A method for extracting positive and negative pole points of a battery cell group and detecting the alignment of positive and negative pole pieces

Abstract

The present invention discloses a method for extracting the positive and negative poles of a cell electrode group and detecting the alignment of positive and negative pole sheets. The extraction method comprises: obtaining an imaging image of a cell electrode group; annotating the positive and negative poles of the pole sheet in the cell electrode group according to the imaging image to generate an annotated image; training a high-resolution model for key point detection according to a training set and a verification set generated from the annotated image to generate an extreme point extraction model; extracting the positive and negative poles in the imaging image of the cell electrode group to be detected according to the extreme point extraction model. By implementing the present invention, the extreme point extraction model obtained by training the high-resolution model for key point detection is used to extract the positive and negative poles in the cell electrode group. Compared with the problem of inaccurate extraction of the positive and negative poles by edge detection and other methods in the prior art, this method can accurately extract the positive and negative poles in the cell electrode group. At the same time, when annotating to form a training set and a verification set, the coordinates of the positive and negative poles in the image are annotated to further improve the accuracy of subsequent extraction of the positive and negative poles.

Description

Technical Field

[0001] The present invention relates to the technical field of battery cell detection, and in particular to a method for extracting positive and negative poles of a battery cell electrode group and detecting the alignment of positive and negative electrode sheets. Background Art

[0002] With the rapid development of the new energy industry, electric vehicles powered by lithium-ion batteries have begun to be widely used, and the demand for power batteries is growing. The quality of power batteries not only affects the battery life, but also affects driving safety.

[0003] The power battery pack is composed of a single battery, and a single battery is composed of several battery cell groups. During the production process of each battery cell group, taking the lamination process as an example, it is necessary to detect whether the state of the positive and negative electrodes in the battery cell group meets the process standards. If the electrode state is unqualified, it needs to be detected and scrapped. Therefore, the development of automatic detection technology for the alignment of positive and negative electrodes in battery cell groups has important engineering and commercial application value.

[0004] At present, most of the industry uses X-rays to image the battery cell pole group, extracts the positive and negative poles in the battery cell pole group through the image obtained, and judges whether the battery cell pole group is qualified based on the distance between the positive and negative poles. However, due to the easy attenuation of X-ray tubes and changes in the power of the rays, the imaging is unstable. Therefore, the current gray threshold extraction, edge detection and other algorithms for extracting the positive and negative poles have the problem of inaccurate extraction and the algorithm cannot effectively detect. Summary of the invention

[0005] In view of this, an embodiment of the present invention provides a method for extracting the cathode and anode points of a battery cell electrode group and detecting the alignment of cathode and anode plates, so as to solve the technical problem of inaccurate extraction of the cathode and anode points of a battery cell electrode group in the prior art.

[0006] The technical solution proposed by the present invention is as follows:

[0007] A first aspect of an embodiment of the present invention provides a method for extracting the positive and negative poles of a battery cell electrode group, comprising: obtaining an imaging image of the battery cell electrode group; marking the positive and negative poles of the electrode pieces in the battery cell electrode group according to the imaging image to generate a marked image, wherein the marked image includes the coordinates of the positive and negative poles; training a high-resolution model for key point detection according to a training set and a validation set generated from the marked image to generate a pole extraction model; and extracting the positive and negative poles in the imaging image of the battery cell electrode group to be detected according to the pole extraction model.

[0008] Optionally, after acquiring the imaging image of the battery cell electrode group, the method further includes: screening the electrode piece imaging image in the imaging image; rotating the screened electrode piece imaging image to obtain a electrode piece image with a consistent imaging direction.

[0009] Optionally, a high-resolution key point detection model is trained according to a training set and a validation set generated from the annotated image to generate a pole extraction model, including: using data enhancement technology to process the annotated image to obtain an enhanced annotated image; and training a high-resolution key point detection model according to a training set and a validation set generated from the enhanced annotated image to generate a pole extraction model.

[0010] Optionally, the positive and negative poles in the imaging image of the pole group of the battery cell to be tested are extracted according to the pole extraction model, including: intercepting the image of the region of interest in the imaging image of the pole group of the battery cell to be tested, and converting the image of the region of interest into a three-channel image; performing affine transformation and torch form conversion on the converted image; and inputting the converted image into the pole extraction model to extract the positive and negative poles.

[0011] A second aspect of an embodiment of the present invention provides a method for detecting the alignment of cathode and cathode plates of a battery cell electrode group, comprising: correcting the anode point extracted by the cathode and cathode point extraction method of the battery cell electrode group described in the first aspect of the embodiment of the present invention and any one of the first aspects to obtain a corrected anode point, wherein the corrected anode point and the extracted cathode point have the same vertical coordinate; and performing alignment detection based on the distance between the corrected anode point and the extracted cathode point as an alignment detection condition.

[0012] Optionally, the extracted anode point is corrected to obtain a corrected anode point, including: performing image binarization and edge detection on the cathode plate area image in the imaging image of the electrode group of the battery cell to be detected to obtain a straight line where any cathode plate is located; calculating the slope of the straight line where any cathode plate is located based on interpolation and least squares method; and correcting the anode point according to the slope and the extracted cathode and anode points to obtain a corrected anode point.

[0013] Optionally, the method for detecting the alignment of the cathode and cathode plates of the battery cell electrode group also includes: calculating the distance between the highest point among the extracted cathode points and the lowest point among the corrected anode points to obtain the maximum drop; calculating the maximum value of the distance between the multiple layers of corrected anode points and the maximum value of the distance between the multiple layers of extracted cathode points to obtain the local drop between the cathode and cathode; performing alignment detection based on the maximum drop and the local drop between the cathode and cathode as alignment detection conditions.

