Lithium battery connecting piece quality detection system and method based on visual identification

Abstract

The invention relates to a lithium battery connecting piece quality detection system and method based on visual identification. The system comprises an optical imaging module which comprises a combined illumination unit and an image acquisition unit; the combined lighting unit is used for lighting the target detection station; the image acquisition unit is used for acquiring a multi-channel color image of a connecting piece identification area on the lithium battery to be detected; the material conveying module is used for conveying the lithium battery to be detected to a target detection station; the material conveying module comprises a diffuse reflection material tray; the calculation processing module comprises an image debugging unit, a channel processing unit, an analysis unit and a size conversion unit; the image debugging unit is used for pre-adjusting the contrast; the channel processing unit is used for receiving the multi-channel color image and carrying out channel separation and target channel extraction to obtain a single-channel image of a connection sheet identification area; the analysis unit is used for identifying connecting piece defects in the single-channel image; and the size conversion unit is used for carrying out defect image physical size conversion.

Description

Technical Field

[0001] This invention relates to the field of lithium battery manufacturing technology, and more specifically to a quality inspection system and method for lithium battery connectors based on visual recognition. Background Technology

[0002] As the core power source for modern electronic devices and new energy vehicles, the reliability of lithium batteries in their production and assembly directly determines the performance and safety of the final product. Among these processes, the welding of the connecting pieces and subsequent adhesive bonding are crucial steps in forming the internal and external electrical connections of the battery. Whether the welded connecting pieces undergo physical deformation will significantly affect the battery's electrochemical performance, long-term stability, and the yield rate of subsequent module assembly.

[0003] In existing technologies, machine vision technology is commonly used to inspect the quality of welded joints. Mainstream inspection methods typically rely on image processing algorithms, such as converting acquired color images into grayscale images, and then using existing defect recognition algorithms to identify and mark defects.

[0004] However, the aforementioned detection technologies still have several significant shortcomings in actual industrial applications. On the one hand, in machine vision technology, image quality is highly susceptible to environmental interference. For example, factors such as material tray reflections and complex backgrounds often result in low contrast in the acquired images, reducing the stability of subsequent image binarization and segmentation algorithms. On the other hand, the logic of existing algorithms is relatively simple and rigid, lacking sufficient sensitivity to identify defects with complex or irregular shapes, such as folded connecting pieces, local warping of connecting pieces, or tiny wrinkles, which can easily lead to missed detections. This increases the risk of misjudgment and product scrapping. Summary of the Invention

[0005] The purpose of this invention is to provide a quality inspection system and method for lithium battery connectors based on visual recognition. This system can perform automated inspection of the quality of connectors after lithium battery welding with high stability and high precision. It has the advantage of high imaging quality, can effectively detect complex defects such as connector folding and connector deformation, is not easily affected by environmental interference, and has high stability and accuracy of inspection results under complex working conditions.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows.

[0007] In a first aspect, the present invention provides a quality inspection system for lithium battery connectors based on visual recognition, the system comprising:

[0008] An optical imaging module includes a combined illumination unit and an image acquisition unit; the combined illumination unit is used to illuminate the target detection station; the image acquisition unit is used to acquire multi-channel color images of the connector identification area on the lithium battery to be inspected.

[0009] A material conveying module is used to convey the lithium battery to be tested to the target testing station; the material conveying module includes a diffuse reflective material tray;

[0010] The computational processing module is connected to the optical imaging module and includes an image adjustment unit, a channel processing unit, an analysis unit, and a size conversion unit. The image adjustment unit is used for contrast pre-adjustment. The channel processing unit is used to receive the multi-channel color image and perform channel separation and target channel extraction to obtain a single-channel image of the connector identification area. The analysis unit is used to identify connector defects in the single-channel image. The size conversion unit is used to perform physical size conversion.

[0011] As a preferred embodiment of the present invention, the L* value of the diffuse reflective material tray in the CIELAB color space is less than or equal to 25.

[0012] As a preferred embodiment of the present invention, the combined lighting unit includes a strip light and a spherical integrating light disposed above the identification area of ​​the connecting piece.

[0013] Secondly, the present invention provides a method for quality inspection of lithium battery connectors based on visual recognition, the method comprising:

[0014] S01, use a diffuse reflective material tray to transfer the lithium battery to be tested to the target testing station;

[0015] S02, Illuminate the target detection station and acquire a multi-channel color image of the connector identification area on the lithium battery to be tested;

[0016] S03, perform contrast pre-adjustment on the multi-channel color image, perform channel separation and target channel extraction, and obtain a single-channel image of the connecting piece recognition area;

[0017] S04, identify the connector defects in the single-channel image and perform physical size conversion to obtain the quality inspection result of the connector defects of the lithium battery to be tested.

