Battery testing method, training method, production method, apparatuses, devices and medium

Abstract

A battery testing method, comprising: acquiring a testing image of a battery to be tested (S10), wherein the testing image comprises a plurality of first electrode sheets (100) and a plurality of second electrode sheets (200) of said battery that are alternately arranged in a first direction, and the first electrode sheets (100) and the second electrode sheets (200) have different polarities; on the basis of the testing image, determining first endpoints (A1, A2 and A3) of the first electrode sheets (100) and second endpoints (B1, B2 and B3) of the second electrode sheets (200) (S11), wherein the first endpoints (A1, A2 and A3) are points of the first electrode sheets (100) that are closest to a first boundary of the testing image, the second endpoints (B1, B2 and B3) are points of the second electrode sheets (200) that are closest to the first boundary of the testing image, and in the testing image, the first boundary extends in the first direction and does not intersect with either the first electrode sheets (100) or the second electrode sheets (200); and on the basis of the first endpoints (A1, A2 and A3) and the second endpoints (B1, B2 and B3) which are adjacent in the first direction, determining overhang amounts of the electrode sheets of said battery (S12). Further disclosed are a battery testing apparatus, a segmentation model training method for battery testing, a segmentation model training apparatus for battery testing, a battery production method, an electronic device, a battery production device and a computer-readable storage medium.

Description

Battery testing methods, training methods, production methods, devices, equipment, and media

[0001] Cross-references

[0002] This application refers to Chinese patent application No. 202310971825.7 filed on August 3, 2023, entitled “Battery detection method, training method, production method, device, equipment and medium”, which is incorporated into this application in its entirety by reference. Technical Field

[0003] The present application relates to the field of battery technology, and in particular to a battery detection method, training method, production method, device, equipment and medium. Background Art

[0004] Energy conservation and emission reduction are key to the sustainable development of the automotive industry. Electric vehicles, due to their energy-saving and environmentally friendly advantages, have become an important component of the sustainable development of the automotive industry. For electric vehicles, battery technology is a key factor in their development.

[0005] The battery consists of a stacked cathode and anode plates, separated by a separator to insulate the cathode and anode plates. The anode plate extends beyond the cathode plate in both width and length. The excess of the anode plate beyond the cathode plate (excess) can affect battery performance and requires testing to ensure that qualified batteries have an excess within an appropriate range.

[0006] In the related art, during the detection of excess electrode amounts of batteries, it is necessary to extract the contours of the electrode pieces, which is labor-intensive and inefficient.

[0007] Summary of the Invention

[0008] The present application aims to solve at least one of the technical problems in the background art. To this end, one object of the present application is to provide a battery testing method, training method, production method, device, equipment and medium to improve testing efficiency.

[0009] An embodiment of the first aspect of the present application provides a battery detection method, the method comprising: obtaining a detection image of a battery to be tested, the detection image comprising a plurality of first pole pieces and a plurality of second pole pieces alternately arranged along a first direction of the battery to be tested, wherein the polarity of the first pole piece and the second pole piece are different; determining a first endpoint of the first pole piece along a second direction and a second endpoint of the second pole piece along the second direction based on the detection image, wherein the first endpoint is a point at which the first pole piece is closest to a first boundary of the detection image, and the second endpoint is a point at which the second pole piece is closest to the first boundary of the detection image, and in the detection image, the first boundary extends along the first direction, and the first boundary does not intersect with the first pole piece and the second pole piece; determining an excess amount of the pole piece of the battery to be tested based on the first endpoint and the second endpoint adjacent to each other along the first direction.

[0010] In the technical solution of the embodiment of the present application, since the pole piece excess is the length of the portion of the first pole piece that exceeds the second pole piece, in the detection image, it is also the distance along the second direction between the first endpoint and the second endpoint. Therefore, after identifying the first endpoint of the first pole piece along the second direction and the second endpoint of the second pole piece along the second direction in the detection image, the pole piece excess of the battery to be tested can be directly determined based on the distance along the second direction between the adjacent first endpoint and the second endpoint. In the detection image, only the first endpoint of the first pole piece and the second endpoint of the second pole piece need to be determined. Compared with the related art that needs to identify the outline of the first pole piece and the outline of the second pole piece, the workload is reduced, the detection speed can be increased, and thus the detection efficiency is improved.

[0011] In some embodiments, determining the first endpoint of the first pole piece and the second endpoint of the second pole piece based on the detection image includes: inputting the detection image into a segmentation model, obtaining a pole piece mask diagram output by the segmentation model, the pole piece mask diagram including a background area and multiple end mask areas, each end mask area corresponding to the area where the end of the first pole piece is located or the area where the end of the second pole piece is located, the end of the first pole piece is the end of the first pole piece close to the first boundary, and the end of the second pole piece is the end of the second pole piece close to the first boundary; determining the first endpoint of the first pole piece and the second endpoint of the second pole piece based on the pole piece mask diagram. The pole piece mask diagram output by the trained segmentation model only needs to mark the head of the pole piece, reducing workload and improving efficiency.

[0012] In some embodiments, the plurality of end mask regions include a plurality of first end mask regions and a plurality of second end mask regions, wherein the first end mask region corresponds to the region where the end of the first pole piece is located, and the second end mask region corresponds to the region where the end of the second pole piece is located; the first endpoint is located at the center of the first end mask region; and the second endpoint is located at the center of the second end mask region. By finding the center in the independent end mask region on the pole piece mask diagram, the pole piece endpoint in the detection image can be better located.

[0013] In some embodiments, determining the amount of overhang of a pole piece of a battery under test based on first and second endpoints adjacent to each other along a first direction includes: determining a first distance between the first and second endpoints along a second direction, the second direction being perpendicular to the first direction, based on the first and second endpoints adjacent to each other along the first direction; and determining the amount of overhang of the pole piece of the battery under test based on the first distance. Characterizing the amount of overhang of a pole piece of a battery under test based on the distance between the first and second endpoints requires only determining the first endpoint of the first pole piece and the second endpoint of the second pole piece, which reduces workload, increases detection speed, and thus improves detection efficiency.

[0014] In some embodiments, obtaining a test image of a battery under test includes: obtaining an initial image of the battery under test; performing gradient compression processing on the initial image to obtain a compressed gradient image of the battery under test; performing image reconstruction based on the compressed gradient image to obtain a reconstructed image of the battery under test; and performing image enhancement processing on the reconstructed image to obtain a test image of the battery under test. Processing the initial image in this manner to obtain a test image can make the electrode features in the test image clearer, facilitating subsequent testing.

[0015] In some embodiments, image enhancement processing is performed on the reconstructed image to obtain a test image of the battery under test, including: contrast enhancement processing is performed on the reconstructed image to obtain a first preprocessed image of the battery under test; histogram equalization is performed on the reconstructed image to obtain a second preprocessed image of the battery under test; and the test image of the battery under test is obtained based on the first preprocessed image and the second preprocessed image. The two different image enhancement processes can distinguish different features of the electrode piece, so that the test image of the battery under test obtained based on the first preprocessed image and the second preprocessed image can highlight different features of the electrode piece.

[0016] In some embodiments, obtaining a test image of the battery under test based on the first preprocessed image and the second preprocessed image includes: performing image fusion processing on the first preprocessed image and the second preprocessed image to obtain the test image of the battery under test. The image fusion processing allows the information in the first preprocessed image and the second preprocessed image to be input into the test image, increasing the data dimension, so that the test image can include more information and improve subsequent test accuracy.

[0017] In some embodiments, obtaining a test image of the battery to be tested based on the first preprocessed image and the second preprocessed image includes: performing image fusion processing on the first preprocessed image and the second preprocessed image to obtain a third preprocessed image of the battery to be tested, the third preprocessed image including a plurality of first pole pieces and a plurality of second pole pieces of the battery to be tested, which are alternately arranged along a first direction, wherein the first pole pieces and the second pole pieces have different polarities; determining a first pole piece fitting line and a second pole piece fitting line based on the third preprocessed image to obtain an auxiliary test image of the battery to be tested, wherein the first pole piece fitting line is a fitting center line of the first pole piece, and the second pole piece fitting line is a fitting center line of the second pole piece; performing image fusion processing on the first preprocessed image, the second preprocessed image, and the auxiliary test image to obtain a test image of the battery to be tested. Compared with the test image obtained by fusing two images, the test image obtained by fusing three images contains more information and can improve subsequent test accuracy.

[0018] In some embodiments, the method further includes: determining the electrode gap of the battery to be tested based on the distance along the first direction between the adjacent first electrode fitting line and the second electrode fitting line in the auxiliary detection image. In the related art, the electrode excess and the electrode gap are detected separately through two images, and the two images need to be processed separately, resulting in low detection efficiency. The method provided in the embodiment of the present application can be used to detect both the electrode excess and the electrode gap, that is, two feature detections of the battery to be tested can be achieved through one image, thereby improving information utilization.