[0014] Optionally, the maximum value of the distance between the anode points after multi-layer correction and the maximum value of the distance between the cathode points extracted from the multi-layer are calculated to obtain the local difference between the anode and cathode, including: calculating the anode difference of each layer between the preset number of layers of anode sheets in the battery cell electrode group, the preset number of layers on both sides of the battery cell electrode group is less than the preset number of layers in the middle of the battery cell electrode group; screening the maximum value between the anode differences as the local anode difference; calculating the cathode difference between each layer on the left and right sides of each cathode sheet in the battery cell electrode group; screening the maximum value between the cathode differences as the local cathode difference.

[0015] A third aspect of an embodiment of the present invention provides a device for extracting the positive and negative poles of a battery cell electrode group, comprising: an image acquisition module for acquiring an imaging image of the battery cell electrode group; a labeling module for labeling the positive and negative poles of the electrode pieces in the battery cell electrode group according to the imaging image, and generating a labeled image, wherein the labeled image includes the coordinates of the positive and negative poles; a model generation module for training a high-resolution model for key point detection according to a training set and a verification set generated from the labeled image, and generating a pole extraction model; an extraction module for extracting the positive and negative poles in the imaging image of the battery cell electrode group to be detected according to the pole extraction model.

[0016] A fourth aspect of an embodiment of the present invention provides an alignment detection device for cathode and cathode plates of a battery cell electrode group, comprising: a correction module, used to correct the anode point extracted by the cathode and cathode point extraction method of the battery cell electrode group according to the first aspect of the embodiment of the present invention and any one of the opposite aspects, to obtain a corrected anode point, wherein the corrected anode point and the extracted cathode point have the same vertical coordinate; a detection module, used to perform alignment detection based on the distance between the corrected anode point and the extracted cathode point as an alignment detection condition.

[0017] A fifth aspect of an embodiment of the present invention provides a computer-readable storage medium, which stores computer instructions, and the computer instructions are used to enable the computer to execute the method for extracting the anode and cathode points of a battery cell electrode group as described in the first aspect of the embodiment of the present invention and any one of the first aspects, and the method for detecting the alignment of the anode and cathode plates of a battery cell electrode group as described in the second aspect of the embodiment of the present invention and any one of the second aspects.

[0018] A sixth aspect of an embodiment of the present invention provides an electronic device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the computer instructions to perform the extraction of anode and cathode points of a battery cell electrode group as described in the first aspect of the embodiment of the present invention and any one of the first aspects, and the detection method for the alignment of anode and cathode plates of a battery cell electrode group as described in the second aspect of the embodiment of the present invention and any one of the second aspects.

[0019] The technical solution provided by the present invention has the following effects:

[0020] The method for extracting the positive and negative poles of a battery cell electrode group provided in an embodiment of the present invention uses a pole extraction model obtained by training a high-resolution model for key point detection to extract the positive and negative poles in the battery cell electrode group. Compared with the problem of inaccurate extraction of positive and negative poles by edge detection and other methods in the prior art, this method can accurately extract the positive and negative poles in the battery cell electrode group. At the same time, when annotating to form a training set and a validation set, the coordinates of the positive and negative poles in the image are annotated to further improve the accuracy of subsequent extraction of the positive and negative poles.

[0021] The method for detecting the alignment of the positive and negative electrodes of a battery cell electrode group provided in an embodiment of the present invention first corrects the determined anode point so that it falls at the same position as the vertical coordinate of the corresponding cathode point, and then connects the extracted cathode point and the corrected anode point, and uses the obtained distance as the detection condition. Compared with the current method of directly connecting the anode point and the cathode point and detecting based on the distance between the two, the detection standard is changed, and the missed detection rate and over-kill rate of the product are reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0023] Figure 1 is a flow chart of a method for extracting cathode and anode points of a battery cell electrode group according to an embodiment of the present invention;

[0024] Figure 2 It is a schematic diagram of an imaging image of a battery cell electrode group;

[0025] Figure 3 and Figure 4 is a schematic diagram of annotated images according to an embodiment of the present invention;

[0026] Figure 5 is a schematic diagram of annotated text according to an embodiment of the present invention;

[0027] Figure 6 is a flow chart of a method for detecting the alignment of cathode and anode pieces of a battery cell electrode group according to an embodiment of the present invention;

[0028] Figure 7 is a flow chart of a method for detecting the alignment of cathode and anode pieces of a battery cell electrode group according to another embodiment of the present invention;

[0029] Figure 8 is a schematic diagram of an extracted cathode point and a corrected anode point according to an embodiment of the present invention;

[0030] Fig. 9 is a flow chart of a method for detecting the alignment of cathode and anode pieces of a battery cell electrode group according to another embodiment of the present invention;

[0031] Fig.10 is a schematic diagram of the detection effect of the method for detecting the alignment of the cathode and anode pieces of a battery cell electrode group according to an embodiment of the present invention;

[0032] Fig.11 is a schematic diagram of the detection effect of a method for detecting the alignment of cathode and anode pieces of a battery cell electrode group according to another embodiment of the present invention;

[0033] Fig.12 is a structural block diagram of a device for extracting cathode and anode points of a battery cell group according to an embodiment of the present invention;

[0034] Fig.13 is a structural block diagram of a device for detecting alignment of cathode and anode pieces of a battery cell electrode group according to an embodiment of the present invention;

[0035] Fig.14 is a schematic diagram of the structure of a computer-readable storage medium provided according to an embodiment of the present invention;

[0036] Fig.15 is a schematic structural diagram of an electronic device provided according to an embodiment of the present invention. DETAILED DESCRIPTION

[0037] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0038] The terms "first", "second", "third", "fourth", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0039] According to an embodiment of the present invention, a method for extracting the anode and cathode points of a battery cell group and detecting the alignment of the anode and cathode plates is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that shown here.

