A method and device for detecting the alignment of positive and negative electrodes of a laminated lithium battery
By standardizing and pre-processing the Overhang area image of laminated lithium batteries, a contour segmentation model is constructed, and combined with deep learning technology to extract the contours and reference points of the cathode, the problem of inaccurate extraction of the cathode and anode points caused by the increase in the number of laminated layers is solved, and the problem of inaccurate extraction of the cathode points is improved. Detection accuracy and robustness.
Patent Information
- Application Number
- CN202310466610.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2023-04-23
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2043-04-23
AI Technical Summary
When detecting a laminated lithium battery, as the number of laminated layers increases, the extraction of the anode point is inaccurate, resulting in misjudgment or misjudgment. Especially in the case of blurred imaging, it is difficult to accurately judge the alignment of the battery cell.
By obtaining the Overhang area images of the four corners of the battery cell, performing direction standardization and preprocessing, a cathode and anode profile segmentation model is constructed, and the contour and reference points are extracted using a semantic segmentation network, combining the maximum inter-class variance method and local adaptive threshold segmentation. Calculate the vertical distance and judge the alignment.
It significantly improves the extraction accuracy of the contour and points of the cathode and anode sheet, enhances the robustness and versatility of the detection, and is suitable for laminated lithium batteries with a large number of layers, solving the problem of inaccurate extraction of the cathode and anode points.
Smart Images

Figure CN116721055B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of lithium battery detection, and in particular to a method and device for detecting the alignment of positive and negative electrode sheets of a laminated lithium battery. Background Art
[0002] With the explosive growth of the new energy vehicle industry, the demand for power lithium batteries has also increased, and the product quality of lithium batteries has also been increasingly valued. The alignment detection of lithium batteries is an important part of the lithium battery quality inspection process. When the cathode of the lithium battery is not covered by the anode, the lithium battery will produce lithium deposition during the charging and discharging process. When the lithium dendrites pierce the diaphragm between the cathode and the cathode, it will cause a short circuit in the battery, causing the risk of battery fire.
[0003] Among lithium batteries, laminated lithium batteries have the advantages of low internal resistance, high energy density, and variable shape, and are relatively widely used. At present, the imaging system of the alignment detection equipment for laminated lithium batteries basically adopts a micro-focus X-ray source and a flat-panel detector. The four corners of the laminated battery are imaged at a certain projection angle with the optical axis, and then the alignment of the cathode and anode plates is detected using image processing algorithms or deep learning algorithms for the acquired images.
[0004] Patent document CN115035061A discloses a method for extracting the cathode and anode points of a laminated battery cell group and detecting the alignment of the cathode and anode poles. The method marks the cathode and anode end points as key points and uses the deep learning key point detection network HRNet to train a model for extracting the cathode and anode points, and then performs alignment detection.
[0005] The anode and cathode point extraction method based on the key point extraction network has a good extraction effect when the number of layers of laminated lithium batteries is small and there is no anode and cathode stacking phenomenon in the imaging. However, with the development of technology, the number of anode and cathode stacking layers of laminated lithium batteries is increasing. Due to the x-ray perspective mechanism, the probability of blurring the anode and cathode on both sides of the x-ray image is getting higher and higher. When using this method, some anode and cathode points will fail to be extracted due to imaging blur, resulting in misjudgment or omission. Summary of the invention
[0006] The purpose of the present invention is to provide a method and device for detecting the alignment of the positive and negative electrodes of a laminated lithium battery, so as to solve the problem in the prior art that the positive and negative electrodes of the battery cell are not accurately extracted when the number of lithium battery laminates is large.
[0007] In order to achieve the above purpose, the following technical solutions are adopted:
[0008] A method for detecting the alignment of positive and negative electrodes of a laminated lithium battery comprises the following steps:
[0009] Obtain the overhang area images of the four corners of the battery cell and standardize the image directions to keep the imaging directions of multiple images consistent;
[0010] In the Overhang area image after normalization in each image direction, the cathode detection area and the anode detection area of interest are extracted and preprocessed;
[0011] Pre-constructing a cathode contour segmentation model and an anode contour segmentation model, and extracting the contour of the cathode pole piece and the contour of the anode pole piece in the pre-processed cathode detection area and the anode detection area respectively according to the cathode contour segmentation model and the anode contour segmentation model;
[0012] Extract the anode reference point of the anode pole piece contour, and extract the cathode reference point of the cathode pole piece contour;
[0013] Based on the extracted cathode reference point and anode reference point, determine whether the alignment of the cathode and anode electrodes is qualified.
