Quality detection method and system for soft package lithium battery

Abstract

The invention belongs to the technical field of image processing, and particularly relates to a quality detection method and system for a soft package lithium battery, and the method comprises the steps: obtaining an edge sealing image of the soft package lithium battery, and carrying out the convolution processing in a plurality of preset directions through a preset filter, so as to obtain the energy amplitudes of all pixel points in different preset directions; determining an anisotropic consistency index of each pixel point according to the energy amplitude of each pixel point in different preset directions; calculating the phase distribution confusion degree and the gray scale gradient of each pixel point in the local area, and evaluating the infiltration and diffusion characteristics of each pixel point to determine the leakage and diffusion disturbance flux of each pixel point; and carrying out nonlinear fusion on the anisotropic consistency index and the leakage diffusion disturbance flux, and evaluating the edge sealing quality state of the soft package lithium battery. According to the invention, accurate identification and reliable evaluation of the edge sealing quality of the soft package lithium battery can be realized.

Description

Technical Field

[0001] This invention relates to the field of image processing technology. More specifically, this invention relates to a method and system for quality inspection of pouch lithium batteries. Background Technology

[0002] As a core energy storage component in new energy vehicles and electronic devices, the safety of pouch lithium batteries has received significant attention from the industry. Pouch lithium batteries are typically encapsulated using an aluminum-plastic composite film, and the tightness of the sealing process directly determines whether electrolyte leakage will occur. In actual production, the heat-sealing head, under high temperature and pressure, causes regular plastic deformation on the surface of the aluminum-plastic composite film, forming a series of tiny stripes distributed parallel to the sealing direction. These tiny stripes are generally considered normal mechanical wrinkles.

[0003] However, in automated visual inspection systems, if the sealing is not tight and a small amount of electrolyte leaks in, the electrolyte will quickly spread along the microstructure of the aluminum-plastic composite film, causing the originally clear and highly directional interference texture to become blurred and scattered. Since traditional detection methods often rely on edge extraction technology based on grayscale gradients, this edge extraction technology is prone to misjudging bright areas as leakage signals when faced with strong reflective interference on the surface of the aluminum-plastic composite film.

[0004] Furthermore, since the grayscale changes of mechanical wrinkles are quite drastic, when the leakage signal of the electrolyte is weak, the leakage signal is often masked by the normal texture background. At this time, it is difficult to accurately peel off the leakage area by simply identifying the brightness distribution or simple geometric features, which leads to a high false alarm rate and false negative rate of the detection system. Summary of the Invention

[0005] To address the technical problem of effectively distinguishing between mechanical wrinkles and electrolyte leakage in the sealing area of ​​pouch lithium batteries, this invention proposes a quality inspection method and system for pouch lithium batteries, which can achieve accurate identification and reliable evaluation of the sealing quality of pouch lithium batteries.

[0006] In a first aspect, the present invention provides a quality inspection method for a pouch lithium battery, comprising: acquiring an edge-sealing image of the pouch lithium battery; performing convolution processing in multiple preset directions using a preset filter to obtain the energy amplitude of each pixel in different preset directions; evaluating the directional feature distribution of each pixel based on the energy amplitude of each pixel in different preset directions to determine the anisotropy consistency index of each pixel; calculating the phase distribution disorder and gray-level gradient of each pixel in a local region, and evaluating the wetting diffusion characteristics of each pixel in combination with the average gray-level gradient of the edge-sealing image to determine the leakage diffusion disturbance flux of each pixel; nonlinearly fusing the anisotropy consistency index and the leakage diffusion disturbance flux, and evaluating the edge-sealing quality status of the pouch lithium battery in combination with the physical distance of the pixel relative to the centerline of the edge-sealing.

[0007] By employing the above technical solution, an edge-sealing image of a pouch lithium battery is obtained, and the energy amplitude is extracted using a multi-directional filter, thereby determining the anisotropy consistency index and leakage diffusion disturbance flux. This method effectively solves the problems in existing technologies, such as the difficulty in distinguishing between mechanical wrinkles and electrolyte leakage, and the susceptibility to interference from the strong reflectivity of the aluminum-plastic composite film, thus achieving accurate monitoring and evaluation of the edge-sealing quality of pouch lithium batteries.