[0040] In this embodiment, a method for extracting the positive and negative poles of a battery cell group and detecting the alignment of the positive and negative pole pieces is provided, which can be used in electronic devices such as computers, mobile phones, tablet computers, etc. Figure 1Flow chart of a method for extracting cathode and anode points of a battery cell group according to an embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:

[0041] Step S101: Acquire an image of a battery cell electrode group. The battery cell electrode group is formed by stacking a plurality of cathode sheets and a plurality of anode sheets. Specifically, Figure 2 As shown, in the battery cell electrode group, the cathode is larger than the anode, the cathode wraps the anode, and is stacked in the order of a cathode sheet, an anode sheet, a cathode sheet, and an anode sheet. In addition, the poles on the pole sheets in the formed battery cell electrode group are all located at the edge of the battery cell electrode group, so when acquiring an imaging image of the battery cell electrode group, an imaging image of the edge of the battery cell electrode group or the four corners of the battery cell electrode group can be acquired.

[0042] For the imaging acquisition method of the battery cell pole group, X-rays can be used to irradiate the battery cell pole group, an image intensifier can be used to convert the X-rays into high-brightness visible light, and then a linear array camera can be used to collect the high-brightness visible light to form an imaging image of the battery cell pole group. In addition, the linear array camera can also be replaced with an image sensor to collect the imaging image, which is not limited in the embodiment of the present invention.

[0043] Step S102: Annotate the positive and negative poles of the electrode pieces in the electrode group of the battery cell according to the imaged image to generate an annotated image, wherein the annotated image includes the coordinates of the positive and negative poles. An annotation tool can be used to annotate the positive and negative poles of the electrode pieces in the image. In order to make the subsequent extraction of the positive and negative poles more accurate, the coordinates of the positive and negative poles in the image are annotated during annotation, thereby generating an annotated image and corresponding annotation text. The annotated image is as follows: Figure 3 and Figure 4 shown.

[0044] Step S103: Train the high-resolution model for key point detection according to the training set and verification set generated by the annotated image to generate a pole extraction model. Specifically, the annotated image and the corresponding annotated text formed by the annotation can be used as training samples of the model. At the same time, in order to verify the model later, part of the training sample is used as the training set and the other part is used as the verification set. Then, the model is trained using the training set, and the trained model is verified using the verification set.

[0045] Among them, in order to accurately extract the poles in the cathode and anode pieces, a key point detection high-resolution (HighResolution Net, HRNet) model is used, which can maintain high-resolution expression throughout the process. The model starts with a high-resolution sub-network as stage 1, gradually adds high-to-low resolution sub-networks one by one to form more stages, and then connects the multi-resolution sub-networks in parallel; throughout the process, information is exchanged over and over again in the parallel multi-resolution sub-networks to complete the repeated multi-scale fusion process; finally, the key points are estimated on the high-resolution feature map output by the network.

[0046] Step S104: extracting the positive and negative poles in the imaging image of the electrode group of the battery cell to be detected according to the pole extraction model. When performing pole extraction, the collected imaging image of the electrode group of the battery cell to be detected is input into the trained pole extraction model, and the detected key points, i.e., the positive and negative poles on the positive and negative electrode sheets, are output.

[0047] The method for extracting the positive and negative poles of a battery cell electrode group provided in an embodiment of the present invention uses a pole extraction model obtained by training a high-resolution model for key point detection to extract the positive and negative poles in the battery cell electrode group. Compared with the problem of inaccurate extraction of positive and negative poles by edge detection and other methods in the prior art, this method can accurately extract the positive and negative poles in the battery cell electrode group. At the same time, when annotating to form a training set and a validation set, the coordinates of the positive and negative poles in the image are annotated to further improve the accuracy of subsequent extraction of the positive and negative poles.

[0048] In one embodiment, after acquiring the imaging image of the battery cell pole group, it also includes: screening the pole piece imaging image in the imaging image; rotating the screened pole piece imaging image to obtain a pole piece image with a consistent imaging direction. After acquiring the image of the battery cell pole group using the X-ray + linear array camera + image intensifier solution, first screen all the acquired images to remove the images that do not contain pole pieces to ensure the validity of the sample. At the same time, the rotation matrix function cv2.getRotationMatrix2D can also be used to rotate part of the image so that the imaging direction of all the images finally obtained is consistent.

[0049] In one embodiment, the key point detection high-resolution model is trained based on the training set and the validation set generated by the annotated image to generate the pole extraction model, including: using data enhancement technology to process the annotated image to obtain an enhanced annotated image; and training the key point detection high-resolution model based on the training set and the validation set generated by the enhanced annotated image to generate the pole extraction model. Specifically, since the anode and cathode of the pole piece have different shapes, during the model training process, the annotated image is processed using data enhancement technology including image flipping, rotation, cropping, scaling, noise, mirroring, etc., thereby improving the robustness of the formed model.

[0050] In one embodiment, the positive and negative poles in the imaging image of the pole group of the battery cell to be detected are extracted according to the pole extraction model, including: intercepting the region of interest (ROI) image in the imaging image of the pole group of the battery cell to be detected, and converting the region of interest image into a three-channel image; performing an affine transformation on the converted image and converting it into a torch form; and inputting the converted image into the pole extraction model to extract the positive and negative poles.