[0014] Furthermore, the extracting of the cathode detection area and the anode detection area of interest specifically includes the following steps:
[0015] Based on the maximum inter-class variance method, the Overhang area image after direction normalization is threshold segmented to obtain the first segmented image I OtsuThresh ;
[0016] Based on the local adaptive threshold segmentation method, the Overhang area image after direction normalization is subjected to threshold segmentation to obtain the second segmented image I adaptThesh ;
[0017] Take the first segmented image I OtsuThresh and the second segmented image I adaptThesh The third segmented image I is obtained thresh ;
[0018] For the third segmented image I thresh , perform Blob analysis and obtain the image region with the largest area I mask ;
[0019] Based on the acquisition of I mask And the known number of battery cell winding layers, extract the cathode detection area and anode detection area of interest.
[0020] Furthermore, the extracted cathode detection area and anode detection area of interest are preprocessed using the following formula:
[0021]
[0022] Among them, I roiTo extract the cathode detection area or anode detection area of interest, For I ro i The gradient along the horizontal direction, η is the magnification factor, I enhance is the preprocessed image.
[0023] Furthermore, the pre-constructing of the cathode contour segmentation model and the anode contour segmentation model specifically comprises the following steps:
[0024] Acquire several pre-processed regional images in advance, and mark the contours of the cathode electrode piece and the contours of the anode electrode piece in the images respectively;
[0025] The sample set of cathode pole piece contour and the sample set of anode pole piece contour generated according to the above annotation are divided into a training set and a validation set at a ratio of 8:2, and the semantic segmentation network Unet++ model and the pytorch deep learning framework are used for model training to obtain the PTH training model of the cathode pole piece contour and the PTH training model of the anode pole piece contour respectively;
[0026] Based on the PTH training model obtained above, it is converted into an ONNX model, and the TensorRT model is used to quantize it to obtain the cathode contour segmentation model and the anode contour segmentation model.
[0027] Furthermore, the extracting of the contour of the cathode electrode piece and the contour of the anode electrode piece respectively specifically comprises the following steps:
[0028] The cathode contour segmentation model and the anode contour segmentation model are respectively called to infer the preprocessed regional image to obtain the cathode pole piece contour segmentation result map and the anode pole piece contour segmentation result map respectively;
[0029] Based on the Otsu threshold segmentation algorithm, the contour of the cathode pole piece and the contour of the anode pole piece are extracted in the cathode pole piece contour segmentation result image and the anode pole piece contour segmentation result image respectively.
[0030] Furthermore, the point coordinates of the end point of the tail of each anode pole piece contour are extracted and used as the anode reference point of the anode pole piece contour.
[0031] Furthermore, the extraction of the cathode reference point of the cathode pole piece contour specifically comprises the following steps:
[0032] Extract the point coordinates of the beginning and end points of each anode pole piece outline;
[0033] Connect the end points of the contours of two adjacent anode plates and draw a perpendicular line between them;
[0034] The point where each perpendicular midline intersects the corresponding end of the cathode pole piece contour is taken as the cathode reference point.
[0035] Further, judging whether the alignment of the anode and cathode plates is qualified specifically includes the following steps:
[0036] Preset the maximum value of the standard Overhang length l max and the minimum value l min ;
[0037] Calculate the vertical distance between the anode reference point of each anode pole piece contour and the cathode reference point of its corresponding cathode pole piece contour;
[0038] If the vertical distance between the anode reference point of each anode pole piece contour and the cathode reference point of its corresponding cathode pole piece contour falls within [l min , l max ] range, the alignment of the anode and cathode plates is considered qualified, otherwise it is unqualified.