[0008] Preferably, the anisotropy consistency index satisfies the following relationship: ; In the formula, Indicates the anisotropy consistency index; Indicates the pixel at the th Energy amplitude in a preset direction; Indicates the first The preset angle of each filter; This represents the first constant term that is preset.

[0009] By employing the above technical solution, an anisotropy consistency index is determined, which can accurately characterize the concentration of energy in a specific direction. When leakage occurs, causing light scattering, this index decreases significantly, thus completing the initial screening of normal textures and anomalous disturbances.

[0010] Preferably, the leakage diffusion disturbance flux satisfies the following relationship: ; In the formula, This represents the leakage diffusion disturbance flux; The standard deviation of pixel phase information within a local window; This represents the local grayscale gradient value of a pixel; This represents the average grayscale gradient of the edge-sealed image; This represents the pre-defined second constant term; This represents the pre-defined third constant term.

[0011] By adopting the above technical solution, the leakage diffusion disturbance flux reflecting the degree of phase distribution disorder in the local area is calculated. Combined with gradient suppression logic, a high-score response can be generated when the texture is blurred and the phase is disordered. This effectively eliminates the interference of image defocus or normal edges on leakage judgment and enhances detection specificity.

[0012] Preferably, the edge sealing quality is determined by calculating a quality assessment index, which satisfies the following relationship: ; In the formula, Indicates a quality assessment index; Indicates the anisotropy consistency index; This represents the leakage diffusion disturbance flux; This represents the normalized distance parameter from a pixel to the preset edge centerline; This represents the pre-defined fourth constant term.

[0013] By adopting the above technical solution, calculating the quality assessment index, and introducing the physical distance from the pixel to the center line of the sealing edge, and using a quadratic nonlinear coupling mechanism, extremely high response is generated in areas with poor directionality and high disorder, thereby achieving accurate positioning and decisive judgment of the leakage location in the core sealing edge area.

[0014] Preferably, the convolution process using a preset filter in multiple preset directions is specifically: using a Log-Gabor filter to perform convolution in 8 preset directions.

[0015] By adopting the above technical solution, the Log-Gabor filter is used to perform convolution processing in 8 preset directions, which effectively captures multi-scale texture information in the edge sealing image. In particular, it can generate representative response vectors for mechanical wrinkles with periodic characteristics, providing robust data support for revealing the physical direction of surface texture.

[0016] Preferably, determining the anisotropy consistency index for each pixel includes: calculating the sum of the cosine and sine components of the energy amplitude in each preset direction; performing vector synthesis on the sum of the cosine and sine components to calculate the magnitude, thereby characterizing the concentration of energy in a specific direction; calculating the cumulative sum of the energy amplitudes in each preset direction, and adding the first constant term as a smoothing factor to the cumulative sum; and determining the anisotropy consistency index based on the ratio between the magnitude and the cumulative sum after adding the smoothing factor.

[0017] By adopting the above technical solution, the sum of cosine components and the sum of sine components are calculated and vector synthesis is performed to accurately characterize the degree of energy concentration in a specific direction. Furthermore, a smoothing factor is introduced to optimize the calculation accuracy of the anisotropy consistency index, thereby improving the ability to characterize mechanical wrinkles.

[0018] Preferably, determining the leakage diffusion disturbance flux of each pixel includes: using a natural logarithm function to perform nonlinear processing on the standard deviation of the pixel phase information within the local window to amplify the variation characteristics of the phase distribution disorder; using an exponential function to construct an inverse threshold based on the local gray-level gradient value and the average gray-level gradient; using the inverse threshold to weight the nonlinearly processed standard deviation, and suppressing the value of the leakage diffusion disturbance flux when the local gray-level gradient value is greater than the average gray-level gradient.