[0018] As a preferred embodiment of the present invention, in step S04, the step of identifying the connection defects in the single-channel image specifically involves: extracting all pixels in the single-channel image that fall into the feature grayscale window and performing pixel area calculation and image output to obtain the feature image area C.

[0019] As a preferred embodiment of the present invention, the grayscale range of the feature grayscale window is [40, 60] and [120, 255].

[0020] As a preferred embodiment of the present invention, step S04, the step of performing physical size conversion, specifically includes:

[0021] Calculate the actual physical feature area S:

[0022] S=d·z·C+e

[0023] Where d is the pixel size conversion coefficient of the circular standard part, z is the pixel size conversion coefficient of the rectangular standard part, and e is the preset compensation coefficient.

[0024] As a preferred embodiment of the present invention, step S04, the step of obtaining the defect quality detection result of the connector piece of the lithium battery to be tested, specifically involves calculating the area deviation value K and comparing the area deviation value K with the defect warning value J. If K ≥ J, then the detection result of the connector piece being unqualified is output; if K < J, then the detection result of the connector piece being qualified is output, wherein:

[0025] K = S / A × 100%

[0026] In the formula, A is the physical area of ​​the connecting piece.

[0027] As a preferred embodiment of the present invention, the defect warning value J is 3.5%.

[0028] In summary, the present invention has the following beneficial effects:

[0029] This invention solves the common problems of tray reflection and background stray light interference in existing production lines from an optical perspective by using a diffuse reflection material tray with specific brightness characteristics on the hardware side, combined with an asymmetrical combination lighting system consisting of white strip lights and white spherical integral lights.

[0030] In this invention, the contrast pre-adjustment unit and channel separation logic in the calculation and processing module are deeply coupled, which enables the system to improve the grayscale contrast between the connecting piece and the background. This solves the problem of binarization segmentation failure caused by excessively high background grayscale and low image contrast in the prior art, and improves the detection robustness and imaging stability of the system under complex production conditions.

[0031] This invention can not only detect defects, but also accurately calculate and warn of the actual physical area of ​​defects, effectively making up for the shortcomings of existing technologies in terms of sensitivity to complex and irregular defects, and reducing the risk of missed detection in the lithium battery production process. Attached Figure Description

[0032] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0033] Figure 1 This is a system structure block diagram of the present invention;

[0034] Figure 2 This is a flowchart of the method of the present invention;

[0035] Figure 3 This is a schematic diagram showing the arrangement of the optical imaging module in an embodiment of the present invention. Detailed Implementation

[0036] The subject matter described herein will now be discussed with reference to exemplary embodiments. It should be understood that these embodiments are discussed merely to enable those skilled in the art to better understand and implement the subject matter described herein, and are not intended to limit the scope, applicability, or examples set forth in the claims. The function and arrangement of the elements discussed may be changed without departing from the scope of this specification. Various processes or components may be omitted, substituted, or added as needed in the various examples. For example, the described methods may be performed in a different order than described, and steps may be added, omitted, or combined. Furthermore, features described in some examples may be combined in other examples.

[0037] Example 1

[0038] like Figure 1 As shown, this embodiment provides a quality inspection system for lithium battery connectors based on visual recognition. The system includes:

[0039] The optical imaging module includes a combined illumination unit and an image acquisition unit;

[0040] The combined lighting unit includes multiple strip lights and spherical integrating lights positioned above the connector identification area. These are used to illuminate the target detection station. The image acquisition unit includes a full-color optical camera for capturing images of the designated target area.

[0041] like Figure 3As shown, a white spherical integrating lamp is installed directly above the identification area of ​​the connector piece, providing soft, diffused light. Since the connector piece inherently possesses dotted areas to increase weld strength, the illumination from the white spherical integrating lamp significantly reduces specular highlights caused by these minor irregularities, minimizing image recognition interference. Simultaneously, it makes the shadow contours of large-scale deformations such as warping and folding of the connector piece clearer. Multiple white strip lights are arranged around the white spherical integrating lamp for directional illumination, utilizing the height difference between the defect and the substrate to create clear light and dark boundaries, ensuring image contrast of the defect contours and providing a basis for subsequent positioning.