[0039] Further, calculating the vertical distance between the anode reference point of each anode pole piece contour and the cathode reference point of its corresponding cathode pole piece contour specifically includes the following steps:
[0040] The anode reference point located in the first column is calculated from its vertical distance to the cathode reference point located in the first column;
[0041] For a number of anode reference points located between the first column and the last column, the vertical distance between each anode reference point and two adjacent cathode reference points located to its left and right is calculated respectively;
[0042] The anode reference point located in the last column is calculated from its vertical distance to the cathode reference point located in the last column.
[0043] Also provided is a device for detecting the alignment of the positive and negative electrodes of a laminated lithium battery, comprising an x-ray source, an x-ray receiver, and a supporting platform mechanism arranged between the x-ray source and the x-ray receiver and capable of driving the lithium battery to be tested to perform rotational motion.
[0044] By adopting the above scheme, the beneficial effects of the present invention are:
[0045] This method significantly improves the contrast between the cathode, anode and the background area through image preprocessing technology, so as to more accurately extract the contours of the cathode and anode sheets and the cathode and anode points, thereby improving the detection accuracy. At the same time, compared with the alignment detection algorithm based on image processing methods such as threshold segmentation and threshold extraction, or the alignment detection algorithm based on key point extraction, this detection algorithm is more versatile and more robust, and can be applied to the detection of stacked lithium batteries with a large number of layers, solving the problem of inaccurate extraction of the cathode and anode points of the battery cell when the number of lithium battery stacking layers is large. BRIEF DESCRIPTION OF THE DRAWINGS
[0046] Figure 1 is a flow chart of the detection method of the present invention;
[0047] Figure 2 It is a structural schematic diagram of the detection device of the present invention;
[0048] Figure 3 In one embodiment of the present invention, an Overhang image of a laminated lithium battery is shown;
[0049] Figure 4 For the present invention, in one embodiment, the direction is normalized;
[0050] Figure 5 In one embodiment of the present invention, the first segmented image diagram;
[0051] Figure 6 In one embodiment of the present invention, the second segmented image diagram;
[0052] Figure 7 In one embodiment of the present invention, a result image of superimposing the first segmented image and the second segmented image;
[0053] Figure 8 In one embodiment of the present invention, Figure 7 The image after screening the contours of the anode and cathode;
[0054] Fig. 9 In one embodiment of the present invention, the result diagram of the cathode detection area and the anode detection area extracted;
[0055] Fig.10 In one embodiment of the present invention, a partial diagram of the cathode and anode detection areas;
[0056] Fig.11 In one embodiment of the present invention, Fig.10 The image after preprocessing;
[0057] Fig.12 In one embodiment of the present invention, a cathode contour segmentation model construction flow chart;
[0058] Fig.13 In one embodiment of the present invention, a result diagram of marking the cathode electrode contour is shown;
[0059] Fig.14 In one embodiment of the present invention, a result diagram of reasoning based on a cathode contour segmentation model is provided;
[0060] Fig.15 In one embodiment of the present invention, Fig.14The result image after threshold segmentation;
[0061] Fig.16 In one embodiment of the present invention, a flow chart of constructing an anode contour segmentation model;
[0062] Fig.17 In one embodiment of the present invention, a result diagram of marking the contour of the anode pole piece is shown;
[0063] Fig.18 In one embodiment of the present invention, a result diagram of reasoning based on the anode contour segmentation model;
[0064] Fig.19 In one embodiment of the present invention, Fig.18 The result image after threshold segmentation;
[0065] Fig. 20 In one embodiment of the present invention, a final test result diagram is shown;
[0066] The accompanying drawings illustrate:
[0067] 1—X-ray source; 2—X-ray receiver;
[0068] 3—cathode electrode; 4—anode electrode;
[0069] 5—Cathode reference point; 6—Anode reference point. DETAILED DESCRIPTION
[0070] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments.