[0019] In some embodiments, before performing gradient compression on the initial image to obtain a compressed gradient image of the battery under test, the method further includes performing noise reduction on the initial image. By performing noise reduction on the initial image, random noise such as quantum noise, aliasing noise, and electronic noise can be removed from the initial image, thereby significantly improving the signal-to-noise ratio of the initial image.

[0020] In some embodiments, gradient compression processing is performed on the initial image to obtain a compressed gradient image of the battery to be tested, including: extracting a gradient image of the initial image; performing gradient compression processing on the gradient image to obtain a compressed gradient image of the battery to be tested. By performing gradient compression processing on the gradient image to obtain a compressed gradient image of the battery to be tested, the gradient value in the compressed gradient image is smaller than the gradient value in the gradient image, and the change in the grayscale value in the reconstructed image obtained by compressing the gradient image is reduced. For the reconstructed image, even if the grayscale values ​​between adjacent pixels are not much different, they can be more easily identified from the reconstructed image, making the detailed features in the reconstructed image clearer and improving the accuracy of the image recognition results. Using this method in processing battery detection images can improve the accuracy of battery detection results.

[0021] An embodiment of the second aspect of the present application provides a battery detection device, which includes: an acquisition module configured to acquire a detection image of a battery to be tested, the detection image including a plurality of first pole pieces and a plurality of second pole pieces alternately arranged along a first direction of the battery to be tested, wherein the polarity of the first pole piece and the second pole piece are different; a first determination module configured to determine a first endpoint of the first pole piece along the second direction and a second endpoint of the second pole piece along the second direction based on the detection image, wherein the first endpoint is the point at which the first pole piece is closest to a first boundary of the detection image, and the second endpoint is the point at which the second pole piece is closest to the first boundary of the detection image, and in the detection image, the first boundary extends along the first direction and does not intersect with the first pole piece and the second pole piece; a second determination module configured to determine the pole piece excess amount of the battery to be tested based on the first endpoint and the second endpoint adjacent to each other along the first direction.

[0022] The embodiment of the third aspect of the present application provides a segmentation model training method for battery detection, the training method comprising: obtaining a sample image containing a sample battery, the sample image comprising a plurality of first sample pole pieces and a plurality of second sample pole pieces of the sample battery arranged alternately along a first direction; obtaining a real pole piece mask image corresponding to the sample image, the real pole piece mask image comprising a real background area and a plurality of real end mask areas, the plurality of real end mask areas comprising a first real end mask area corresponding to the area where the end of the first sample pole piece is located and / or a second real end mask area corresponding to the area where the end of the second sample pole piece is located; inputting the sample image into the segmentation model to be trained, obtaining a predicted pole piece mask image output by the segmentation model; calculating a loss value based on the real pole piece mask image and the predicted pole piece mask image; and adjusting the parameters of the segmentation model based on the loss value. The pole piece mask image output by the trained segmentation model can only mark the head of the pole piece in the detection image and output a complete pole piece mask area. Using the real mask area as the true value of the training segmentation model can improve the accuracy of the segmentation model prediction, restore the characteristics of the pole piece to a certain extent, and facilitate subsequent measurement.

[0023] In some embodiments, obtaining a sample image containing a sample battery includes: obtaining a sample initial image of the sample battery; performing gradient compression processing on the sample initial image to obtain a sample compressed gradient image of the sample battery; performing image reconstruction based on the sample compressed gradient image to obtain a sample reconstructed image of the sample battery; and performing image enhancement processing on the sample reconstructed image to obtain a sample image of the sample battery. The method for obtaining the sample image of the sample battery is the same as that for obtaining the test image of the battery to be tested, which makes the subsequent training model more accurate in processing the test image.

[0024] In some embodiments, image enhancement processing is performed on the sample reconstructed image to obtain a sample image of the sample battery, including: contrast enhancement processing is performed on the sample reconstructed image to obtain a first sample preprocessed image of the sample battery; histogram equalization is performed on the sample reconstructed image to obtain a second sample preprocessed image of the sample battery; and image fusion processing is performed on the first sample preprocessed image and the second sample preprocessed image to obtain a sample image of the sample battery. After the two preprocessed images are fused into a sample image and annotated to produce a training data set, the sample detection image used for training and the detection image used for detection are obtained by the same type of side processing method, which can make the trained segmentation model process the detection image more accurately.

[0025] In some embodiments, obtaining a true pole piece mask corresponding to a sample image includes: marking a first annotation point corresponding to an end of a first sample pole piece based on the sample image, and marking a second annotation point corresponding to an end of a second sample pole piece; obtaining a first true end mask region corresponding to the end of the first sample pole piece based on the marked first annotation point and a preset first expansion value; and obtaining a second true end mask region corresponding to the end of the second sample pole piece based on the marked second annotation point and a preset second expansion value. Marking only two annotation points for a sample pole piece can reduce the difficulty of annotation, reduce the annotation workload, and improve annotation efficiency.

[0026] An embodiment of the fourth aspect of the present application provides a segmentation model training device for battery detection, the training device comprising: a sample acquisition module configured to acquire a sample image containing a sample battery, the sample image comprising a plurality of first sample pole pieces and a plurality of second sample pole pieces of the sample battery arranged alternately along a first direction; a truth value module configured to acquire a real pole piece mask image corresponding to the sample image, the real pole piece mask image comprising a real background area and a plurality of real end mask areas, the plurality of real end mask areas comprising a first real end mask area corresponding to an area where an end of the first sample pole piece is located and / or a second real end mask area corresponding to an area where an end of the second sample pole piece is located; a sample prediction module configured to input the sample image into a segmentation model to be trained and acquire a predicted pole piece mask image output by the segmentation model; a loss module configured to calculate a loss value based on the real pole piece mask image and the predicted pole piece mask image; and a parameter adjustment module configured to adjust the parameters of the segmentation model based on the loss value.

[0027] In a fifth aspect, an embodiment of the present application provides a battery production method, comprising: using any of the battery testing methods described in the above embodiments to detect excess electrode plate volume in a battery to be tested, thereby determining a quality test result for the battery to be tested. By testing the battery using the above battery testing method, the battery can be determined to be qualified based on the excess electrode plate volume, thereby determining a quality test result for the battery to be tested.

[0028] In some embodiments, the method further includes: in response to the quality inspection result of the battery to be tested being unqualified, rejecting the unqualified battery. Rejecting the unqualified battery can reduce the possibility of safety accidents caused by the unqualified battery.

[0029] An embodiment of the sixth aspect of the present application provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the battery detection method of any one of the above embodiments or the training method of any one of the above embodiments or the battery production method of the above embodiments.

[0030] An embodiment of the seventh aspect of the present application provides a battery production device, including the electronic device in the above embodiment.

[0031] An embodiment of the eighth aspect of the present application provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the battery detection method of any one of the above embodiments, the training method of any one of the above embodiments, or the battery production method of any one of the above embodiments.

[0032] An embodiment of the ninth aspect of the present application provides a computer program product, comprising a computer program, which, when executed by a processor, implements the battery detection method of any one of the above embodiments, the training method of any one of the above embodiments, or the battery production method of the above embodiments.

[0033] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the multiple drawings represent the same or similar parts or elements. These drawings are not necessarily drawn to scale. It should be understood that these drawings only depict some embodiments disclosed in the present application and should not be regarded as limiting the scope of the present application. In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on the drawings without paying creative work.

[0035] FIG1 is a flow chart of a battery detection method provided in an embodiment of the present application;

[0036] FIG2 is a detection image of a battery to be tested provided in an embodiment of the present application;

[0037] FIG3 is a schematic structural diagram of a wound battery provided in an embodiment of the present application;

[0038] FIG4 is a schematic structural diagram of a laminated battery provided in an embodiment of the present application;

[0039] FIG5 is a block diagram of a battery detection device provided in an embodiment of the present application;

[0040] FIG6 is a flowchart of a segmentation model training method for battery detection provided in an embodiment of the present application;

[0041] FIG7 is a block diagram of a segmentation model training device for battery detection provided in an embodiment of the present application. DETAILED DESCRIPTION

[0042] The following embodiments of the technical solution of the present application will be described in detail with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present application and are therefore only examples and are not intended to limit the scope of protection of the present application.

[0043] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned figure descriptions are intended to cover non-exclusive inclusions.

[0044] In the description of the embodiments of this application, the technical terms "first" and "second" are used only to distinguish different objects and should not be understood to indicate or imply relative importance or implicitly specify the quantity, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, the meaning of "plurality" is more than two, unless otherwise clearly and specifically defined.