[0042] The material conveying module is used to transport the lithium batteries to be tested to the target testing station. The module includes a diffuse reflective material tray. The surface of the diffuse reflective material tray is treated with a dark color in the CIELAB color space (L*≤25) and possesses diffuse reflective properties. The dark color maximizes the absorption of ambient stray light and provides a clear contrast with the white light source. The diffuse reflective material tray, in conjunction with the spherical integrating lamp, creates a stable, uniform, and low-noise imaging background environment, which helps ensure the consistency of the testing system's results.

[0043] The computational processing module is connected to the optical imaging module and includes an image debugging unit, a channel processing unit, an analysis unit, and a size conversion unit. The image debugging unit is used for contrast pre-adjustment. The channel processing unit is used to receive multi-channel color images and perform channel separation and target channel extraction to obtain a single-channel image of the connector identification area. The analysis unit is used to identify connector defects in the single-channel image. The size conversion unit is used to perform physical size conversion.

[0044] The computational processing module can be implemented based on a microprocessor (MCU). It is mainly used to receive multi-channel color images and extract defect images from them. It mainly includes four functional execution units: image debugging unit, channel processing unit, analysis unit, and size conversion unit.

[0045] The task of the image debugging unit is to output image features most conducive to defect identification through the synergy of hardware parameter adjustment and software algorithm compensation. The image debugging unit first controls the combined illumination unit and camera to perform the step of creating a low-exposure state. Specifically, it synchronously reduces the illumination intensity of the spherical integrating light source and shortens the camera's (camera's) exposure time, so that the overall grayscale value of the connecting piece's main surface in the acquired multi-channel color image is controlled within the lower range of 25 to 40. This low-exposure condition can maximally suppress random highlight noise caused by the microscopic roughness of the connecting piece's surface, while ensuring a basic contrast (not less than 20 grayscale) between the connecting piece and the background.

[0046] Next, the low-exposure original image is compensated to improve the feature contrast to a usable range. Specifically, based on the black level parameter adjustment method in the existing technology, the overall image grayscale is reduced without changing the relative relationship between pixels. This is mainly used to finely adjust the background to a darker shade (target 20-30 grayscale) to highlight the main body of the connecting piece. Then, based on gamma correction technology, the overall image brightness is linearly increased, raising the grayscale of the main body of the connecting piece from the low exposure range (25-40) and uniformly increasing the grayscale of the normal area of ​​the connecting piece to the target range (80-100). The grayscale of the defective area is separated into two distinct ranges: 40-60 (shadow) and 120-255 (highlight).

[0047] The channel processing unit prioritizes extracting the optimal processing channel for each connecting piece of different materials. For example, the contrast between the copper foil and the background is highest in the red channel, so the channel processing unit will prioritize extracting the image of that channel. After selecting the target channel, the channel processing unit performs channel separation and target channel extraction to obtain a single-channel image of the connecting piece recognition area.

[0048] The analysis unit deploys a pre-trained multi-class segmentation and recognition model, which is iteratively trained using pre-prepared deformed samples of connected pieces. It can analyze various defects in connected pieces, extract the pixels at the defect locations, and output a feature image. After inputting the single-channel image of the connected piece recognition area into the multi-class segmentation and recognition model, the model outputs defect location prediction information. The analysis unit binarizes the defect location mask output by the model, uses blob analysis to count the total number of pixels occupied by the defects, and finally outputs the area C of the feature image.

[0049] The size conversion unit is used for physical size conversion. During actual production line installation, it is necessary to photograph and measure both circular and rectangular standard parts (with known precise physical dimensions). By calculating the correspondence between their pixel areas and actual areas, the size conversion coefficients d and z are determined. Based on the conversion formula, the physical size of the image is then converted.

[0050] S=d·z·C+e

[0051] Where d is the pixel size conversion coefficient of the circular standard part, z is the pixel size conversion coefficient of the rectangular standard part, and e is the preset compensation coefficient.

[0052] It should be noted that a compensation coefficient e is introduced during the conversion process to compensate for systematic hardware errors such as camera lens distortion and minor uneven lighting. This coefficient can be obtained through calibration on a standard testing platform.

[0053] By converting physical dimensions, it is possible to accurately calculate and provide early warnings of the true physical area of ​​defects.

[0054] Example 2

[0055] like Figure 2 As shown, this embodiment provides a method for quality inspection of lithium battery connectors based on visual recognition. The method includes:

[0056] S01, use a diffuse reflective material tray to transfer the lithium battery to be tested to the target testing station;

[0057] S02, Illuminate the target detection station and acquire a multi-channel color image of the connector identification area on the lithium battery to be inspected;

[0058] S03, perform contrast pre-adjustment on the multi-channel color image, perform channel separation and target channel extraction to obtain a single-channel image of the connecting piece recognition area;

[0059] S04. Identify the connector defects in the single-channel image and perform physical size conversion to obtain the quality inspection results of the connector defects of the lithium battery to be tested.