[0071] Reference Figures 1 to 20 As shown, the present invention provides a method for detecting the alignment of positive and negative electrodes of a laminated lithium battery, comprising the following steps:
[0072] Obtain the overhang area images of the four corners of the battery cell and standardize the image directions to keep the imaging directions of multiple images consistent;
[0073] In the Overhang area image after normalization in each image direction, the cathode detection area and the anode detection area of interest are extracted and preprocessed;
[0074] Pre-constructing a cathode contour segmentation model and an anode contour segmentation model, and extracting the contour of the cathode pole piece 3 and the contour of the anode pole piece 4 in the pre-processed cathode detection area and the anode detection area according to the cathode contour segmentation model and the anode contour segmentation model;
[0075] Extracting the anode reference point 6 of the contour of the anode pole piece 4, and extracting the cathode reference point 5 of the contour of the cathode pole piece 3;
[0076] Based on the extracted cathode reference point 5 and anode reference point 6, determine whether the alignment of the cathode and anode electrodes is qualified.
[0077] Continue to refer to Figure 2 As shown, it is first necessary to obtain the Overhang area images of the four corners of the battery cell. Therefore, in one embodiment, a detection device for the alignment of the cathode and anode sheets of a laminated lithium battery cell is also provided, which is used to detect the alignment of the cathode and anode sheets of the battery cell, including an x-ray source 1, an x-ray receiver 2, and a carrier mechanism arranged between the x-ray source 1 and the x-ray receiver 2 and capable of driving the lithium battery to be tested to rotate, wherein the x-ray source 1 and the x-ray receiver 2 are arranged at intervals (can be vertical), the carrier mechanism is arranged between the two, the x-ray source 1 is used as the light source, and the x-ray receiver 2 (can be a flat plate) is used as the carrier mechanism. The detector is used as a ray receiver, and the lithium battery to be detected is driven to rotate through the carrier mechanism, and then the images of the four corners of the battery cell are collected at a certain shooting angle (the battery cell is divided into four areas in advance, that is, the images of the four areas are collected); after the images are collected (usually four images), due to the different collection viewing angles and angles, the image needs to be standardized in direction for subsequent analysis and processing. In this embodiment, the direction of the image in the lower left corner of the battery cell can be used as a reference, and the remaining images can be rotated to the lower left corner (the transpose and flip functions in OpenCV can be used to rotate the image) to keep the image direction consistent, and obtain a direction-standardized image, such as Figure 4 As shown, it is convenient to uniformly extract the cathode detection area and the anode detection area of interest, and pre-process them.
[0078] After the image acquisition is completed, in one embodiment, in order to improve the detection accuracy, it is necessary to extract the cathode detection area and the anode detection area of interest. Considering that a single threshold segmentation method cannot completely segment the battery mask, in this application, the maximum inter-class variance method is combined with the local adaptive threshold segmentation method to segment the battery mask, thereby completing the extraction of the cathode detection area and the anode detection area. The specific steps are as follows:
[0079] First, if Figure 5 As shown, the Overhang area image I after direction normalization based on the maximum inter-class variance method standard , perform threshold segmentation to obtain the first segmented image I OtsuThresh ;
[0080] Then, based on the local adaptive threshold segmentation method, the direction-normalized Overhang region image I standard , perform threshold segmentation to obtain the second segmented image I adaptThesh, the result is as follows Figure 6 As shown;
[0081] Then, take the first segmented image I OtsuThresh and the second segmented image I adaptThesh The third segmented image I is obtained thresh , that is I thresh =I OtsuThresh YI adaptThesh , the result is as follows Figure 7 As shown;
[0082] Then, for the third segmented image I thresh , perform Blob analysis and obtain the image region with the largest area I mask , the result is as follows Figure 8 As shown;
[0083] Finally, if Fig. 9 As shown, based on the obtained I mask and known prior information (such as the number of battery cell winding layers) to extract the cathode detection area and anode detection area of interest.
