environment, the image may be fogged. The dark channel information in the image is extracted by the dark channel prior method (DCP), and the clarity of the image is restored based on this. The preprocessed image is feature extracted using a deep learning model (such as ResNet-50). The model can automatically learn complex features in images and provide strong support for subsequent defect identification. The extracted features are input into the trained CNN model, and the model outputs the probability distribution of each pixel belonging to different defect types. By analyzing the output results, the defect type on the electrode (such as holes, cracks, impurities, etc.) is determined, and the detection threshold is dynamically adjusted according to the real-time monitored environmental parameters (such as temperature and humidity). The mapping relationship between environmental parameters and preset thresholds ensures that the detection accuracy can be maintained in different environments. The electrode is inspected from different angles (such as front view, side view, and top view) to fully cover the surface of the electrode and reduce the possibility of missed detection. Lightweight algorithms such as MobileNet-V2 are used to optimize the detection model, reduce computational complexity, and improve detection speed. The detected defect information (including type, location, and size) is output for subsequent processing and analysis.

[0078] Specific workflow:

[0079] Image acquisition: Industrial cameras capture the pole piece images.

[0080] Image preprocessing: denoising, contrast enhancement, color correction and dehazing of images.

[0081] Feature extraction: Extract image features using deep learning models.

[0082] Defect identification: Identify defect types through CNN model.

[0083] Adaptive threshold adjustment: adjust the detection threshold according to environmental parameters.

[0084] Multi-angle inspection: Repeat image preprocessing, feature extraction and defect identification from different angles.

[0085] Result output: Output defect information.

[0086] Data transmission: Transmit defect information to the monitoring center.

[0087] By adopting the embodiments of the present invention, the following beneficial effects are achieved:

[0088] 1. Through adaptive image preprocessing, deep learning feature extraction and adaptive threshold setting, this embodiment can maintain high-precision detection in different environments and reduce the impact of environmental factors on the detection results. For example, through denoising and contrast enhancement, defects on the pole piece can be clearly identified even in insufficient light or in the presence of noise.

[0089] 2. The lightweight algorithm MobileNet-V2 is used to optimize the detection model, which significantly improves the detection speed while maintaining high accuracy. For example, by using MobileNet-V2, the detection time can be shortened from several minutes to a few seconds, greatly improving the efficiency of the production line.

[0090] 3. When the ambient temperature rises, the material may expand, causing the defect size to increase, so the threshold needs to be increased to avoid false detection. Adjustment strategy: For every 1°C increase in temperature, the high threshold increases by 0.5% and the low threshold increases by 0.25%. Humidity impact: High humidity may cause water vapor to adhere to the surface of the material, affecting the image contrast, and the threshold may need to be lowered to enhance defect recognition. Adjustment strategy: For every 10% increase in humidity, the high threshold decreases by 1% and the low threshold decreases by 0.5%. Before each inspection, a new threshold is calculated based on the current temperature and humidity and applied to the defect detection process. The updated threshold should take effect immediately to ensure the real-time and accuracy of the detection process.

[0091] Device Embodiment

[0092] According to an embodiment of the present invention, a lithium battery pole piece visual inspection device is provided. Figure 3 This is a flow chart of a lithium battery pole piece visual inspection device according to an embodiment of the present invention. Figure 3 As shown, the lithium battery pole piece visual inspection device of the embodiment of the present invention specifically includes:

[0093] Multi-angle detection module, image preprocessing module, deep learning algorithm module, adaptive threshold setting module, result transmission module and monitoring center; the multi-angle detection module is specifically used for:

[0094] Use multi-angle industrial cameras or camera arrays to ensure that every angle of the lithium battery pole piece is covered.

[0095] The multi-angle detection module is used to collect multi-angle images of lithium battery pole pieces using an industrial camera;

[0096] The image preprocessing module is used to preprocess the multi-angle image; the image preprocessing module is specifically used to:

[0097] De-noise, contrast enhancement, and white balance correction are performed on multi-angle images.