[0060] The specific steps for obtaining the defect quality inspection results of the connector of the lithium battery under test are as follows: calculate the area deviation value K and compare the area deviation value K with the defect warning value J. If K ≥ J, output the inspection result that the connector is unqualified; if K < J, output the inspection result that the connector is qualified.

[0061] K = S / A × 100%

[0062] In the formula, A represents the physical area of ​​the connecting piece. The preferred value for the defect warning value J is 3.5%.

[0063] In summary, this invention enables highly stable and precise automated inspection of the quality of welded connectors in lithium batteries; it boasts high imaging quality, effectively detecting complex defects such as connector folding and deformation, is not easily affected by environmental interference, and exhibits high stability and accuracy of inspection results under complex working conditions.

[0064] Several embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, and are not limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical applications, or technological improvements to the embodiments in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A quality inspection system for lithium battery connectors based on visual recognition, characterized in that, The system includes: An optical imaging module, comprising a combined illumination unit and an image acquisition unit; The combined lighting unit is used to illuminate the target inspection station; the image acquisition unit is used to acquire multi-channel color images of the connector identification area on the lithium battery to be inspected; A material conveying module is used to convey the lithium battery to be tested to the target testing station; the material conveying module includes a diffuse reflective material tray; The computational processing module is connected to the optical imaging module and includes an image adjustment unit, a channel processing unit, an analysis unit, and a size conversion unit. The image adjustment unit is used for contrast pre-adjustment. The channel processing unit is used to receive the multi-channel color image and perform channel separation and target channel extraction to obtain a single-channel image of the connector identification area. The analysis unit is used to identify connector defects in the single-channel image. The size conversion unit is used to perform physical size conversion of the defect image.

2. The lithium battery connector quality inspection system based on visual recognition according to claim 1, characterized in that, The L* value of the diffuse material tray in the CIELAB color space is less than or equal to 25.

3. The lithium battery connector quality inspection system based on visual recognition according to claim 2, characterized in that, The combined lighting unit includes a strip light and a spherical integrating light disposed above the identification area of ​​the connecting piece.

4. A method for quality inspection of lithium battery connectors based on visual recognition, characterized in that, The methods include: S01, use a diffuse reflective material tray to transfer the lithium battery to be tested to the target testing station; S02, Illuminate the target detection station and acquire a multi-channel color image of the connector identification area on the lithium battery to be tested; S03, perform contrast preprocessing on the multi-channel color image, perform channel separation, target channel extraction and fixed color extraction to obtain a single-channel grayscale image of the connecting piece recognition area; S04, identify the connector defects in the single-channel image and perform physical size conversion to obtain the quality inspection result of the connector defects of the lithium battery to be tested.

5. The method for quality inspection of lithium battery connectors based on visual recognition according to claim 4, characterized in that, In step S04, the step of identifying the connection defects in the single-channel image specifically involves: extracting all pixels in the grayscale processed image that fall into the feature grayscale threshold window, calculating the pixel area, and outputting the image to obtain the feature image area C.

6. The method for quality inspection of lithium battery connectors based on visual recognition according to claim 5, characterized in that, The grayscale range of the feature grayscale window is [40, 60] and [120, 255].

7. The method for quality inspection of lithium battery connectors based on visual recognition according to claim 4, characterized in that, In step S04, the specific steps for physical size conversion are as follows: Calculate the actual physical feature area S: S = d·z·C + e Where d is the pixel size conversion coefficient of the circular standard part, z is the pixel size conversion coefficient of the rectangular standard part, and e is the preset compensation coefficient.

8. The method for quality inspection of lithium battery connectors based on visual recognition according to claim 7, characterized in that, In step S04, the specific steps for obtaining the defect quality detection results of the connecting piece of the lithium battery to be tested are as follows: calculate the area deviation value K and compare the area deviation value K with the defect warning value J. If K≥J, then output the detection result that the connecting piece is unqualified. If K < J, then output the test result indicating that the connector is qualified, where: K = S / A × 100% In the formula, A is the physical area of ​​the connecting piece.

9. A method for quality inspection of lithium battery connectors based on visual recognition according to claim 8, characterized in that, The defect warning value J is 3.5%.

Images