[0084] In order to improve the contrast between the contours of the cathode electrode piece 3 and the anode electrode piece 4 extracted subsequently and the background, in one embodiment, the extracted cathode detection area and the anode detection area of interest are subjected to image preprocessing to enhance the image. From the images of the cathode detection area and the anode detection area obtained, it can be seen that the contours of the cathode and anode electrode pieces 4 of the laminated lithium battery have a relatively strong gradient in the horizontal direction. Therefore, it can be considered to multiply the gradient in the horizontal direction by a certain coefficient and superimpose it on the original image. The expression is as follows:
[0085]
[0086] Among them, I roi To extract the cathode detection area or anode detection area of interest, For I ro i The gradient along the horizontal direction, η is the magnification factor (here it is 1), I enhance is the preprocessed image, in the form of the first-order partial derivative of a two-dimensional Gaussian function, Discretize it into a 5x5 first-order differential operator as follows:
[0087]
[0088] Combined with the above expression, the convolution kernel of the following formula is used to perform convolution operation on the cathode detection area or the anode detection area to obtain an enhancement effect to obtain a preprocessed image, such as Fig.11 As shown,
[0089]
[0090] After image preprocessing and enhancement, the contours of the cathode and anode sheets need to be extracted. The extraction operation is divided into two parts: offline and online. The offline part includes sample annotation, model training, construction and deployment, and the online part includes anode model reasoning and contour screening. First, the cathode contour segmentation model and the anode contour segmentation model need to be constructed. The construction of the cathode contour segmentation model is used as an example (the process is as follows Fig.12 shown):
[0091] After obtaining several pre-processed enhanced regional images, the wizard annotation assistant can be used to annotate the cathode pole piece 3 contour to obtain a mask image of the cathode pole piece 3 contour (such as Fig.13 As shown); then, for the sample set generated by the above annotation, it is divided into a training set and a validation set at a ratio of 8:2. In order to accurately extract the contour of the cathode electrode 3, the semantic segmentation network Unet++ model and the pytorch deep learning framework are used for model training; for the PTH model obtained after training, it is converted into an ONNX (Open Neural Network Exchange) model. ONNX is an open file format designed for machine learning and is used to store trained models; in order to reduce the time consumption of model reasoning, TensorRT is used for model reasoning, which can significantly improve the reasoning performance of the network. Therefore, TensorRT is used to quantize ONNX to obtain the engine model (cathode contour segmentation model). The construction and training of the anode contour segmentation model are similar to it, which will not be repeated here.
[0092] In a preferred embodiment, after constructing the cathode contour segmentation model and the anode contour segmentation model, the model can be called using TensorRT C++ to infer the direction-annotated image and obtain the segmentation result map of the model, that is, based on the cathode contour segmentation model and the anode contour segmentation model, the contour segmentation result map of the cathode pole piece 3 and the contour segmentation result map of the anode pole piece 4 are obtained respectively. In order to enhance the image segmentation accuracy and effect and eliminate some interfering contours, the contour of the cathode pole piece 3 and the contour of the anode pole piece 4 can be extracted in the cathode pole piece 3 contour segmentation result map and the anode pole piece 4 contour segmentation result map respectively based on the Otsu threshold segmentation algorithm.
[0093] After obtaining the contours of the cathode electrode piece 3 and the anode electrode piece 4, it is necessary to extract the cathode reference point 5 and the anode reference point 6. In one embodiment, the point coordinates of the end point of the contour of each anode electrode piece 4 are extracted (eg Fig. 20 The topmost end point of the anode pole piece 4 outline shown in FIG. 4 is used as the anode reference point 6 of the anode pole piece 4 outline, thereby obtaining the point coordinate set A of the anode reference point 6 = {a 1 ,a2 , Λ, a n , a n+1}, where n is the number of contours of the anode plate 4. For the extraction of the cathode reference point 5, to improve the accuracy of its extraction, in one embodiment, the following method is used for extraction, that is, first extract the point coordinates of the starting endpoint of each contour of the anode plate 4; then, connect the starting endpoints of every two adjacent contours of the anode plate 4 and draw the perpendicular bisector thereof; take the point where each perpendicular bisector intersects the end of the corresponding contour of the cathode plate 3 as the cathode reference point 5, so as to obtain the set C of the cathode reference points 5 = {c 1 , c 2 , Λ, c n}, where n is the number of contours of the cathode plate 3.