[0045] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0046] In the description of the embodiments of this application, the term "and / or" is simply a description of the association relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent the following three situations: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this document generally indicates that the associated objects are in an "or" relationship.

[0047] In the description of the embodiments of the present application, the term "plurality" refers to more than two (including two). Similarly,

[0048] "Multiple groups" refers to two or more groups (including two groups), and "multiple tablets" refers to two or more tablets (including two tablets).

[0049] In the description of the embodiments of the present application, the technical terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing the embodiments of the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the embodiments of the present application.

[0050] In the description of the embodiments of the present application, unless otherwise expressly specified or limited, technical terms such as "installed," "connected," "connected," and "fixed" should be understood in a broad sense. For example, they can refer to fixed connections, detachable connections, or integration; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; internal connections between two components or interactions between two components. Those skilled in the art can understand the specific meanings of the above terms in the embodiments of the present application based on specific circumstances.

[0051] Currently, market developments indicate that power batteries are becoming increasingly widely used. They are not only used in energy storage systems such as hydropower, thermal, wind, and solar power plants, but are also widely used in electric vehicles like electric bicycles, electric motorcycles, and electric vehicles, as well as in military equipment and aerospace. As power battery applications continue to expand, market demand is also growing.

[0052] The battery includes a stacked cathode electrode and an anode electrode, which are separated by a diaphragm to insulate the anode electrode from the cathode electrode. The anode electrode exceeds the cathode electrode in both its width and length directions, and the electrode overhang (OH) will affect the performance of the battery. Taking lithium batteries as an example, during the charge and discharge process of the battery, lithium ions will repeatedly deintercalate between the cathode and cathode electrodes. However, when lithium ions are released from the cathode but there are not enough anode channels to embed, it is easy to cause lithium ion accumulation and lithium precipitation, causing certain product safety risks. Therefore, it is necessary to ensure that the cathode electrode is within the coating range of the anode as a whole, that is, to ensure that the anode exceeds the cathode, that is, the electrode overhang is greater than 0. However, the greater the electrode overhang, the lower the battery energy density. Therefore, it is necessary to detect the electrode overhang so that the battery electrode overhang is within an appropriate range.

[0053] In the related art, a ray collection device is used to irradiate the battery to obtain a detection image of the battery, and then the outline of the pole piece is identified in the detection image. The pole piece excess amount is determined based on the outline of the pole piece. However, identifying the outline of the pole piece is labor-intensive and inefficient.

[0054] An embodiment of the present application provides a battery detection method, the method comprising: obtaining a detection image of a battery to be tested, the detection image comprising a plurality of first pole pieces and a plurality of second pole pieces alternately arranged along a first direction of the battery to be tested, wherein the polarities of the first pole piece and the second pole piece are different; determining a first endpoint of the first pole piece and a second endpoint of the second pole piece based on the detection image, wherein the first endpoint is the point at which the first pole piece is closest to a first boundary of the detection image, and the second endpoint is the point at which the second pole piece is closest to the first boundary of the detection image, and in the detection image, the first boundary extends along the first direction, and the first boundary does not intersect with the first pole piece and the second pole piece; determining the amount of excess of the pole piece of the battery to be tested based on the first endpoint and the second endpoint adjacent to each other along the first direction. The amount of excess of the pole piece of the battery to be tested can be determined by only identifying the endpoints of the pole piece in the detection image, without identifying the outline of the entire pole piece, thus reducing the workload of identification, speeding up the detection speed, and improving the detection efficiency.

[0055] The battery detection method disclosed in the embodiments of the present application can be used to detect batteries that have just left the factory, and can also be used to detect batteries that are in service.

[0056] The present invention provides a battery detection method. FIG1 is a flow chart of a battery detection method provided by the present invention. Referring to FIG1 , the method includes:

[0057] Step S10: Acquire a detection image of the battery to be tested.

[0058] The detection image includes a plurality of first pole pieces and a plurality of second pole pieces alternately arranged along a first direction of the battery to be tested, wherein the polarities of the first pole pieces and the second pole pieces are different.

[0059] In an embodiment of the present application, the battery to be tested may be a wound battery, or a laminated battery.

[0060] In an embodiment of the present application, the detection image is an image obtained by capturing the battery to be tested through an image capture device. Exemplarily, the detection image may be a processed image.

[0061] Exemplarily, the image acquisition device may be a radiographic image acquisition device, a CT scanner, or a PET scanner, and the detection image may be a detection image acquired by the radiographic image acquisition device. Exemplarily, the radiographic image acquisition device may include a radiation source and a flat-panel detector, wherein the detection image may be acquired by the radiation source and the flat-panel detector. Radiation emitted by the radiation source is projected onto the flat-panel detector through the battery under test, and the flat-panel detector acquires the detection image based on the received radiation. The radiation emitted by the radiation source may be X-rays, gamma rays, beta rays, or the like.

[0062] In an embodiment of the present application, the first pole piece is an anode pole piece, and the second pole piece is a cathode pole piece; or, the first pole piece is a cathode pole piece, and the second pole piece is an anode pole piece.

[0063] In an embodiment of the present application, the cathode and anode plates of the battery are made of different materials, and different materials absorb radiation to different degrees, forming a light and dark contrast pattern on the detection image, thereby being able to identify the first and second plates in the detection image.

[0064] Step S20: determining a first endpoint of the first pole piece and a second endpoint of the second pole piece based on the detection image.

[0065] Among them, the first endpoint is the point where the first pole piece is closest to the first boundary of the detection image, and the second endpoint is the point where the second pole piece is closest to the first boundary of the detection image. In the detection image, the first boundary extends along the first direction, and the first boundary does not intersect with the first pole piece and the second pole piece.

[0066] For example, the first endpoint and the second endpoint can be obtained by algorithmic prediction or by manual annotation. When the first endpoint and the second endpoint are obtained by algorithmic prediction, the first endpoint corresponding to the first electrode or the second endpoint corresponding to the second electrode can be obtained by the electrode prediction result of the trained model. When the first endpoint and the second endpoint are obtained by manual annotation, the annotator can obtain the first endpoint corresponding to the first electrode or the second endpoint corresponding to the second electrode by punctuating on the detection image.

[0067] In an embodiment of the present application, the first endpoint corresponding to each first pole piece in the detection image and the second endpoint corresponding to each second pole piece can be determined. Alternatively, only the first endpoint corresponding to some of the first pole pieces in the detection image and the second endpoint corresponding to some of the second pole pieces can be determined. For example, the edge area of ​​the detection image may be distorted due to shooting limitations, so the first endpoint and the second endpoint can be determined by selecting the first pole piece and the second pole piece in the partial area (non-distorted area).

[0068] Step S30: determining the electrode sheet overhang of the battery to be tested based on the first endpoint and the second endpoint adjacent to each other along the first direction.

[0069] The pole piece overhang is the length of the portion of the first pole piece that overhangs the second pole piece, or the length of the portion of the second pole piece that overhangs the first pole piece.

[0070] FIG2 is a test image of a battery under test provided in an embodiment of the present application. Referring to FIG2 , the first electrode piece 100 and the second electrode piece 200 in the test image are brighter than the background, allowing the first electrode piece 100 and the second electrode piece 200 to be observed in the test image. In FIG2 , the first electrode piece 100 is an anode electrode piece, and the second electrode piece 200 is a cathode electrode piece. In other embodiments, the first electrode piece 100 may be a cathode electrode piece, and the second electrode piece 200 may be an anode electrode piece.

[0071] In some embodiments of the present application, the first endpoint of the first electrode piece along the second direction Y can be understood as the point in the battery under test where the first electrode piece is closest to the outer casing along the second direction Y, i.e., point A in FIG2 . The second endpoint of the second electrode piece along the second direction Y can be understood as the point in the battery under test where the second electrode piece is closest to the outer casing along the second direction Y, and the first endpoint and the second endpoint are located on the same side of the battery under test, i.e., point B in FIG2 . In other words, the first endpoint of the first electrode piece is a boundary point of the first electrode piece, and the second endpoint of the second electrode piece is a boundary point of the second electrode piece.

[0072] In an embodiment of the present application, a plurality of first pole pieces and a plurality of second pole pieces are alternately arranged along a first direction X. Taking any first pole piece except the end portion as an example, the first pole piece is adjacent to two second pole pieces, and the positions of the second end points of the two second pole pieces may deviate in the second direction. Therefore, for the first pole piece, the excess amount of the first pole piece except the two pole pieces can be determined, and the battery under test is subsequently tested based on the excess amount of all pole pieces. For example, the first direction X is perpendicular to the second direction Y.