[0051] Among them, when performing operations such as interception and conversion on the image, the open source framework Opencv can be used for implementation. The sensing area area image can be an area to be processed that is outlined by a box, circle, ellipse or irregular polygon. In this embodiment, the operator and function in Opencv are used to obtain the region of interest, and the region of interest can specifically be the electrode area in the battery cell electrode group. The three-channel image is an RGB three-channel image. The affine transformation process can be achieved by multiplying the affine transformation matrix and the original image to realize the image transformation process.

[0052] In one embodiment, the method for extracting the cathode and anode points of the battery cell group is implemented by the following process:

[0053] Use X-rays, linear array cameras, and image intensifiers to collect images of multiple battery cell pole groups; filter the collected images to obtain pole piece images; use cv2.getRotationMatrix2D to rotate the filtered pole piece images to ensure that the image imaging direction is consistent. Use the image annotation tool Labelme to annotate the positive and negative poles of the pole piece image, and annotate the positive and negative poles of the positive and negative poles to generate annotated images (jpg images) and annotation text (json text). Among them, the annotation text is as follows: Figure 5 As shown, the coordinates of the yin and yang poles are given.

[0054] The training samples are formed based on the annotated images and the annotated texts. One part of the training samples is used as the training set (train) and the other part is used as the validation set (val). Before forming the training samples, the annotated images are processed using data enhancement technology to expand the sample set. After obtaining the training set and the validation set, the key point detection HRNet model built based on the Pytorch framework is trained using the training set. The model parameters need to be set during the training process. The specific setting parameters include: the size of the model input resolution (image_size); the size of the output heatmap (heatmap_size); setting the number of output channels (num_output_channels); the number of key points (num_joints); the number of channels supported by the dataset (dataset_channel); the number of inference output channels (inference_channel); whether to perform adaptive non-maximum suppression threshold during inference (soft_nms); non-maximum suppression threshold (nms_thr); object key point similarity score threshold during non-maximum suppression threshold (oks_thr); key point visibility threshold (vis_thr); whether to use manually annotated bounding boxes during testing (use_gt_bbox); threshold of detected bounding box scores (det_bbox_thr); the path of the bounding box detection file to be used when use_gt_bbox=True (bbox_file).

[0055] In this embodiment, the specific model setting parameters are as follows:

[0056] data_cfg = dict(

[0057] image_size = [288, 384],

[0058] heatmap_size = [72, 96],

[0059] num_output_channels=channel_cfg['num_output_channels'],

[0060] num_joints=channel_cfg['dataset_joints'],

[0061] dataset_channel=channel_cfg['dataset_channel'],

[0062] inference_channel=channel_cfg['inference_channel'],

[0063] soft_nms=False,

[0064] nms_thr=1.0,

[0065] oks_thr=0.9,

[0066] vis_thr=0.2,

[0067] use_gt_bbox=True,

[0068] det_bbox_thr=0.0,

[0069] bbox_file='',)

[0070] After completing the model training and verification to obtain the pole detection model, the pole detection model is used to extract the positive and negative poles of the imaging image of the battery cell pole group to be detected. Before the imaging image to be detected is input into the model for extraction, Opencv is used to load the battery cell pole group image to be detected, and the ROI (region of interest) image is intercepted and converted into an RGB three-channel image. The ROI image is affine transformed to 384*288*3, and then converted to torch form (1, 3, 384, 288), and passed into the trained model.

[0071] The prediction results of the model are decoded to obtain the coordinates of 33 cathode points (green points above) and 32 anode points (green points below), and the corrected anode points (yellow points) are calculated. The cathode and anode points and the corrected anode points are shown in the figure below.

[0072] The embodiment of the present invention also provides a method for detecting the alignment of the positive and negative electrodes of a battery cell electrode group, such as Figure 6 As shown, the following steps are included:

[0073] Step S201: Correct the anode point extracted by the method for extracting cathode and cathode points of a battery cell electrode group described in any of the above embodiments to obtain a corrected anode point, wherein the corrected anode point has the same ordinate as the extracted cathode point. Since the battery cell electrode group is composed of anode and cathode sheets stacked in sequence, each cathode sheet and an adjacent anode sheet form a group starting from the starting position of the battery cell electrode group. When correcting the anode point, the ordinate of the cathode pole of the cathode sheet in the corresponding group is used as a reference, so that the ordinate of the corrected anode point is the same as the ordinate of the cathode sheet pole in the group.

[0074] Step S202: Perform alignment detection based on the distance between the corrected anode point and the extracted cathode point as the alignment detection condition. Since the battery cell group to be detected includes multiple anode sheets and cathode sheets, the extracted cathode points and the corrected anode points in the group can be connected to obtain multiple distances corresponding to the battery cell group to be detected. The maximum and minimum distances can be first screened out, and the two can be compared with the corresponding distance standard range values. Based on whether the two are within the range, it is determined whether the battery cell group to be detected is qualified. In addition, each calculated distance value can be compared with the distance standard range value. When any distance is not within the range, the product is unqualified.

[0075] The method for detecting the alignment of the positive and negative electrodes of a battery cell electrode group provided in an embodiment of the present invention first corrects the determined anode point so that it falls at the same position as the vertical coordinate of the corresponding cathode point, and then connects the extracted cathode point and the corrected anode point, and uses the obtained distance as the detection condition. Compared with the current method of directly connecting the anode point and the cathode point and detecting based on the distance between the two, the detection standard is changed, and the missed detection rate and over-kill rate of the product are reduced.