[0019] By adopting the above technical solution, the chaotic phase distribution features are amplified by utilizing the natural logarithm function, and a reverse threshold mechanism based on gray-level gradient is constructed. This can sensitively detect the microscopic phase drift caused by liquid wetting, and effectively suppress interference values ​​when the local gray-level gradient is large, thus significantly improving the accuracy of leakage identification.

[0020] Preferably, the evaluation of the sealing quality status of the soft-pack lithium battery includes: using the square of the leakage diffusion disturbance flux as the dominant judgment factor, and using the anisotropy consistency index as a suppression factor to adjust the dominant judgment factor to improve the evaluation value in areas with missing directionality and high disorder; calculating spatial weights based on the physical distance from the pixel to the preset sealing centerline, and using the spatial weights to weight the adjusted dominant judgment factor to obtain the quality evaluation index.

[0021] By adopting the above technical solution, the square value of the leakage diffusion disturbance flux is used as the dominant judgment factor, and the anisotropy consistency index is combined for adjustment. The evaluation value is significantly improved in the directionally missing area. At the same time, spatial weighting is used to ensure that the detection results conform to the natural physical process of liquid wetting.

[0022] Preferably, a quality inspection method for a soft-pack lithium battery further includes: setting a defect judgment threshold; comparing the quality assessment index with the defect judgment threshold; if the quality assessment index is greater than the defect judgment threshold, determining that there is electrolyte leakage in the area corresponding to the pixel, and generating defect location coordinates.

[0023] By adopting the above technical solution and comparing the quality assessment index with the preset defect judgment threshold, the electrolyte leakage area is automatically determined and the defect location coordinates are generated, thereby realizing automated real-time monitoring and reliable recording of the packaging quality of soft-pack lithium batteries.

[0024] In a second aspect, the present invention provides a quality inspection system for pouch lithium batteries, including a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the aforementioned quality inspection method for pouch lithium batteries is implemented.

[0025] By adopting the above technical solution, a computer program for the quality inspection method of a soft-pack lithium battery is generated and stored in a memory so that it can be loaded and executed by a processor. This allows for the creation of a terminal device based on the memory and processor, making it convenient to use.

[0026] This invention utilizes the anisotropic destructive effect of electrolyte wetting on the microstructure of aluminum-plastic composite film. It observes that normal mechanical wrinkles exhibit significant anisotropy, while the diffusion of leaked liquid presents an isotropic energy distribution. Furthermore, it achieves deep stripping of the background texture by using an anisotropic consistency index.

[0027] Furthermore, by introducing leakage diffusion perturbation flux, this invention, based on phase domain analysis and combined with dynamic threshold suppression of grayscale gradients, can keenly capture microscopic phase drift caused by liquid wetting. In addition, through multi-dimensional feature fusion, the problem of feature obscuration under strong reflective backgrounds is effectively solved, thereby improving the stability of the detection system. Attached Figure Description

[0028] Figure 1 This is a flowchart of a quality testing method for a soft-pack lithium battery according to an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the multidimensional feature space separation effect in an embodiment of the present invention; Figure 3 This is a schematic diagram of the thermal distribution for quality defect identification in an embodiment of the present invention. Detailed Implementation

[0029] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0030] This invention discloses a quality testing method for soft-pack lithium batteries, referring to... Figure 1 This includes steps S1-S4: S1. Obtain the edge-sealing image of the soft-pack lithium battery, and perform convolution processing in multiple preset directions using a preset filter to obtain the energy amplitude of each pixel in different preset directions.

[0031] In an optional embodiment, the sealing edge of the soft-pack lithium battery is first captured in real time by an industrial camera deployed on the production line of the soft-pack lithium battery. Since the aluminum-plastic film material has strong specular reflection characteristics, small disturbances in ambient light can cause drastic changes in the absolute grayscale value of the pixel. Therefore, in this embodiment of the invention, a Log-Gabor filter bank is used to perform frequency domain analysis on the image.