[0073] Obtaining the first endpoint of the first pole piece based on the detection image can be understood as obtaining the feature point corresponding to the first pole piece on the detection image. This feature point can be a feature point that characterizes the end of the first pole piece. It can be understood that these feature points do not actually represent the actual starting point of the first pole piece, but are used to characterize the starting point of the first pole piece that can be observed on the detection image (that is, the head feature point).

[0074] Obtaining the second endpoint of the second pole piece based on the detection image can be understood as obtaining the feature point corresponding to the second pole piece on the detection image. This feature point can be a feature point that characterizes the end of the second pole piece. It can be understood that these feature points do not actually represent the actual starting point of the second pole piece, but are used to characterize the starting point of the second pole piece that can be observed on the detection image (that is, the head feature point).

[0075] The first endpoint and the second endpoint can be obtained by manual punctuation or by machine punctuation. For example, the Point type in labelme is used for annotation.

[0076] In an embodiment of the present application, the first endpoint and the second endpoint should be selected based on the same marked reference so that the distance along the second direction between the first endpoint and the second endpoint adjacent to each other along the first direction is equal to the excess amount of the electrode of the battery to be tested in the detection image.

[0077] 2 , for the three first pole pieces 100 and the three second pole pieces 200 in the dotted box, the first endpoints of the three first pole pieces 100 are A1, A2, and A3, respectively, and the second endpoints of the three second pole pieces 200 are B1, B2, and B3, respectively. In the dotted box, the pole piece overhang includes: the distance between A1 and B1 along the second direction, the distance between B1 and A2 along the second direction, the distance between A2 and B2 along the second direction, the distance between B3 and A3 along the second direction, and the distance between A3 and B3 along the second direction.

[0078] In the embodiment of the present application, since the pole piece excess is the length of the portion of the first pole piece that exceeds the second pole piece, in the detection image, it is also the distance along the second direction between the first endpoint and the second endpoint. Therefore, after identifying the first endpoint of the first pole piece along the second direction and the second endpoint of the second pole piece along the second direction in the detection image, the pole piece excess of the battery to be tested can be directly determined based on the distance along the second direction between the adjacent first endpoint and the second endpoint. In the detection image, only the first endpoint of the first pole piece and the second endpoint of the second pole piece need to be determined. Compared with the related art that needs to identify the outline of the first pole piece and the outline of the second pole piece, the workload is reduced, the detection speed can be increased, and thus the detection efficiency is improved.

[0079] Referring to Figure 2, the detection image only includes a part of the battery to be tested, and the other boundary point between the first electrode and the second electrode is not displayed. Then in Figure 2, the first boundary is the upper boundary of the detection image. If the detection image includes all the batteries to be tested, and the other boundary point between the first electrode and the second electrode is also displayed in the detection image, then the first boundary can be the upper boundary of the detection image or the lower boundary of the detection image, but it is necessary to ensure that the first endpoint and the second endpoint are on the same side of the detection image.

[0080] For the detection image, the pole piece detail features are represented by pixels of different grayscale values. Since the quality of the detection image may be good or bad, the corresponding pole piece detail features may be complete or missing. By marking the end of the pole piece, the missing part of the pole piece feature in the detection image can be ignored, which is convenient for subsequent measurement.

[0081] In the embodiment of the present application, the first boundary is easily identifiable in the detection image, and it is more convenient to determine the first endpoint and the second endpoint based on the first boundary.

[0082] According to some embodiments of the present application, the battery to be tested includes a wound battery, and the detection image is obtained by projecting rays onto the surface of the wound battery for shooting, and the detection image includes any one of the four corners of the surface of the wound battery, and the surface of the wound battery is perpendicular to the thickness direction of the wound battery.

[0083] According to other embodiments of the present application, the battery to be tested includes a laminated battery, the detection image is obtained by projecting rays onto the side of the laminated battery for shooting, and the detection image includes any one of the four corners of the side of the laminated battery, and the side of the laminated battery is parallel to the thickness direction of the laminated battery.

[0084] A battery composed of cells formed by winding is a wound battery. Figure 3 is a schematic structural diagram of a wound battery provided in an embodiment of the present application. Referring to Figure 3, the wound battery 300 includes a first surface 301, a second surface 302, and two first side surfaces 303 and two second side surfaces 304 connecting the first surface 301 and the second surface 302. Among them, the second surface 302 is on the back of the wound battery 300 shown in Figure 3, and one of the first side surfaces 303 is on the bottom surface of the wound battery 300 shown in Figure 3. It cannot be directly seen from Figure 3, and only its side can be shown. The two first side surfaces 303 are both planes, and the two second side surfaces 304 are both curved surfaces. The thickness direction Z1 of the wound battery 300 is perpendicular to the first surface 301 and the second surface 302. The electrode sheet in the wound battery 300 is bent at the two second side surfaces 304. When the ray is projected onto the first surface 301 or the second surface 302 of the wound battery 300 for imaging, and includes any of the four corner positions of the first surface 301 or the second surface 302 of the wound battery 300, the bent portion of the electrode sheet can be displayed in the detection image. Since the electrode sheet boundary is also present at the corner position, the detection image can be presented as shown in Figure 2, which can be used for detection.

[0085] A laminated battery refers to a battery produced using a lamination process, in which the pole pieces are arranged in layers. FIG4 is a schematic structural diagram of a laminated battery provided in an embodiment of the present application. Referring to FIG4 , a laminated battery 400 includes a third surface 401, a fourth surface 402, and four third side surfaces 403 connecting the third surface 401 and the fourth surface 402. The thickness direction Z2 of the laminated battery 400 is parallel to the third side surface 403. The fourth surface 402 is on the back of the wound battery 300 shown in FIG3 , and cannot be directly seen in FIG3 . Only its side can be shown, and the two third side surfaces 403 cannot be directly seen in FIG3 . Since the electrodes in the laminated battery 400 are arranged in stacked layers, when the rays are projected onto the side of the laminated battery 400 for shooting, and including any of the four corners of the side of the laminated battery 400, the bent portion of the stacked electrodes can be displayed in the detection image, and since there are also boundaries of the electrodes at the corners, the detection image can also present the situation shown in Figure 2, which can be used for detection.

[0086] In the embodiments of the present application, different methods are used to obtain the detection images of different battery types, so that the method provided in the embodiments of the present application can be used to detect different types of batteries.

[0087] According to some embodiments of the present application, step S20 includes:

[0088] Step S21: input the detection image into the segmentation model to obtain the pole piece mask image output by the segmentation model.

[0089] In an embodiment of the present application, the pole piece mask diagram includes a background area and multiple end mask areas, each end mask area corresponds to the area where the end of the first pole piece is located or the area where the end of the second pole piece is located, the end of the first pole piece is the end of the first pole piece close to the first boundary, and the end of the second pole piece is the end of the second pole piece close to the first boundary.

[0090] Step S22: determining a first endpoint of the first pole piece and a second endpoint of the second pole piece based on the pole piece mask image.

[0091] The segmentation model can be trained using a gradient descent method until it can stably identify the positive and negative electrodes. Based on the grayscale features of adjacent pixels, the grayscale value of each pixel in the detection image is compared with the grayscale values ​​of adjacent pixels, and the pixels are classified into appropriate categories based on the comparison results to achieve segmentation of each target in the image. In an embodiment of the present application, the segmentation model is a pre-trained model that can directly output a pole piece mask image based on the input detection image.

[0092] In an embodiment of the present application, for the detection of excess pole piece of the battery to be tested, it is necessary to distinguish the areas where the first pole piece, the second pole piece are located and the background area. Therefore, the segmentation model can better classify each pixel point based on different grayscale ranges, output a pole piece mask map, separate the pole piece and the background, and use multiple mask end film areas to represent the area where the end of the first pole piece or the end of the second pole piece is located.

[0093] The pole piece mask output by the trained segmentation model only needs to mark the head of the pole piece, reducing workload and improving efficiency.

[0094] According to some embodiments of the present application, the multiple end mask areas include multiple first end mask areas and multiple second end mask areas, the first end mask area corresponds to the area where the end of the first pole piece is located, and the second end mask area corresponds to the area where the end of the second pole piece is located; the first endpoint is located at the center of the first end mask area; the second endpoint is located at the center of the second end mask area.

[0095] In an embodiment of the present application, since it is necessary to measure the pole piece excess between the first pole piece and the second pole piece, it is necessary to distinguish the first pole piece and the second pole piece again, and divide the multiple end mask areas into multiple independent pole piece end masks, wherein each independent pole piece end mask can include a first end mask area representing the end of the first pole piece, and also include a second end mask area representing the end of the second pole piece.