[0094] After the cathode reference point 5 and the anode reference point 6 are extracted, the alignment degree of the anode and cathode plates can be detected next. According to the structure of the stacked lithium battery, as Fig. 20 shown, from left to right, except that only one column of the corresponding cathode plate 3 can be found for the first column of the anode plate 4 and the last column of the anode plate 4, the middle anode plates 4 correspond to one cathode plate 3 on each of their left and right sides. Therefore, when calculating the vertical distance (i.e., the distance in the y-axis direction) between the cathode reference point 5 and the anode reference point 6, it is necessary to calculate it by classification:
[0095] 1) When i = 1, at this time, a i is the contour of the first column of the anode plate 4, and only the cathode reference point 5 of the first column corresponds to it:
[0096] l i_right = η(c 1 .y - a 1 .y),
[0097] 2) When 1 < i < n + 1, at this time, a i is the contour of the anode plate 4 in the middle column, and there are two cathode reference points 5 on its left and right corresponding to it:
[0098]
[0099] 3) When i = n + 1, at this time, a i is the contour of the last column of the anode plate 4, and only the cathode reference point 5 of the last column corresponds to it:
[0100] l i_left = η(c n .y - a n+1 .y),
[0101] where l i_right is the alignment length from the contour of the i-th anode plate 4 to the contour of the cathode plate 3 in the right column; l i_leftis the alignment length of the contour of the i-th anode electrode piece 4 to the contour of the cathode electrode piece 3 in the left column; η represents how many physical dimensions a pixel represents, and the unit is mm / pixel, c i .y is the y-axis coordinate value of the i-th cathode reference point 5, a i .y is the y-axis coordinate value of the i-th anode reference point 6.
[0102] After calculating the vertical distance between the anode reference point 6 of each anode pole piece 4 contour and the cathode reference point 5 of the corresponding cathode pole piece 3 contour, compare it with the preset maximum value of the Overhang length l max and the minimum value l min Compare, if all in [l min , l max ] range, the alignment of the anode and cathode plates is judged to be qualified, that is, the battery cell is ok, otherwise it is unqualified. In one embodiment, the final test result is as follows Fig. 20 shown.
[0103] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.
Claims
1. A method for detecting the alignment of positive and negative electrodes of a laminated lithium battery, characterized in that: The steps include: Obtain the overhang area images of the four corners of the battery cell and standardize the image directions to keep the imaging directions of multiple images consistent; In the Overhang area image after normalization in each image direction, the cathode detection area and the anode detection area of interest are extracted and preprocessed; Pre-constructing a cathode contour segmentation model and an anode contour segmentation model, and extracting the contour of the cathode pole piece and the contour of the anode pole piece in the pre-processed cathode detection area and the anode detection area respectively according to the cathode contour segmentation model and the anode contour segmentation model; Extract the anode reference point of the anode pole piece contour, and extract the cathode reference point of the cathode pole piece contour; Based on the extracted cathode reference point and anode reference point, determine whether the alignment of the cathode and anode electrodes is qualified; The extracting of the cathode detection area and the anode detection area of interest specifically comprises the following steps: Based on the maximum inter-class variance method, the Overhang area image after direction normalization is threshold segmented to obtain the first segmented image ; Based on the local adaptive threshold segmentation method, the Overhang area image after direction normalization is threshold segmented to obtain the second segmented image ; Take the first segmented image and the second segmented image The third segmented image is obtained by ; For the third segmented image , perform Blob analysis to obtain the largest image area ; Acquisition-based And the known number of battery cell winding layers, extract the cathode detection area and anode detection area of interest.
2. The method for detecting the alignment of the positive and negative electrodes of a laminated lithium battery according to claim 1, characterized in that: The extracted cathode detection area and anode detection area of interest are preprocessed using the following formula: , in, To extract the cathode detection area or anode detection area of interest, for The gradient along the horizontal direction, is the magnification factor, is the preprocessed image.