[0032] Specifically, the method for frequency domain analysis of the image using a Log-Gabor filter bank is as follows: First, eight preset directions are defined, with an angle of 22.5 degrees between each pair of adjacent preset directions, thus covering a 180-degree texture feature space. Then, through convolution operations between the filter and the image, the response energy of each pixel in each preset direction is extracted. .

[0033] Specifically, the Log-Gabor filter can effectively capture multi-scale texture information in real-time captured images of the sealing edge of a soft-pack lithium battery. When processing mechanical wrinkles with periodic characteristics, it can generate highly representative response vectors. For example, when the filter direction is consistent with the direction of the mechanical wrinkle, the energy response in that direction will be much higher than in other directions. This set of energy values ​​serves as the basic feature flow for subsequent analysis, providing data support for revealing the physical orientation of the surface texture.

[0034] Thus, through multi-directional filter convolution processing, texture energy distribution features unaffected by light intensity can be effectively extracted, laying a solid foundation for distinguishing normal textures from abnormal defects.

[0035] S2. Evaluate the directional feature distribution of each pixel based on the energy amplitude of each pixel in different preset directions, so as to determine the anisotropy consistency index of each pixel.

[0036] In an optional embodiment, in the qualified sealing area of ​​the pouch lithium battery, the mechanical wrinkles physically manifest as a cluster of parallel peaks and troughs. This highly ordered structure means that the energy response exhibits an absolute peak in the direction perpendicular to the wrinkles, while the response is weaker in directions orthogonal to or deviating from them. Therefore, this invention constructs an anisotropic consistency index. This characterizes the concentration of energy in a specific direction and the degree of order in texture direction. Anisotropy uniformity index. The calculation method is as follows: ; in, For the first Energy amplitude in each direction; The corresponding filter angle is 1.5, which is the first constant term. It is added as a smoothing factor to the energy accumulation sum. The core logic of constructing the anisotropy consistency index is as follows: by calculating the sum of the cosine components and the sum of the sine components of the energy amplitude in each preset direction, the magnitude of these two components is calculated by vector synthesis to characterize the concentration of energy in a specific direction.

[0037] To more clearly illustrate the role and calculation process of the anisotropy consistency index, the following example will be used: First, assume that at a certain pixel, the energy amplitudes in its eight directions are as follows: other directions arrive The values ​​are all 0.1. At this point, the vector synthesis result of the molecules in the anisotropy consistency index calculation formula will mainly be determined by... The dominant value is close to 10, and the total energy in the denominator is approximately Calculated A value of approximately 0.82, close to 1, indicates that this point exhibits extremely strong directional consistency and belongs to a normal mechanical wrinkle region. Conversely, if leakage occurs, the energy in all directions becomes uniform, and the molecular components cancel each other out due to their opposite directions, causing the modulus to approach 0, resulting in a rapid drop in the anisotropy consistency index.

[0038] Thus, by calculating the anisotropic consistency index, the degree of directional order of the texture can be quantitatively described, enabling effective characterization and identification of normal mechanically wrinkled backgrounds.

[0039] S3. Calculate the phase distribution disorder and gray-level gradient of each pixel in the local area, and evaluate the infiltration and diffusion characteristics of each pixel in combination with the average gray-level gradient of the edge-sealing image, so as to determine the leakage and diffusion disturbance flux of each pixel.

[0040] In an optional embodiment, although anisotropy can reflect a decrease in directionality, simple image blurring can also lead to a decrease in consistency. The essential characteristic of electrolyte leakage is that the texture phase at the microscopic level of the aluminum-plastic film becomes completely disordered due to the wetting effect of the liquid. Therefore, this invention constructs a leakage diffusion perturbation flux. To obtain the phase standard deviation within a statistical local window Leakage is located by combining gray-scale gradient suppression factors and leakage diffusion disturbance flux. The calculation method is as follows: ; in, The local gradient is calculated using the Sobel operator. This represents the average gradient across the entire graph. 3 represents the second constant term. 2 is the third constant term. The core logic for constructing the leakage diffusion disturbance flux is as follows: The phase standard deviation is nonlinearly amplified using the natural logarithm function to characterize the disorder. Simultaneously, an inverse threshold is constructed using an exponential function; when the local gray-level gradient value is greater than the average gray-level gradient, the exponential term is negative, thereby suppressing [the disturbance]. The value.