[0096] Each independent end mask region includes a center that characterizes the end of the pole piece.

[0097] Refer to Figure 2. It includes a first pole piece 100, a second pole piece 200, and a background area 500. The area where the end of the first pole piece 100 is located is a first end mask area 101, and the center of the first end mask area 101 is a first endpoint A. The area where the end of the second pole piece 200 is located is a second end mask area 201, and the center of the second end mask area 201 is a second endpoint B. By measuring the distance between the first endpoint A and the second endpoint B along the second direction Y, the pole piece overhang of the battery under test can be obtained.

[0098] By finding the center in the independent end mask area on the pole piece mask image, the end point of the pole piece in the detection image can be better located.

[0099] According to some embodiments of the present application, step S30 includes:

[0100] Step S31: determining a first distance between the adjacent first endpoint and the adjacent second endpoint along the second direction based on the adjacent first endpoint and the adjacent second endpoint along the first direction.

[0101] The second direction is perpendicular to the first direction, and in the detection image, both the first pole piece and the second pole piece extend along the second direction.

[0102] Step S32: determining the excess amount of the electrode of the battery to be tested based on the first distance.

[0103] In an embodiment of the present application, the boundary point of the first electrode in the detection image can be characterized by the first endpoint, and the boundary point of the second electrode in the detection image can be characterized by the second endpoint. Since the first electrode and the second electrode are alternately arranged along the first direction and have high consistency, the first endpoint and the second endpoint should also be fitted based on the same standard, so that the obtained first endpoint and the second endpoint are more consistent with the corresponding electrode characteristics on the basis of conforming to the arrangement rule of the first electrode and the second electrode, so that the first distance between the adjacent first endpoint and the second endpoint can be measured, and the electrode excess of the battery to be tested can be characterized by the first distance. Wherein, the adjacent first endpoint and the second endpoint mean that there is no other endpoint between the two endpoints.

[0104] The excess amount of the electrode of the battery to be tested is characterized based on the distance between the first endpoint and the second endpoint. It is only necessary to determine the first endpoint of the first electrode and the second endpoint of the second electrode, which reduces the workload and can increase the detection speed, thereby improving the detection efficiency.

[0105] According to some embodiments of the present application, step S10 includes:

[0106] Step S11: Acquire an initial image of the battery to be tested.

[0107] Step S12: performing gradient compression processing on the initial image to obtain a compressed gradient image of the battery to be tested.

[0108] Step S13: performing image reconstruction based on the compressed gradient image to obtain a reconstructed image of the battery to be tested.

[0109] Step S14: performing image enhancement processing on the reconstructed image to obtain a detection image of the battery to be tested.

[0110] In an embodiment of the present application, the initial image is an image acquired by a radiographic image acquisition device, and the original image may be subjected to a simple noise reduction process to remove noise in the original image.

[0111] In an embodiment of the present application, the initial image is a grayscale image, and the first compressed gradient image and the second compressed gradient image are both gradient images and cannot be used for detection. Image reconstruction is performed on the first compressed gradient image and the second compressed gradient image, that is, the first compressed gradient image and the second compressed gradient image are restored to the original grayscale images to obtain a reconstructed image.

[0112] In the embodiment of the present application, the detection image is obtained by processing the initial image in the above manner, which can make the electrode features in the detection image clearer and more conducive to subsequent detection.

[0113] According to some embodiments of the present application, step S14 includes:

[0114] Step S141: performing contrast enhancement processing on the reconstructed image to obtain a first pre-processed image of the battery to be tested.

[0115] Step S142: performing histogram equalization processing on the reconstructed image to obtain a second pre-processed image of the battery to be tested.

[0116] Step S143: obtaining a detection image of the battery to be tested based on the first preprocessed image and the second preprocessed image.

[0117] Exemplarily, the contrast enhancement processing can be a unidirectional contrast enhancement, which can enhance the contrast in the image while avoiding amplifying noise that may exist in the image, making the object to be detected in the image more obvious and easier to identify than other areas, and highlighting the outline of the pole piece.

[0118] Exemplarily, histogram equalization processing can be local histogram processing, which is also a contrast enhancement method. It can also enhance the contrast in the image and avoid amplifying the noise that may exist in the image, making the object to be detected in the image more obvious and easier to identify than other areas, and can highlight the details of the pole piece.

[0119] In an embodiment of the present application, the above-mentioned step S141 and step S142 can be performed simultaneously or in a distributed manner. Step S141 can be performed before step S142 or after step S142.

[0120] Different features of the electrode can be distinguished by two different image enhancement processes. Therefore, the detection image of the battery to be tested obtained based on the first pre-processed image and the second pre-processed image can highlight the different features of the electrode.

[0121] According to some embodiments of the present application, step S143 includes:

[0122] Step S1431: performing image fusion processing on the first pre-processed image and the second pre-processed image to obtain a detection image of the battery to be tested.

[0123] The purpose of image fusion is not only to reduce the amount of data, but also to construct images that are more suitable and easier to understand for human and machine perception. Image fusion is the process of combining relevant information from two or more images into one image, so that the resulting image will contain more information than either of the input images.

[0124] Through image fusion processing, the information in the first preprocessed image and the second preprocessed image can be input into the detection image, increasing the data dimension so that the detection image can include more information and improving the subsequent detection accuracy.

[0125] According to some embodiments of the present application, step S143 includes:

[0126] Step S1431: performing image fusion processing on the first preprocessed image and the second preprocessed image to obtain a third preprocessed image of the battery to be tested.

[0127] The third pre-processed image includes a plurality of first pole pieces and a plurality of second pole pieces of the battery to be tested, which are alternately arranged along a first direction, wherein the polarities of the first pole pieces and the second pole pieces are different.

[0128] Step S1432: Determine the first pole piece fitting line and the second pole piece fitting line based on the third preprocessed image to obtain an auxiliary detection image of the battery to be tested, where the first pole piece fitting line is the fitting center line of the first pole piece, and the second pole piece fitting line is the fitting center line of the second pole piece.

[0129] Step S1433: performing image fusion processing on the first pre-processed image, the second pre-processed image and the auxiliary detection image to obtain a detection image of the battery to be tested.

[0130] In an embodiment of the present application, the first pole piece in the detection image can be characterized by a continuous first pole piece fitting line, and the second pole piece in the detection image can be characterized by a continuous second pole piece fitting line. Since the first pole piece and the second pole piece are arranged alternately along the first direction and have high consistency, the first pole piece fitting line and the second pole piece fitting line should also be fitted based on the same standard, so that the obtained first pole piece fitting line and the second pole piece fitting line are more in line with the corresponding pole piece characteristics on the basis of conforming to the arrangement rules of the first pole piece and the second pole piece.

[0131] In the auxiliary detection image of the battery under test, the first electrode piece in the detection image is represented by the first electrode piece fitting line, and the second electrode piece in the detection image is represented by the continuous second electrode piece fitting line. Even if the electrode piece contour or electrode piece feature in the auxiliary detection image is discontinuous, the continuous electrode piece fitting line can still be used to obtain more accurate electrode piece features. This makes the electrode piece features more obvious in the auxiliary detection image compared to the first and second pre-processed images.

[0132] In the embodiment of the present application, the detection image obtained by fusing three images has more information than the detection image obtained by fusing two images, and can improve the subsequent detection accuracy.

[0133] The detection image obtained by fusing the two images has fewer processing steps and higher detection efficiency.

[0134] According to some embodiments of the present application, the method further includes:

[0135] Step S40: determining the electrode gap of the battery to be tested based on the distance along the first direction between the adjacent first electrode fitting line and the second electrode fitting line in the auxiliary detection image.

[0136] The first pole piece fitting line and the second pole piece fitting line are used to represent a plurality of first pole pieces and a plurality of second pole pieces arranged alternately along a first direction. Even when the outline or pole piece features in the detection image are discontinuous, more accurate pole piece gap detection data can be obtained by using the continuous pole piece fitting line.

[0137] In related technologies, the amount of electrode excess and the gap between the electrodes are detected separately using two images, which require separate processing, resulting in low detection efficiency. The method provided in the embodiments of the present application can be used to detect both the amount of electrode excess and the gap between the electrodes. In other words, two characteristics of the battery under test can be detected using a single image, thereby improving information utilization.

[0138] According to some embodiments of the present application, before step S12, the method further includes:

[0139] Step S15: performing noise reduction processing on the initial image.

[0140] Exemplarily, the noise reduction process includes at least one of mean filtering noise reduction, Gaussian filtering noise reduction, and three-dimensional block matching noise reduction.