[0076] In one embodiment, if Figure 7 As shown, the extracted anode point is corrected to obtain the corrected anode point, including the following steps:

[0077] Step S301: binarize and detect the edge of the cathode sheet area image in the imaging image of the electrode group of the battery cell to be tested, and obtain the straight line where any cathode sheet is located. Specifically, binarize the cathode sheet area image, that is, set the gray value of the pixel on the image to 0 or 255, that is, the entire image presents an obvious visual effect of only black and white; then perform Canny Cauchy edge detection on the binarized image, and do edge, and grab the edge point of the straight line where the corresponding cathode sheet is located for the refined edge image.

[0078] Step S302: Calculate the slope of any cathode line based on interpolation and least squares method. For the edge points of the cathode line captured, interpolation method is used to calculate the corresponding sub-pixel points, and finally the slope k of the cathode line is calculated using least squares method.

[0079] Step S303: Correct the anode point according to the slope and the extracted cathode and cathode points to obtain the corrected anode point. The above steps can be used to extract the coordinates of the cathode point and the anode point, and the cathode point coordinates and the slope k are substituted into the linear function y=kx+b to obtain the b value, thereby determining the linear function. Then the ordinate of the extracted anode point is substituted into the linear function to obtain the abscissa of the corrected anode point. The corrected anode point can be obtained based on the ordinate of the extracted anode point and the abscissa of the corrected anode point. The extracted cathode point (diamond black dot) and the corrected anode point (square white dot) are shown in Figure 2. Figure 8 shown.

[0080] In one embodiment, the method for detecting the alignment of the cathode and cathode plates of the battery cell electrode group also includes: calculating the distance between the highest point among the extracted cathode points and the lowest point among the corrected anode points to obtain the maximum drop; calculating the maximum value of the distance between the multiple layers of corrected anode points and the maximum value of the distance between the multiple layers of extracted cathode points to obtain the local drop between the cathode and cathode; performing alignment detection based on the maximum drop and the local drop between the cathode and cathode as alignment detection conditions.

[0081] Specifically, when performing alignment detection, in addition to the distance between the extracted cathode point and the corrected anode point mentioned above, the maximum drop and local drop are also calculated as alignment detection conditions. The three are used together as screening conditions for screening, thereby further reducing the missed detection rate and over-kill rate.

[0082] In one embodiment, if Fig. 9 As shown, the maximum value of the distance between the anode points after multi-layer correction and the maximum value of the distance between the cathode points extracted by multi-layer are calculated to obtain the local drop between the cathode and the anode, including the following steps:

[0083] Step S401: Calculate the anode difference of each layer between the preset number of layers of anode sheets in the battery cell electrode group, and the preset number of layers on both sides of the battery cell electrode group is less than the preset number of layers in the middle of the battery cell electrode group. Specifically, in order to facilitate the calculation of the local drop, before the calculation, the anode sheets in the battery cell electrode group are numbered, starting from 0 to the last one. At the start and end positions of the battery cell electrode group, the preset number of layers selected can be smaller, and the preset number of layers selected in the middle position can be larger.

[0084] For example, there are 32 anode sheets in a battery cell electrode group, numbered 0 to 31. When you start selecting the preset number of layers, you can first select three layers, that is, first calculate the anode difference from layer 0 to layer 2, specifically calculate the anode difference of 3 layers between layer 0 and layer 1, layer 1 and layer 2, and layer 0 and layer 2. Then select four layers, that is, the anode difference from layer 0 to layer 3, specifically calculate the anode difference of 4 layers between layer 0 and layer 1, layer 1 and layer 2, layer 2 and layer 3, and layer 0 to layer 3. Then select five layers, that is, the anode difference from layer 0 to layer 4, specifically calculate the anode difference of 5 layers between layer 0 and layer 1, layer 1 and layer 2, layer 2 and layer 3, layer 3 and layer 4, and layer 0 to layer 4. After that, 5 layers are selected to calculate the anode difference, that is, the anode difference of 5 layers from the 1st layer to the 5th layer, the anode difference of 5 layers from the 2nd layer to the 6th layer, the anode difference of 5 layers from the 3rd layer to the 7th layer, ... the anode difference of 5 layers from the 26th layer to the 30th layer, and the anode difference of 5 layers from the 27th layer to the 31st layer. At this time, the end position of the battery cell electrode group has been calculated, and finally the anode difference of 4 layers from the 28th layer to the 31st layer and the anode difference of 3 layers from the 29th layer to the 31st layer are selected.

[0085] When calculating the anode difference, the corrected anode point coordinates corresponding to the selected layer are determined, and then the vertical coordinates of the anode points in the two layers are subtracted to obtain the corresponding anode difference distance. At the same time, since the anode difference is calculated each time the preset number of layers is selected, that is, multiple anode differences are calculated, multiple anode differences calculated when the preset number of layers is selected can be compared, and the maximum value thereof is obtained as the anode difference corresponding to the preset number of layers.

[0086] Step S402: Screening the maximum value between the anode differences as the local anode drop. Specifically, for each battery cell electrode group, multiple preset layers are selected for anode difference calculation, so the maximum value among the multiple anode differences corresponding to the multiple preset layers is screened as the local anode drop.

[0087] Step S403: Calculate the cathode difference between each layer on the left and right sides of each cathode sheet in the battery cell electrode group. Specifically, before calculating the cathode difference, the cathode sheets are numbered first, and then the preset number of layers on the left and right sides of each cathode sheet are selected in turn to calculate the cathode difference between each layer. The calculated cathode difference of the preset number of layers on the left is the left extreme difference, and the calculated cathode difference of the preset number of layers on the right is the right extreme difference.