3. The method for detecting the alignment of the positive and negative electrodes of a laminated lithium battery according to claim 1, characterized in that: The pre-construction of the cathode contour segmentation model and the anode contour segmentation model specifically includes the following steps: Acquire several pre-processed regional images in advance, and mark the contours of the cathode electrode piece and the contours of the anode electrode piece in the images respectively; The sample set of cathode pole piece contour and the sample set of anode pole piece contour generated according to the above annotation are divided into a training set and a validation set at a ratio of 8:2, and the semantic segmentation network Unet++ model and the pytorch deep learning framework are used for model training to obtain the PTH training model of the cathode pole piece contour and the PTH training model of the anode pole piece contour respectively; Based on the PTH training model obtained above, it is converted into an ONNX model, and the TensorRT model is used to quantize it to obtain the cathode contour segmentation model and the anode contour segmentation model.
4. The method for detecting the alignment of the positive and negative electrodes of a laminated lithium battery according to claim 3, characterized in that: The steps of respectively extracting the contour of the cathode electrode piece and the contour of the anode electrode piece specifically include the following steps: The cathode contour segmentation model and the anode contour segmentation model are respectively called to infer the preprocessed regional image to obtain the cathode pole piece contour segmentation result map and the anode pole piece contour segmentation result map respectively; Based on the Otsu threshold segmentation algorithm, the contour of the cathode pole piece and the contour of the anode pole piece are extracted in the cathode pole piece contour segmentation result image and the anode pole piece contour segmentation result image respectively.
5. The method for detecting the alignment of positive and negative electrodes of a laminated lithium battery according to claim 1, characterized in that: The point coordinates of the end point of each anode pole piece contour are extracted and used as the anode reference point of the anode pole piece contour.
6. The method for detecting the alignment of positive and negative electrodes of a laminated lithium battery according to claim 5, characterized in that: The cathode reference point of the cathode electrode contour is extracted, which specifically includes the following steps: Extract the point coordinates of the beginning and end points of each anode pole piece outline; Connect the end points of the contours of two adjacent anode plates and draw a perpendicular line between them; The point where each perpendicular midline intersects the corresponding end of the cathode pole piece contour is taken as the cathode reference point.
7. The method for detecting the alignment of positive and negative electrodes of a laminated lithium battery according to claim 1, characterized in that: Judging whether the alignment of the anode and cathode plates is qualified includes the following steps: Preset the maximum value of the standard overhang length and minimum value ; Calculate the vertical distance between the anode reference point of each anode pole piece contour and the cathode reference point of its corresponding cathode pole piece contour; If the vertical distance between the anode reference point of each anode pole piece contour and the cathode reference point of its corresponding cathode pole piece contour falls within If the alignment degree of the cathode and anode pieces is within the interval, it is judged to be qualified, otherwise it is unqualified.
8. The method for detecting the alignment of positive and negative electrodes of a laminated lithium battery according to claim 7, characterized in that: Calculating the vertical distance between the anode reference point of each anode pole piece contour and the cathode reference point of the corresponding cathode pole piece contour specifically includes the following steps: The anode reference point located in the first column is calculated from its vertical distance to the cathode reference point located in the first column; For a number of anode reference points located between the first column and the last column, the vertical distance between each anode reference point and two adjacent cathode reference points located to its left and right is calculated respectively; The anode reference point located in the last column is calculated from its vertical distance to the cathode reference point located in the last column.
9. A device for detecting the alignment of positive and negative electrodes of a laminated lithium battery, used for detecting the alignment of positive and negative electrodes of a laminated lithium battery using the detection method described in any one of claims 1 to 8, characterized in that: The invention comprises an x-ray source, an x-ray receiver, and a bearing platform mechanism which is arranged between the x-ray source and the x-ray receiver and can drive the lithium battery to be tested to perform rotational motion.
Citation Information
Patent Citations
Method for extracting cathode and anode points of battery cell pole group and detecting alignment degree of cathode and anode pieces
CN115035061A
An infrared target detection and tracking method
CN109902578A
Method and device for detecting alignment degree of pole pieces of battery cell pole group
CN112330623A
Battery cell detection method and device and equipment
CN112577421A
Battery cell anomaly detection method and system of laminated battery
CN114581446A