[0141] Exemplarily, step S15 includes: first performing mean filtering noise reduction on the initial image, then performing Gaussian filtering noise reduction on the initial image, then performing three-dimensional block matching noise reduction on the initial image, and finally extracting the reflected image from the noise reduction image. Gradient compression is then performed on the initial image according to step S12 above to obtain a compressed gradient image of the battery under test.

[0142] In the embodiment of the present application, by performing noise reduction processing on the initial image, random noise such as quantum noise, aliasing noise, electronic noise, etc. in the initial image can be removed, thereby improving the signal-to-noise ratio of the initial image.

[0143] According to some embodiments of the present application, step S12 includes:

[0144] Step S121: extracting a gradient image of the initial image.

[0145] Step S122: performing gradient compression processing on the gradient image to obtain a compressed gradient image of the battery to be tested.

[0146] In an embodiment of the present application, the above step S121 may include:

[0147] Step S1211: extracting a first gradient image and a second gradient image of the initial image along a first direction and a second direction respectively.

[0148] In an embodiment of the present application, the first direction may be a pixel row direction in the initial image, and the second direction may be a pixel column direction in the initial image.

[0149] In an embodiment of the present application, the above step S122 may include:

[0150] Step S1221: Processing the first gradient image and the second gradient image using a feature pyramid algorithm to obtain a contour image, where each pixel in the contour image includes a gradient amplitude.

[0151] Step S1222: using a gradient domain dynamic range compression algorithm to process the gradient amplitude of each pixel in the contour image to obtain a compressed value of each pixel.

[0152] In an embodiment of the present application, the compression value of each pixel includes a first compression component along a first direction and a second compression component along a second direction.

[0153] Step S1223: extracting the gradient value of each pixel in the first gradient image.

[0154] Step S1224: obtaining a compressed gradient value for each pixel in the first gradient image based on the gradient value of each pixel and the first component of the compressed value of each pixel.

[0155] Step S1225: Obtain a first compressed gradient image based on the compressed gradient value of each pixel.

[0156] Step S1226: extracting the gradient value of each pixel in the second gradient image.

[0157] Step S1227: obtaining a compressed gradient value for each pixel based on the gradient value of each pixel in the second gradient image and the second component of the compressed value of each pixel.

[0158] Step S1228: Obtain a second compressed gradient image based on the compressed gradient value of each pixel.

[0159] The compressed gradient image of the battery to be tested includes a first compressed gradient image and a second compressed gradient image.

[0160] For the first gradient image, the gradient value of a pixel therein can be understood as the difference between the grayscale value of the corresponding pixel on the initial image and the grayscale value of the previous pixel on the initial image, where the corresponding pixel on the initial image and the previous pixel on the initial image are adjacently arranged along a first direction, and the previous pixel on the initial image is located in front of the corresponding pixel on the initial image along the first direction. The magnitude of the gradient value indicates the magnitude of the grayscale change in the detection image between the pixel and the adjacent pixel. A larger gradient value indicates a greater grayscale change in the detection image between the pixel and the adjacent pixel.

[0161] The compression processing compresses the gradient values ​​so that the gradient values ​​in the first compressed gradient image are smaller than those in the first gradient image, and the gradient values ​​in the second compressed gradient image are smaller than those in the second gradient image. Image reconstruction is then performed based on the first compressed gradient image and the second compressed gradient image to obtain a reconstructed image of the object to be detected. Compared to the detected image, the grayscale value variation in the reconstructed image is reduced.

[0162] In an embodiment of the present application, a gradient image is subjected to gradient compression processing to obtain a compressed gradient image of the battery to be tested, so that the gradient value in the compressed gradient image is smaller than the gradient value in the gradient image, and the change in the grayscale value in the reconstructed image obtained by compressing the gradient image is reduced. For the reconstructed image, even if the grayscale values ​​between adjacent pixels are not much different, they can be more easily identified from the reconstructed image, making the detailed features in the reconstructed image clearer and improving the accuracy of the image recognition results. Using this method in the processing of battery detection images can improve the accuracy of battery detection results.

[0163] An embodiment of the present application provides a battery detection device. Figure 5 is a block diagram of a battery detection device provided by an embodiment of the present application. Referring to Figure 5, the device includes: an acquisition module 501, configured to acquire a detection image of a battery to be tested, the detection image including a plurality of first pole pieces and a plurality of second pole pieces alternately arranged along a first direction of the battery to be tested, wherein the polarity of the first pole piece and the second pole piece are different; a first determination module 502, configured to determine a first endpoint of the first pole piece and a second endpoint of the second pole piece based on the detection image, wherein the first endpoint is the point at which the first pole piece is closest to the first boundary of the detection image, and the second endpoint is the point at which the second pole piece is closest to the first boundary of the detection image, and in the detection image, the first boundary extends along the first direction, and the first boundary does not intersect with the first pole piece and the second pole piece; a second determination module 503, configured to determine the pole piece excess amount of the battery to be tested based on the first endpoint and the second endpoint adjacent to each other along the first direction.

[0164] The definition of the battery detection device can be found in the definition of the battery detection method above and will not be repeated here. Each module in the above-mentioned battery detection device can be implemented in whole or in part by software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of the processor in the electronic device in the form of hardware, or can be stored in the memory of the electronic device in the form of software, so that the processor can call and execute the corresponding operations of each of the above modules.

[0165] The embodiment of the present application provides a segmentation model training method for battery detection. FIG6 is a flow chart of a segmentation model training method for battery detection provided by the embodiment of the present application. Referring to FIG6 , the training method includes:

[0166] Step S61: Acquire a sample image containing a sample battery.

[0167] The sample image includes a plurality of first sample pole pieces and a plurality of second sample pole pieces of the sample battery that are alternately arranged along a first direction.

[0168] Step S62: Obtain the real electrode mask image corresponding to the sample image.

[0169] The real pole piece mask image includes a real background area and multiple real end mask areas, and the multiple real end mask areas include a first real end mask area corresponding to the area where the end of the first sample pole piece is located and / or a second real end mask area corresponding to the area where the end of the second sample pole piece is located.

[0170] Step S63: Input the sample image into the segmentation model to be trained, and obtain the predicted pole piece mask image output by the segmentation model.

[0171] Step S64: Calculate the loss value based on the actual pole piece mask diagram and the predicted pole piece mask diagram.

[0172] Step S65: Adjust the parameters of the segmentation model based on the loss value.

[0173] In an embodiment of the present application, a plurality of sample images containing sample batteries may be obtained as a dataset for training a segmentation model. The sample images are images acquired by a radiographic image acquisition device. For example, the plurality of sample images may be images that have undergone simple processing.

[0174] The pole piece area in the sample image is marked to obtain the true pole piece mask corresponding to the sample image. The true pole piece mask is used as the ground truth of the segmentation model for supervised learning.

[0175] After the segmentation model outputs a predicted pole mask, the actual pole mask is compared with the predicted one, and the cross-entropy loss function is used to calculate the loss value. If the predicted pole mask differs significantly from the actual one, the loss value will also be large, requiring model parameter adjustment and iteration. If the predicted pole mask is highly consistent with the actual one, the loss value converges and approaches the optimal value, or an iteration is completed when a local optimal value is found. At this point, backpropagation can be used to update the internal parameters of the segmentation model to complete training.

[0176] The pole piece mask output by the trained segmentation model can be used to annotate only the pole piece head in the detection image, outputting the complete pole piece mask area. Using the actual mask area as the ground truth for the trained segmentation model can improve the accuracy of the segmentation model's predictions, restore the pole piece's characteristics to a certain extent, and facilitate subsequent measurements.

[0177] According to some embodiments of the present application, step S61 includes:

[0178] Step S611: Acquire a sample initial image of a sample battery.

[0179] Step S612: performing gradient compression processing on the sample initial image to obtain a sample compressed gradient image of the sample battery.

[0180] Step S613: performing image reconstruction based on the sample compressed gradient image to obtain a sample reconstructed image of the sample battery.

[0181] Step S614: performing image enhancement processing on the sample reconstructed image to obtain a sample image of the sample battery.

[0182] In the embodiment of the present application, the method for obtaining the sample image of the sample battery is the same as the method for obtaining the detection image of the battery to be tested, so that the subsequent trained model can process the detection image more accurately.

[0183] According to some embodiments of the present application, step S614 includes:

[0184] Step S6141: performing contrast enhancement processing on the sample reconstructed image to obtain a first sample pre-processed image of the sample battery.

[0185] Step S6142: Perform histogram equalization processing on the sample reconstructed image to obtain a second sample preprocessed image of the sample battery.

[0186] Step S6143: performing image fusion processing on the first sample preprocessed image and the second sample preprocessed image to obtain a sample image of the sample battery.