[0088] For example, there are 32 cathode plates in a battery cell electrode group, numbered from 0 to 32. First, calculate the left extreme difference, starting from layer 0. There is no cathode plate on the left side of layer 0, and calculate the cathode difference of layer 0 and layer 0, which is 1 layer in total. The cathode difference is 0; there is one cathode plate on the left side of layer 1, and calculate the cathode difference of the cathode plate on the left side of layer 1, that is, the cathode difference of layer 0 and layer 0, which is 1 layer in total. The cathode difference is 0; calculate the cathode difference on the left side of layer 2, that is, the cathode difference of layer 0 and layer 1, specifically calculate the cathode difference of layer 0 and layer 1 and the cathode difference of layer 1 and layer 1; calculate the cathode difference on the left side of layer 3, that is, the cathode difference of layer 1 and layer 2; calculate the cathode difference on the left side of layer 4, that is, the cathode difference of layer 2 to layer 3, which is 2 layers in total. Range; calculate the cathode difference on the left side of the 5th layer, that is, the cathode difference of the 3rd to 4th layers, a total of 2 layers; calculate the cathode difference on the left side of the 6th layer, that is, the cathode difference of the 4th to 5th layers, a total of 2 layers; ... calculate the cathode difference on the left side of the 28th layer, that is, the cathode difference of the 26th to 27th layers, a total of 2 layers; calculate the cathode difference on the left side of the 29th layer, that is, the cathode difference of the 27th to 28th layers, a total of 2 layers; calculate the cathode difference on the left side of the 30th layer, that is, the cathode difference of the 28th to 29th layers, a total of 2 layers; calculate the cathode difference on the left side of the 31st layer, that is, the cathode difference of the 29th to 30th layers, a total of 2 layers; calculate the cathode difference on the left side of the 32nd layer, that is, the cathode difference of the 30th to 31st layers, a total of 2 layers.

[0089] Then calculate the right extreme difference: starting from layer 0, calculate the anode difference on the right side of layer 0, that is, the cathode difference of 3 layers from layer 0 to layer 2, specifically calculate the cathode difference between layer 0 and layer 1, the cathode difference between layer 1 and layer 2, and the cathode difference between layer 0 and layer 2; calculate the anode difference on the right side of layer 1, that is, the cathode difference of 3 layers from layer 1 to layer 3; calculate the anode difference on the right side of layer 2, that is, the cathode difference of 3 layers from layer 2 to layer 4; calculate the anode difference on the right side of layer 3, that is, the cathode difference of 3 layers from layer 3 to layer 5; calculate the anode difference on the right side of layer 4, that is, the cathode difference of 3 layers from layer 4 to layer 6; calculate the anode difference on the right side of layer 5 The extreme difference is the cathode difference of the 5th to 7th layers, a total of 3 layers; the anode difference on the right side of the 6th layer is calculated, that is, the cathode difference of the 6th to 8th layers, a total of 3 layers... The anode difference on the right side of the 28th layer is calculated, that is, the cathode difference of the 28th to 30th layers, a total of 3 layers; the anode difference on the right side of the 29th layer is calculated, that is, the cathode difference of the 29th to 31st layers, a total of 3 layers; the anode difference on the right side of the 30th layer is calculated, that is, the cathode difference of the 30th to 32nd layers, a total of 3 layers; the anode difference on the right side of the 31st layer is calculated, that is, the cathode difference of the 31st to 32nd layers, a total of 2 layers; the anode difference on the right side of the 32nd layer is calculated, that is, the cathode difference of 1 layer from the 32nd to 32nd layers.

[0090] Step S404: Screening the maximum value between the cathode differences as the cathode local drop. Specifically, compare the multiple left extreme differences and the multiple right extreme differences calculated above, and select the maximum value therebetween as the cathode local drop.

[0091] In one embodiment, when the alignment detection method for the positive and negative electrode sheets of a battery cell group is used to detect the alignment, when the scale of the collected imaging image is different, the alignment detection conditions set therein will also change. Fig.10 As shown in the figure, when the scale is 0.023mm / pixel, the four alignment detection conditions are set as follows: the distance between the extracted cathode point and the corrected anode point is [0.8mm~3.29mm], the maximum drop is 4.5mm, the local drop of the anode is 1mm, and the local drop of the cathode is 1mm. At this time, the maximum value of the distance between the extracted cathode point and the corrected anode point calculated by the above steps according to the imaging image is 2.692mm, and the minimum value is 1.358mm, both of which are within the detection conditions; the calculated maximum drop is 2.989mm, the calculated local drop of the anode is 0.276mm, and the calculated local drop of the cathode is 1.265mm. It can be seen that the local drop of the cathode is greater than the detection condition of 1mm, so the product corresponding to the imaging image is unqualified.

[0092] like Fig.11 As shown in the figure, when the scale is 0.014mm / pixel, the four alignment detection conditions are set as follows: the distance between the extracted cathode point and the corrected anode point is [0.8mm~2.97mm], the maximum drop is 4.5mm, the local drop of the anode is 1mm, and the local drop of the cathode is 1mm. At this time, the maximum value of the distance between the extracted cathode point and the corrected anode point calculated by the above steps according to the imaging image is 2.464mm, and the minimum value is 1.499mm, both of which are within the detection conditions; the calculated maximum drop is 2.674mm, the calculated local drop of the anode is 1.036mm, and the calculated local drop of the cathode is 0.448mm. It can be seen that the local drop of the anode is greater than the detection condition of 1mm, so the product corresponding to the imaging image is unqualified.