[0041] To more clearly illustrate the role and calculation process of leakage diffusion disturbance flux, the present invention will now provide an example: First, assume that the phase distribution of a suspected leakage area is highly disordered. The value is 5, and the local gradient is affected by the blurring of the edges due to liquid wetting. The local gradient is 1, which is much lower than the average gradient of 10. Substituting this into the formula, we can obtain the result at this time. If the detection area is a relatively clear mechanical edge, then the local gradient... It is 20. Then its The final score will be suppressed.

[0042] Thus, by constructing a leakage diffusion perturbation flux, the phase disorder characteristics caused by liquid wetting can be accurately captured, effectively eliminating interference from normal structures and improving the specificity of leakage identification.

[0043] S4. The anisotropic consistency index and the leakage diffusion disturbance flux are nonlinearly fused, and the edge sealing quality of the soft-pack lithium battery is evaluated by combining the physical distance of the pixel point relative to the edge sealing centerline.

[0044] In an optional embodiment, this invention achieves a closed-loop evaluation of the sealing quality status of pouch lithium batteries by nonlinearly fusing two indicators: anisotropic consistency index and leakage diffusion perturbation flux, and combining this with the physical distance of a pixel relative to the sealing edge centerline. Since electrolyte leakage typically diffuses outward from the centerline where the sealing edge is most severely pressured, anomalies closer to the centerline have higher weights. Therefore, this invention constructs a quality evaluation index. To complete the final classification and quality assessment index. The calculation method is as follows: ; in, This represents the normalized physical parameter from the pixel to the edge centerline, with 0.5 being the fourth constant term. The logic for constructing the quality assessment index is as follows: based on the leakage diffusion disturbance flux. The squared value is used as the dominant criterion, and the anisotropy consistency index is utilized. As a suppression factor, when a pixel is determined to be leaking, its Large and Minimal, in the denominator It serves to adjust sensitivity, making The index increases quadratically. After obtaining the quality assessment index, it is compared with a preset defect judgment threshold. If the quality assessment index exceeds the preset defect judgment threshold, electrolyte leakage is determined and defect coordinates are generated.

[0045] Reference Figure 2 Normal edge sealing feature sampling points and electrolyte leakage feature sampling points are in Horizontal axis and The coordinate system formed by the vertical axis has excellent distinguishability. Normal edge sealing feature sampling points are distributed in the lower right, while electrolyte leakage feature sampling points are distributed in... Figure 2 In the upper left corner, there is a clear decision gap between the two. (Refer to...) Figure 3 High score The area appears as a bright, deep red spot, the shape of which conforms to the natural physical process of liquid wetting.

[0046] Thus, through nonlinear fusion and spatial weighting, a closed-loop evaluation of the sealing quality of pouch lithium batteries can be achieved, ensuring the accuracy and reliability of the test results.

[0047] This invention also discloses a quality inspection system for pouch lithium batteries, including a processor and a memory. The memory stores computer program instructions, which, when executed by the processor, implement a quality inspection method for pouch lithium batteries according to the present invention.

[0048] The system also includes other components well known to those skilled in the art, such as communication buses and communication interfaces, the settings and functions of which are known in the art and will not be described in detail here.

[0049] In the description of this specification, "multiple" or "several" means at least two, such as two, three or more, unless otherwise expressly and specifically defined.

Claims

1. A quality inspection method for soft-pack lithium batteries, characterized in that, include: The edge-sealing image of the soft-pack lithium battery is obtained, and a preset filter is used to perform convolution processing in multiple preset directions to obtain the energy amplitude of each pixel in different preset directions. Based on the energy amplitude of each pixel in different preset directions, the directional feature distribution of each pixel is evaluated to determine the anisotropy consistency index of each pixel. The phase distribution disorder and gray-level gradient of each pixel in the local region are calculated, and the wetting and diffusion characteristics of each pixel are evaluated by combining the average gray-level gradient of the edge-sealing image, so as to determine the leakage and diffusion perturbation flux of each pixel. The anisotropic consistency index and the leakage diffusion disturbance flux are nonlinearly fused, and the edge sealing quality of the soft-pack lithium battery is evaluated by combining the physical distance of the pixel point relative to the edge sealing centerline.