[0187] The two pre-processed images are fused into sample images and then annotated to create a training dataset, so that the sample detection images used for training and the detection images used for detection are obtained through the same type of side processing method, which can make the trained segmentation model process the detection images more accurately.

[0188] According to some embodiments of the present application, step S62 includes:

[0189] Step S621: marking a first marking point corresponding to the end of the first sample pole piece based on the sample image, and marking a second marking point corresponding to the end of the second sample pole piece.

[0190] Step S622: Based on the marked first marking point and the preset first expansion value, obtain the first real end mask area corresponding to the end of the first sample pole piece; based on the marked second marking point and the preset second expansion value, obtain the second real end mask area corresponding to the end of the second sample pole piece.

[0191] In an embodiment of the present application, based on the sample image, the steps of marking a first marking point corresponding to the first sample pole piece and marking a second marking point corresponding to the second sample pole piece can be completed by manual marking.

[0192] The preset first expansion value can be obtained through the empirical value of the first sample pole piece. As an exemplary embodiment, the first marked area can be a rectangle. After the first marked point is obtained, the first marked point is expanded along the first direction and the second direction respectively. Based on the preset first expansion value, the first marked point is the first marked area (that is, the real mask area corresponding to the end of the first pole piece). At this time, the first marked area can be a parallelogram, and the long side and the short side of the parallelogram are both 2 times the first expansion value. As another exemplary embodiment, the first marked area can be a circle. After the first marked point is obtained, the circle is formed after the first marked point, and the first expansion value is used as the radius for expansion. The first marked point is the first marked area (that is, the real mask area corresponding to the end of the first pole piece). At this time, the first marked area is a circle.

[0193] The preset second extension value can be obtained through the experience value of the second sample pole piece. As an exemplary embodiment, the second annotation area can be a rectangle. After the second annotation point is obtained, the second annotation point is expanded along the first direction and the second direction respectively. Based on the preset second extension value, the second annotation point is set as the second annotation area (that is, the real mask area corresponding to the end of the second pole piece). At this time, the second annotation area can be a parallelogram, and the long side and the short side of the parallelogram are both 2 times the second extension value. As another exemplary embodiment, the second annotation area can be a circle. After the second annotation point is obtained, the circle is formed after the second annotation point, and the second extension value is used as the radius for expansion. The second annotation point is set as the second annotation area (that is, the real mask area corresponding to the end of the second pole piece). At this time, the second annotation area is a circle.

[0194] Marking only two points on a sample pole piece can reduce the difficulty of marking, reduce the workload of marking, and improve the efficiency of marking.

[0195] An embodiment of the present application provides a segmentation model training device for battery detection. FIG7 is a block diagram of a segmentation model training device for battery detection provided by an embodiment of the present application. 7 , the training apparatus includes: a sample acquisition module 701, configured to acquire a sample image containing a sample battery, the sample image including a plurality of first sample pole pieces and a plurality of second sample pole pieces of the sample battery arranged alternately along a first direction; a truth module 702, configured to acquire a real pole piece mask image corresponding to the sample image, the real pole piece mask image including a real background area and a plurality of real end mask areas, the plurality of real end mask areas including a first real end mask area corresponding to the area where the end of the first sample pole piece is located and / or a second real end mask area corresponding to the area where the end of the second sample pole piece is located; a sample prediction module 703, configured to input the sample image into the segmentation model to be trained and acquire a predicted pole piece mask image output by the segmentation model; a loss module 704, configured to calculate a loss value based on the real pole piece mask image and the predicted pole piece mask image; and a parameter adjustment module 705, configured to adjust the parameters of the segmentation model based on the loss value.

[0196] Regarding the definition of the segmentation model training device for battery detection, please refer to the definition of the segmentation model training method for battery detection above, which will not be repeated here. The various modules in the above-mentioned segmentation model training device for battery detection can be implemented in whole or in part by software, hardware and their combination. The above-mentioned modules can be embedded in or independent of the processor in the electronic device in the form of hardware, or can be stored in the memory of the electronic device in the form of software, so that the processor can call and execute the operations corresponding to the above modules.

[0197] An embodiment of the present application provides a battery production method, which includes: step S801: detecting the excess amount of the electrode of the battery to be tested to determine the quality test result of the battery to be tested.

[0198] In step S801 , any of the battery detection methods in the above embodiments may be used to detect the excess amount of the electrode of the battery to be tested.

[0199] By testing the battery using the above-mentioned battery testing method, it is possible to determine whether the battery is qualified based on the excess amount of the electrode, thereby determining the quality test result of the battery to be tested.

[0200] For example, when all the electrode excess amounts are within a preset range, the battery to be tested can be determined to be qualified; when the excess amount of any battery among all the electrode excess amounts is not within the preset range, the battery to be tested can be determined to be unqualified.

[0201] According to some embodiments of the present application, the method further includes:

[0202] Step S802: In response to the quality inspection result of the battery to be tested being unqualified, the unqualified battery is discarded.

[0203] Eliminating unqualified batteries can reduce the possibility of safety accidents caused by unqualified batteries.

[0204] An embodiment of the present application provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the battery detection method of any one of the above embodiments, the training method of any one of the above embodiments, or the battery production method of the above embodiments.

[0205] An embodiment of the present application provides a battery production device, including the electronic device in the above embodiment.

[0206] An embodiment of the present application provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the battery detection method of any one of the above embodiments, the training method of any one of the above embodiments, or the battery production method of any one of the above embodiments.

[0207] An embodiment of the present application provides a computer program product, including a computer program, which, when executed by a processor, implements the battery detection method of any one of the above embodiments, the training method of any one of the above embodiments, or the battery production method of any one of the above embodiments.

[0208] An embodiment of the present application provides a battery detection method, the method comprising:

[0209] Step S901: obtaining an initial image of the battery to be tested, and performing noise reduction processing on the initial image.

[0210] Step S902: performing gradient compression processing on the initial image after noise reduction processing to obtain a compressed gradient image of the battery to be tested.

[0211] Step S903: performing image reconstruction based on the compressed gradient image to obtain a reconstructed image of the battery to be tested.

[0212] Step S904: performing contrast enhancement processing on the reconstructed image to obtain a first pre-processed image of the battery to be tested.

[0213] Step S905: performing histogram equalization processing on the reconstructed image to obtain a second pre-processed image of the battery to be tested.

[0214] Step S906: performing image fusion processing on the first pre-processed image and the second pre-processed image to obtain a detection image of the battery to be tested.

[0215] The detection image includes a plurality of first pole pieces and a plurality of second pole pieces of the battery to be tested, which are alternately arranged along a first direction, wherein the first pole pieces and the second pole pieces have different polarities. The first pole pieces and the second pole pieces in the detection image both extend along a second direction, and the first direction is perpendicular to the second direction.

[0216] Step S907: input the detection image into the segmentation model to obtain the pole piece mask image output by the segmentation model.

[0217] The pole piece mask diagram includes a background area and multiple end mask areas, each end mask area corresponds to the area where the end of the first pole piece is located or the area where the end of the second pole piece is located, the end of the first pole piece is the end of the first pole piece close to the first boundary, and the end of the second pole piece is the end of the second pole piece close to the first boundary.

[0218] Step S908: Determine the first endpoint of the first pole piece and the second endpoint of the second pole piece based on the pole piece mask image.

[0219] Step S909: determining the electrode sheet overhang of the battery to be tested based on the distance along the second direction between the first endpoint and the second endpoint adjacent to each other along the first direction.

[0220] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application, and they should all be included in the scope of the claims and specification of the present application. In particular, as long as there is no structural conflict, the various technical features mentioned in the various embodiments can be combined in any way. The present application is not limited to the specific embodiments disclosed herein, but includes all technical solutions that fall within the scope of the claims.

Claims

1. A battery detection method, the method comprising: Acquire a detection image of the battery to be tested, wherein the detection image includes a plurality of first pole pieces and a plurality of second pole pieces of the battery to be tested that are alternately arranged along a first direction, wherein the first pole pieces and the second pole pieces have different polarities; Determine a first endpoint of the first pole piece and a second endpoint of the second pole piece based on the detection image, wherein the first endpoint is a point of the first pole piece closest to a first boundary of the detection image, and the second endpoint is a point of the second pole piece closest to the first boundary of the detection image, and in the detection image, the first boundary extends along the first direction, and the first boundary does not intersect with the first pole piece and the second pole piece; The pole piece excess amount of the battery to be tested is determined based on the first endpoint and the second endpoint adjacent to each other along the first direction.