[0093] In one embodiment, in order to facilitate the application of the method for extracting the positive and negative poles of the battery cell electrode group and the method for detecting the alignment of the positive and negative poles of the battery cell electrode group on the device, the two methods can be deployed in C++. Specifically, the pole extraction model trained by Pytorch is converted into an ONNX (Open Neural Network Exchange) model. The microsoft / onnxruntime interface is called to pre-process, reason and post-process the ONNX model to obtain the coordinate positions of the positive and negative poles, and the three distances are calculated according to the methods of steps S201 to S202 and steps S301 to S303, and the calculation results are drawn on the image using Opencv. Through the above steps, unqualified images (NG images) and qualified images (OK images) can be screened out, and then the NG images and OK images are saved separately to the local (for data recall), and the result signal is transmitted to the receiving end through communication (such as the PLC (Programmable Logic Controller) / MES (manufacturing execution system) system in the device to directly intercept the NG pole piece).

[0094] The embodiment of the present invention also provides a device for extracting positive and negative poles of a battery cell electrode group, such as Fig.12 As shown, the device comprises:

[0095] The image acquisition module is used to acquire the imaging images of the four corners of the battery cell electrode group; the specific content refers to the corresponding part of the above method embodiment, which will not be repeated here.

[0096] The marking module is used to mark the positive and negative poles of the pole pieces in the battery cell pole group according to the imaging image to generate a marked image, wherein the marked image includes the coordinates of the positive and negative poles; the specific content refers to the corresponding part of the above method embodiment, which will not be repeated here.

[0097] The model generation module is used to train the high-resolution model for key point detection based on the training set and sample set generated by the labeled image to generate a pole extraction model; the specific content refers to the corresponding part of the above method embodiment, which will not be repeated here.

[0098] The extraction module is used to extract the positive and negative poles in the imaging image of the pole group of the battery cell to be detected according to the pole extraction model. For details, please refer to the corresponding part of the above method embodiment, which will not be repeated here.

[0099] The device for extracting the positive and negative poles of a battery cell electrode group provided by the embodiment of the present invention uses a pole extraction model obtained by training a high-resolution model for key point detection to extract the positive and negative poles of a battery cell electrode group. Compared with the problem of inaccurate extraction of positive and negative poles by edge detection and other methods in the prior art, the device can accurately extract the positive and negative poles of a battery cell electrode group. At the same time, when annotating to form a training set and a validation set, the coordinates of the positive and negative poles in the image are annotated to further improve the accuracy of subsequent extraction of the positive and negative poles.

[0100] For a detailed description of the function of the device for extracting the cathode and anode points of a battery cell electrode group provided in an embodiment of the present invention, please refer to the description of the method for extracting the cathode and anode points of a battery cell electrode group in the above embodiment.

[0101] The embodiment of the present invention also provides a device for detecting the alignment of the positive and negative electrodes of a battery cell electrode group, such as Fig.13 As shown, the device comprises:

[0102] A correction module is used to correct the anode point extracted by the method for extracting the cathode and cathode points of the battery cell group described in any of the above embodiments to obtain a corrected anode point, wherein the corrected anode point has the same vertical coordinate as the extracted cathode point; for specific contents, please refer to the corresponding part of the above method embodiment, which will not be repeated here.

[0103] The detection module is used to perform alignment detection according to the distance between the corrected anode point and the extracted cathode point as the alignment detection condition. For details, please refer to the corresponding part of the above method embodiment, which will not be repeated here.

[0104] The device for detecting the alignment of the cathode and anode pieces of a battery cell electrode group provided in an embodiment of the present invention first corrects the determined anode point so that it falls at the same position as the vertical coordinate of the corresponding cathode point, and then connects the extracted cathode point and the corrected anode point, and uses the obtained distance as the detection condition. Compared with the current method of directly connecting the anode point and the cathode point and detecting based on the distance between the two, the detection standard is changed, and the missed detection rate and over-kill rate of the product are reduced.

[0105] The functional description of the device for detecting the alignment of the cathode and cathode plates of a battery cell electrode group provided in the embodiment of the present invention is detailed in the description of the method for detecting the alignment of the cathode and cathode plates of a battery cell electrode group in the above embodiment.

[0106] The embodiment of the present invention also provides a storage medium, such as Fig.14As shown, a computer program 601 is stored thereon, and when the instruction is executed by the processor, the steps of the method for extracting the positive and negative poles of the battery cell group and detecting the alignment of the positive and negative pole pieces in the above-mentioned embodiment are implemented. The storage medium also stores audio and video stream data, feature frame data, interaction request signaling, encrypted data, and preset data size, etc. Among them, the storage medium can be a disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), a flash memory (Flash Memory), a hard disk (Hard Disk Drive, abbreviated: HDD) or a solid-state drive (SSD), etc.; the storage medium can also include a combination of the above-mentioned types of memory.

[0107] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium, and when the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the storage medium can be a disk, an optical disk, a read-only memory (ROM), a random access memory (RAM), a flash memory, a hard disk drive (HDD) or a solid-state drive (SSD), etc.; the storage medium can also include a combination of the above-mentioned types of memory.

[0108] The embodiment of the present invention further provides an electronic device, such as Fig.15 As shown, the electronic device may include a processor 51 and a memory 52, wherein the processor 51 and the memory 52 may be connected via a bus or other means. Fig.15 The example of connecting through bus is taken in the following.

[0109] The processor 51 may be a central processing unit (CPU). The processor 51 may also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, or a combination of the above chips.