[0098] The adaptive threshold setting module sets the threshold according to the environmental parameters; the adaptive threshold setting module is specifically used for:

[0099] The high threshold is set to the first preset percentile of the image gradient intensity, and the low threshold is set to the preset ratio of the high threshold. The ambient temperature and humidity are monitored in real time. For every 1°C increase in temperature, the high threshold is increased by 0.5% and the low threshold is increased by 0.25%. For every 10% increase in humidity, the high threshold is reduced by 1% and the low threshold is reduced by 0.5%.

[0100] The deep learning algorithm module uses a deep learning model to extract features from the preprocessed multi-angle images to obtain multi-angle features; uses a trained neural network model to perform defect recognition on the multi-angle features to obtain defect recognition results, and the trained neural network model uses an adaptive threshold strategy that takes environmental parameters into consideration;

[0101] The result transmission module sends the defect identification results to the monitoring center.

[0102] The lithium battery pole piece visual inspection device further includes: a lightweight algorithm module, which is used to optimize the neural network model.

[0103] Figure 2 is a structural diagram of a lithium battery electrode visual inspection device according to an embodiment of the present invention. Figure 2 It can be seen that the working principle of the lithium battery pole piece visual inspection device of the embodiment of the present invention is:

[0104] The lithium battery pole pieces to be tested in the present invention are placed on a conveyor belt 1 in sequence, and the lithium battery pole pieces to be tested are transported to a test box 2 by the conveyor belt, and the lithium battery pole pieces are tested in the test box, which is implemented by the following steps:

[0105] The lithium battery pole pieces to be inspected are placed on the conveyor belt in sequence, and the conveyor belt is used to transport the lithium battery pole pieces to the inspection box, and the lithium battery pole pieces are inspected in the inspection box, and the images of the lithium battery pole pieces are captured by using the top industrial camera 3, the side industrial camera 5 and the bottom industrial camera 4. The camera converts the light reflected from the surface of the pole piece into an electrical signal through the optical system to form a digital image, and the random noise in the image is removed by the median filter to improve the clarity of the image. First, the image of the lithium battery pole piece is captured by the industrial camera to ensure that the image quality meets the requirements of subsequent processing. The collected image is pre-processed, including denoising, contrast enhancement and color correction operations to improve the image quality. The deep learning algorithm is applied to extract features from the pre-processed image, including edge detection and region segmentation operations, and the trained deep learning model is used to identify defects on the features, including defect classification and positioning. At the same time, the processed image, classification and other information can be directly displayed on the display screen 6 to improve the real-time interactive effect. The detection threshold is adjusted according to the real-time environmental parameters to ensure accurate detection in different environments. The pole piece is inspected from different angles to increase the comprehensiveness and accuracy of the detection. The lightweight algorithm is applied to optimize the detection process, increase the detection speed and reduce the consumption of computing resources. The detected defect information is output, including the defect type, location and size, and the detection results are transmitted to the monitoring center for further analysis and processing.

[0106] Through the above-mentioned implementation steps, the present invention has the following effects: under normal temperature and humidity environment, an initial threshold is set according to experience or experimental data for defect detection, and the threshold range is: a reasonable threshold range is set, for example, the high threshold is set to the high percentile of the image gradient intensity (such as 90% or 95%), and the low threshold is set to a certain proportion of the high threshold (such as 1:2 to 1:3), environmental parameter monitoring, real-time monitoring of ambient temperature and humidity, to ensure that the detection equipment can obtain environmental change information in a timely manner.

[0107] When the ambient temperature rises, the material may expand, causing the defect size to increase, so the threshold needs to be increased to avoid false detection. Adjustment strategy: For every 1°C increase in temperature, the high threshold increases by 0.5% and the low threshold increases by 0.25%. Humidity impact: High humidity may cause water vapor to adhere to the surface of the material, affecting the image contrast, and the threshold may need to be lowered to enhance defect recognition. Adjustment strategy: For every 10% increase in humidity, the high threshold decreases by 1% and the low threshold decreases by 0.5%. Before each inspection, a new threshold is calculated based on the current temperature and humidity and applied to the defect detection process. The updated threshold should take effect immediately to ensure the real-time and accuracy of the detection process.

[0108] By adopting the embodiments of the present invention, the following beneficial effects are achieved:

[0109] 1. Through adaptive image preprocessing, deep learning feature extraction and adaptive threshold setting, this embodiment can maintain high-precision detection in different environments and reduce the impact of environmental factors on the detection results. For example, through denoising and contrast enhancement, defects on the pole piece can be clearly identified even in insufficient light or in the presence of noise.