2. The method according to claim 1, wherein: The determining, based on the detection image, a first endpoint of the first pole piece and a second endpoint of the second pole piece comprises: Input the detection image into a segmentation model to obtain a pole piece mask image output by the segmentation model, wherein the pole piece mask image includes a background area and a plurality of end mask areas, each of the end mask areas corresponds to an area where an end of the first pole piece is located or an area where an end of the second pole piece is located, the end of the first pole piece is an end of the first pole piece close to the first boundary, and the end of the second pole piece is an end of the second pole piece close to the first boundary; A first endpoint of the first pole piece and a second endpoint of the second pole piece are determined based on the pole piece mask image.

3. The method according to claim 2, wherein: The plurality of end mask regions include a plurality of first end mask regions and a plurality of second end mask regions, wherein the first end mask regions correspond to regions where the ends of the first pole piece are located, and the second end mask regions correspond to regions where the ends of the second pole piece are located; The first endpoint is located at the center of the first end mask area; The second end point is located at the center of the second end mask area.

4. The method according to any one of claims 1 to 3, wherein: The determining of the electrode sheet excess amount of the battery to be tested based on the first endpoint and the second endpoint adjacent to each other along the first direction includes: Based on the first endpoint and the second endpoint that are adjacent to each other along the first direction, determining a first distance between the first endpoint and the second endpoint that are adjacent to each other along the second direction, wherein the second direction is perpendicular to the first direction; The pole piece excess amount of the battery to be tested is determined based on the first distance.

5. The method according to any one of claims 1 to 4, wherein: Obtain the test image of the battery to be tested, including: Acquiring an initial image of the battery to be tested; Performing gradient compression processing on the initial image to obtain a compressed gradient image of the battery to be tested; Performing image reconstruction based on the compressed gradient image to obtain a reconstructed image of the battery to be tested; Perform image enhancement processing on the reconstructed image to obtain a detection image of the battery to be tested.

6. The method according to claim 5, wherein: Performing image enhancement processing on the reconstructed image to obtain a detection image of the battery to be tested, including: Performing contrast enhancement processing on the reconstructed image to obtain a first preprocessed image of the battery to be tested; Performing histogram equalization processing on the reconstructed image to obtain a second preprocessed image of the battery to be tested; A detection image of the battery to be tested is obtained based on the first preprocessed image and the second preprocessed image.

7. The method according to claim 6, wherein: Obtaining a detection image of the battery to be tested based on the first preprocessed image and the second preprocessed image, comprising: The first preprocessed image and the second preprocessed image are subjected to image fusion processing to obtain a detection image of the battery to be tested.

8. The method according to claim 6, wherein: Obtaining a detection image of the battery to be tested based on the first preprocessed image and the second preprocessed image, comprising: Performing image fusion processing on the first preprocessed image and the second preprocessed image to obtain a third preprocessed image of the battery to be tested, wherein the third preprocessed image includes a plurality of first pole pieces and a plurality of second pole pieces of the battery to be tested that are alternately arranged along the first direction, wherein the first pole pieces and the second pole pieces have different polarities; Determine a first pole piece fitting line and a second pole piece fitting line based on the third preprocessed image to obtain an auxiliary detection image of the battery to be tested, wherein the first pole piece fitting line is a fitting center line of the first pole piece, and the second pole piece fitting line is a fitting center line of the second pole piece; Perform image fusion processing on the first preprocessed image, the second preprocessed image and the auxiliary detection image to obtain a detection image of the battery to be tested.

9. The method according to claim 8, wherein: The method further comprises: Based on the distance along the first direction between the first pole piece fitting line and the second pole piece fitting line adjacent to each other in the auxiliary detection image, the pole piece gap of the battery to be tested is determined.

10. The method according to any one of claims 5 to 9, wherein: Before performing gradient compression processing on the initial image to obtain a compressed gradient image of the battery to be tested, the method further includes: The initial image is subjected to noise reduction processing.

11. The method according to any one of claims 5 to 10, wherein: Performing gradient compression processing on the initial image to obtain a compressed gradient image of the battery to be tested includes: Extracting a gradient image of the initial image; Performing gradient compression processing on the gradient image to obtain a compressed gradient image of the battery to be tested.

12. A battery detection device, comprising: An acquisition module is configured to acquire a detection image of the battery to be tested, wherein the detection image includes a plurality of first pole pieces and a plurality of second pole pieces of the battery to be tested that are alternately arranged along a first direction, wherein the first pole pieces and the second pole pieces have different polarities; A first determination module is configured to determine a first endpoint of the first pole piece and a second endpoint of the second pole piece based on the detection image, wherein the first endpoint is a point of the first pole piece closest to a first boundary of the detection image, and the second endpoint is a point of the second pole piece closest to the first boundary of the detection image, and in the detection image, the first boundary extends along the first direction, and the first boundary does not intersect with the first pole piece and the second pole piece; The second determination module is configured to determine the pole piece excess amount of the battery to be tested based on the first endpoint and the second endpoint adjacent to each other along the first direction.

13. A segmentation model training method for battery detection, the training method comprising: Acquire a sample image containing a sample battery, wherein the sample image includes a plurality of first sample pole pieces and a plurality of second sample pole pieces of the sample battery alternately arranged along a first direction; Acquire a real pole piece mask image corresponding to the sample image, wherein the real pole piece mask image includes a real background area and multiple real end mask areas, wherein the multiple real end mask areas include a first real end mask area corresponding to an area where an end of the first sample pole piece is located and / or a second real end mask area corresponding to an area where an end of the second sample pole piece is located; Inputting the sample image into the segmentation model to be trained, and obtaining a predicted pole piece mask image output by the segmentation model; Calculating a loss value based on the actual pole piece mask diagram and the predicted pole piece mask diagram; Parameters of the segmentation model are adjusted based on the loss value.

14. The training method according to claim 13, wherein: Get a sample image containing a sample battery, including: Acquiring a sample initial image of the sample battery; Performing gradient compression processing on the sample initial image to obtain a sample compressed gradient image of the sample battery; Performing image reconstruction based on the sample compressed gradient image to obtain a sample reconstructed image of the sample battery; Perform image enhancement processing on the sample reconstructed image to obtain a sample image of the sample battery.

15. The training method according to claim 14, wherein: Performing image enhancement processing on the sample reconstructed image to obtain a sample image of the sample battery includes: Performing contrast enhancement processing on the sample reconstructed image to obtain a first sample preprocessed image of the sample battery; Performing histogram equalization processing on the sample reconstructed image to obtain a second sample preprocessed image of the sample battery; The first sample preprocessed image and the second sample preprocessed image are subjected to image fusion processing to obtain a sample image of the sample battery.

16. The training method according to any one of claims 13 to 15, wherein: Obtaining a real pole piece mask image corresponding to the sample image, including: Marking a first marking point corresponding to an end of the first sample pole piece based on the sample image, and marking a second marking point corresponding to an end of the second sample pole piece; Based on the marked first annotation point and the preset first expansion value, a first real end mask area corresponding to the end of the first sample pole piece is obtained; based on the marked second annotation point and the preset second expansion value, a second real end mask area corresponding to the end of the second sample pole piece is obtained.

17. A segmentation model training device for battery detection, the training device comprising: A sample acquisition module is configured to acquire a sample image containing a sample battery, wherein the sample image includes a plurality of first sample pole pieces and a plurality of second sample pole pieces of the sample battery alternately arranged along a first direction; a truth value module, configured to obtain a real pole piece mask image corresponding to the sample image, the real pole piece mask image comprising a real background area and a plurality of real end mask areas, the plurality of real end mask areas comprising a first real end mask area corresponding to an area where an end of the first sample pole piece is located and / or a second real end mask area corresponding to an area where an end of the second sample pole piece is located; A sample prediction module is configured to input the sample image into a segmentation model to be trained, and obtain a predicted polar mask image output by the segmentation model; A loss module, configured to calculate a loss value based on the actual pole piece mask diagram and the predicted pole piece mask diagram; A parameter adjustment module is configured to adjust the parameters of the segmentation model based on the loss value.

18. A method for producing a battery, the method comprising: The battery detection method according to any one of claims 1 to 11 is used to detect the excess amount of the electrode of the battery to be tested, so as to determine the quality detection result of the battery to be tested.

19. The method according to claim 18, wherein: The method further comprises: In response to the quality inspection result of the battery to be tested being unqualified, the unqualified battery is discarded.

20. An electronic device, comprising: at least one processor; as well as a memory communicatively coupled to the at least one processor; The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the battery detection method described in any one of claims 1 to 11 or the training method described in any one of claims 13 to 16 or the battery production method described in claim 18 or 19.

21. A battery production device comprising the electronic device according to claim 20.

22. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the battery detection method according to any one of claims 1 to 11 or the training method according to any one of claims 13 to 16 or the battery production method according to claim 18 or 19 is implemented.

Images