[0110] The memory 52 is a non-transient computer-readable storage medium that can be used to store non-transient software programs, non-transient computer executable programs and modules, such as the corresponding program instructions / modules in the embodiment of the present invention. The processor 51 executes various functional applications and data processing of the processor by running the non-transient software programs, instructions and modules stored in the memory 52, that is, to implement the battery cell electrode group anode and cathode pole point extraction and anode and cathode pole piece alignment detection method in the above method embodiment.

[0111] The memory 52 may include a program storage area and a data storage area, wherein the program storage area may store an application required for operating the device and at least one function; the data storage area may store data created by the processor 51, etc. In addition, the memory 52 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In some embodiments, the memory 52 may optionally include a memory remotely arranged relative to the processor 51, and these remote memories may be connected to the processor 51 via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0112] The one or more modules are stored in the memory 52 and when executed by the processor 51, perform the following steps: Figure 1 -11 shows a method for extracting the positive and negative poles of the battery cell electrode group and detecting the alignment of the positive and negative pole pieces.

[0113] For details of the above electronic equipment, please refer to Figures 1 to 11 The corresponding related descriptions and effects in the illustrated embodiments can be understood and will not be repeated here.

[0114] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for detecting the alignment of cathode and anode plates of a battery cell electrode group, characterized in that: include: Acquire an imaging image of a battery cell electrode group; Marking the positive and negative poles of the pole pieces in the battery cell pole group according to the imaging image to generate a marked image, wherein the marked image includes the coordinates of the positive and negative poles; Training a high-resolution model for key point detection based on a training set and a validation set generated from the labeled images to generate a pole extraction model; Extracting positive and negative poles in the imaging image of the pole group of the battery cell to be tested according to the pole extraction model; Correcting the extracted anode point to obtain a corrected anode point, the corrected anode point is determined in the following manner: substituting the ordinate of the extracted anode point into the straight line where the cathode point is located to obtain the abscissa of the corrected anode point, and obtaining the corrected anode point based on the ordinate of the extracted anode point and the abscissa of the corrected anode point; The alignment detection is performed based on the distance between the corrected anode point and the extracted cathode point as the alignment detection condition.

2. The method according to claim 1, characterized in that After obtaining the imaging image of the battery cell group, it also includes: screening the polar piece imaging image in the imaging image; The screened pole piece imaging images are rotated to obtain pole piece images with consistent imaging directions.

3. The method according to claim 1, characterized in that The key point detection high-resolution model is trained according to the training set and the verification set generated by the labeled image to generate a pole extraction model, including: Processing the labeled image using data enhancement technology to obtain an enhanced labeled image; The key point detection high-resolution model is trained based on the training set and validation set generated by the enhanced labeled images to generate a pole extraction model.

4. The method according to claim 1, characterized in that: Extracting positive and negative poles in the pole group imaging image of the battery cell to be tested according to the pole extraction model includes: Intercepting an image of a region of interest in the imaging image of the electrode group of the battery cell to be detected, and converting the image of the region of interest into a three-channel image; Perform affine transformation and torch form conversion on the converted image; The converted image is input into the pole extraction model to extract the positive and negative poles.

5. The method according to claim 1, characterized in that Correct the extracted anode points to obtain the corrected anode points, including: Perform image binarization and edge detection on the cathode sheet area image in the electrode group imaging image of the battery cell to be detected, and obtain the straight line where any cathode sheet is located; The slope of the straight line where any cathode piece is located is calculated based on the interpolation method and the least square method; The anode point is corrected according to the slope and the extracted cathode and anode points to obtain a corrected anode point.

6. The method according to claim 1, characterized in that Also includes: Calculate the distance between the highest point among the extracted cathode points and the lowest point among the corrected anode points to obtain the maximum drop; Calculate the maximum value of the distance between the anode points after multi-layer correction and the maximum value of the distance between the cathode points extracted from the multi-layer to obtain the local drop between the cathode and the anode; The alignment detection is performed based on the maximum drop and the local drop of the anode and cathode as alignment detection conditions.

7. The method according to claim 6, characterized in that Calculate the maximum value of the distance between the anode points after multi-layer correction and the maximum value of the distance between the cathode points extracted from the multi-layer to obtain the local drop between the cathode and the anode, including: Calculating the anode difference of each layer between the preset number of layers of anode sheets in the battery cell electrode group, the preset number of layers on both sides of the battery cell electrode group is less than the preset number of layers in the middle of the battery cell electrode group; The maximum value between the anode differences is selected as the anode local drop; Calculate the cathode difference between each layer on the left and right sides of each cathode sheet in the battery cell electrode group; The maximum value between the cathode differences is screened as the cathode local drop.

8. A device for detecting the alignment of positive and negative electrodes of a battery cell electrode group, characterized in that: include: An image acquisition module, used to acquire an imaging image of the battery cell electrode group; A marking module, used for marking the positive and negative poles of the pole pieces in the battery cell pole group according to the imaging image to generate a marked image, wherein the marked image includes the coordinates of the positive and negative poles; A model generation module, used to train a high-resolution model for key point detection based on a training set and a verification set generated from the annotated image to generate a pole extraction model; An extraction module, used for extracting positive and negative poles in the imaging image of the pole group of the battery cell to be tested according to the pole extraction model; A correction module is used to correct the anode point extracted by the extraction method to obtain a corrected anode point. The corrected anode point is determined in the following manner: the ordinate of the extracted anode point is brought into the straight line where the cathode point is located to obtain the abscissa of the corrected anode point, and the corrected anode point is obtained based on the ordinate of the extracted anode point and the abscissa of the corrected anode point; The detection module is used to perform alignment detection based on the distance between the corrected anode point and the extracted cathode point as an alignment detection condition.

Images