2. The quality inspection method for a soft-pack lithium battery according to claim 1, characterized in that, The anisotropy consistency index satisfies the following relationship: ； In the formula, Indicates the anisotropy consistency index; Indicates the pixel at the th Energy amplitude in a preset direction; Indicates the first The preset angle of each filter; This represents the first constant term that is preset.

3. The quality inspection method for a soft-pack lithium battery according to claim 1, characterized in that, The leakage diffusion disturbance flux satisfies the following relationship: ； In the formula, This represents the leakage diffusion disturbance flux; The standard deviation of pixel phase information within a local window; This represents the local grayscale gradient value of a pixel; This represents the average grayscale gradient of the edge-sealed image; This represents the pre-defined second constant term; This represents the pre-defined third constant term.

4. The quality inspection method for a soft-pack lithium battery according to claim 1, characterized in that, The edge banding quality status is determined by calculating a quality assessment index, which satisfies the following relationship: ； In the formula, Indicates a quality assessment index; Indicates the anisotropy consistency index; This represents the leakage diffusion disturbance flux; This represents the normalized distance parameter from a pixel to the preset edge centerline; This represents the pre-defined fourth constant term.

5. The quality inspection method for a soft-pack lithium battery according to claim 1, characterized in that, The process of performing convolution in multiple preset directions using a preset filter is specifically as follows: Convolution is performed in 8 preset directions using a Log-Gabor filter.

6. The quality inspection method for a soft-pack lithium battery according to claim 2, characterized in that, The determination of the anisotropy consistency index for each pixel includes: The sum of the cosine and sine components of the energy amplitude in each preset direction is calculated separately. The sum of the cosine and sine components is then vector-synthesized to calculate the magnitude, which characterizes the concentration of energy in a specific direction. Calculate the sum of energy amplitudes in each preset direction, and add the first constant term as a smoothing factor to the sum; The anisotropy consistency index is determined based on the ratio of the modulus to the sum after adding the smoothing factor.

7. The quality inspection method for a soft-pack lithium battery according to claim 3, characterized in that, The determination of the leakage diffusion perturbation flux at each pixel includes: The standard deviation of the pixel phase information within the local window is nonlinearly processed using the natural logarithm function to amplify the variation characteristics of the phase distribution disorder. An inverse threshold based on the local gray-level gradient value and the average gray-level gradient is constructed using an exponential function; The standard deviation after nonlinear processing is weighted using the inverse threshold, and the leakage diffusion disturbance flux is suppressed when the local gray-level gradient value is greater than the average gray-level gradient.

8. The quality inspection method for a soft-pack lithium battery according to claim 4, characterized in that, The evaluation of the sealing quality of the pouch lithium battery includes: The square value of the leakage diffusion disturbance flux is used as the dominant judgment factor, and the anisotropy consistency index is used as the inhibition factor to adjust the dominant judgment factor, so as to improve the evaluation value in areas with missing directionality and high disorder. The spatial weight is calculated based on the physical distance from the pixel to the preset edge centerline. The adjusted dominant judgment factor is then weighted using the spatial weight to obtain the quality evaluation index.

9. A quality inspection method for a soft-pack lithium battery according to claim 4, characterized in that, Also includes: A preset defect detection threshold is set. Compare the quality assessment index with the defect determination threshold; If the quality assessment index is greater than the defect determination threshold, it is determined that there is electrolyte leakage in the area corresponding to the pixel, and the defect location coordinates are generated.

10. A quality inspection system for soft-pack lithium batteries, characterized in that, include: A processor and a memory, the memory storing computer program instructions that, when executed by the processor, implement a quality testing method for a soft-pack lithium battery according to any one of claims 1-9.

Images