[0110] 2. The lightweight algorithm MobileNet-V2 is used to optimize the detection model, which significantly improves the detection speed while maintaining high accuracy. For example, by using MobileNet-V2, the detection time can be shortened from several minutes to a few seconds, greatly improving the efficiency of the production line.

[0111] 3. When the ambient temperature rises, the material may expand, causing the defect size to increase, so the threshold needs to be increased to avoid false detection. Adjustment strategy: For every 1°C increase in temperature, the high threshold increases by 0.5% and the low threshold increases by 0.25%. Humidity impact: High humidity may cause water vapor to adhere to the surface of the material, affecting the image contrast, and the threshold may need to be lowered to enhance defect recognition. Adjustment strategy: For every 10% increase in humidity, the high threshold decreases by 1% and the low threshold decreases by 0.5%. Before each inspection, a new threshold is calculated based on the current temperature and humidity and applied to the defect detection process. The updated threshold should take effect immediately to ensure the real-time and accuracy of the detection process.

[0112] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for visual inspection of lithium battery pole pieces, characterized in that include: Industrial cameras are used to collect multi-angle images of lithium battery pole pieces, and temperature and humidity sensors are used to monitor environmental parameters in real time; Preprocess multi-angle images; A deep learning model is used to extract features from the preprocessed multi-angle images to obtain multi-angle features; After using a trained neural network model to perform defect recognition on the multi-angle features, a defect recognition result is obtained, wherein the trained neural network model uses an adaptive threshold strategy that takes environmental parameters into consideration; The defect identification result is sent to a monitoring center.

2. The method according to claim 1, characterized in that The method of using an industrial camera to collect multi-angle images of lithium battery pole pieces specifically includes: Use multi-angle industrial cameras or camera arrays to ensure that every angle of the lithium battery pole piece is covered.

3. The method according to claim 1, characterized in that The preprocessing of the multi-angle images specifically includes: De-noise, contrast enhancement, and white balance correction are performed on multi-angle images.

4. The method according to claim 1, characterized in that The adaptive threshold strategy specifically includes: The high threshold is set to the first preset percentile of the image gradient intensity, and the low threshold is set to the preset ratio of the high threshold. The ambient temperature and humidity are monitored in real time, and the high and low thresholds are adjusted according to temperature and humidity changes.

5. The method according to claim 1, characterized in that The lithium battery pole piece visual inspection method further uses a lightweight algorithm to optimize the neural network model.

6. A lithium battery pole piece visual inspection device, characterized in that include: Multi-angle detection module, image preprocessing module, deep learning algorithm module, adaptive threshold setting module, result transmission module and monitoring center; The multi-angle detection module is used to collect multi-angle images of lithium battery pole pieces using an industrial camera; The image preprocessing module is used to preprocess the multi-angle image; The adaptive threshold setting module sets the threshold according to the environmental parameters; The deep learning algorithm module uses a deep learning model to extract features from the preprocessed multi-angle images to obtain multi-angle features; uses a trained neural network model to perform defect recognition on the multi-angle features to obtain defect recognition results, and the trained neural network model uses an adaptive threshold strategy that takes environmental parameters into consideration; The result transmission module sends the defect identification results to the monitoring center.

7. The device according to claim 6, characterized in that The multi-angle detection module is specifically used for: Use multi-angle industrial cameras or camera arrays to ensure that every angle of the lithium battery pole piece is covered.

8. The device according to claim 6, characterized in that The image preprocessing module is specifically used for: De-noise, contrast enhancement, and white balance correction are performed on multi-angle images.

9. The device according to claim 6, characterized in that The adaptive threshold setting module is specifically used for: The high threshold is set to the first preset percentile of the image gradient intensity, and the low threshold is set to the preset ratio of the high threshold. The ambient temperature and humidity are monitored in real time, and the high and low thresholds are adjusted according to temperature and humidity changes.

10. The device according to claim 6, characterized in that The lithium battery pole piece visual inspection device further includes: a lightweight algorithm module, which is used to optimize the neural network model.

Images