Battery post-burning traceability method and system based on heterogeneous double code

Abstract

The present application relates to the technical field of battery safety monitoring, and in particular to a battery combustion post-tracing method and system based on heterogeneous double codes, which solves the technical problem that the prior art is difficult to identify and trace under abnormal conditions. The method comprises: acquiring a micro-depth image of the surface of the battery tab, and performing image depth analysis based on the micro-depth image to determine the character sequence corresponding to the plaintext character distribution area and the etching point coordinate set corresponding to the two-dimensional code distribution area of the surface of the battery tab; database traversal matching based on the character sequence obtains at least one candidate factory identification code and at least one candidate factory identification code corresponding two-dimensional code digital bitmap respectively; based on the etching point coordinate set, structure comparison analysis is performed on the two-dimensional code digital bitmap corresponding to the at least one candidate factory identification code respectively, and the target factory identification code is traced from the at least one candidate factory identification code.

Description

Technical Field

[0001] This invention relates to the field of battery safety monitoring technology, specifically to a method and system for tracing batteries after combustion based on heterogeneous dual codes. Background Technology

[0002] Traceability of the entire life cycle of new energy batteries is a core prerequisite for conducting battery safety monitoring, failure analysis, and accident tracing. To meet this traceability requirement, battery manufacturers generally use laser etching technology to mark heterogeneous dual codes on the metal surface of the cell tabs. These heterogeneous dual codes are a combination of high-density QR codes and low-density plain text characters as identification identifiers. Under normal temperature and routine maintenance conditions, industrial code reading systems can quickly confirm the cell's factory identity by recognizing these heterogeneous dual codes, thus meeting the basic requirements for battery life cycle traceability.

[0003] After a battery experiences extreme high-temperature damage, such as a fire, the surface of the tab will be covered with carbonized deposits and an oxide layer. Furthermore, the thin metal foil of the tab will wrinkle and undergo asymmetrical curling deformation. Current mainstream industrial barcode reading systems rely on image grayscale contrast and a two-dimensional global coordinate system for projection distance alignment and feature registration. However, when optical grayscale features on the tab surface are lost and nonlinear three-dimensional deformation exists, the registration algorithm will experience large-scale misalignment, and the registration error will be continuously amplified. This causes conventional heterogeneous dual-code recognition and decoding technologies to completely fail, making it impossible to accurately confirm the factory identity of the damaged battery cell. This makes it difficult to effectively trace the source of battery fires, resulting in poor overall traceability and failing to meet the actual industry demand for accurate tracing of the battery cell's factory identity after a new energy battery accident. Summary of the Invention

[0004] To address the technical problem of existing technologies struggling to perform identification and traceability under abnormal circumstances, the present invention aims to provide a method and system for tracing batteries after combustion based on heterogeneous dual-code technology. The specific technical solution adopted is as follows: This application provides a method for tracing battery combustion based on heterogeneous dual-code, including: A microscopic depth image of the battery tab surface is acquired, and image depth analysis is performed based on the microscopic depth image to determine the character sequence corresponding to the plaintext character distribution area on the battery tab surface and the set of etching point coordinates corresponding to the QR code distribution area; the set of etching point coordinates includes the three-dimensional absolute coordinates of each etching point within the QR code distribution area. Database traversal and matching are performed based on character sequences to obtain at least one candidate factory identification code and the QR code bitmap corresponding to each candidate factory identification code. Based on the set of etched point coordinates, a structural comparison analysis is performed on the QR code bitmap corresponding to at least one candidate factory identification code, and the target factory identification code is obtained by tracing back from at least one candidate factory identification code.

[0005] In one possible implementation, the method includes: Multi-angle illumination scanning was performed on the battery tabs to acquire shadow image sequences under different illumination directions; The relative height field of the battery tab surface is constructed based on the shadow image sequence, and the etching foreground binary image is extracted based on the relative height field; the relative height field is used to characterize the depth value of each pixel on the battery tab surface relative to the camera field of view reference plane; Based on the binary image of the etching foreground, determine the character sequence corresponding to the plaintext character distribution area and the set of etching point coordinates corresponding to the QR code distribution area.

[0006] In one possible implementation, the method includes: Optical character recognition is performed on the plaintext character distribution area in the etched foreground binary image, and the characters that fail to be recognized are mapped as placeholders to generate the character sequence corresponding to the plaintext character distribution area. Geometric centroids of each etching point in the QR code distribution area of ​​the etching foreground binary image are extracted, and the three-dimensional absolute coordinates of each etching point are determined by combining the camera intrinsic parameter matrix, thus generating a set of etching point coordinates.

[0007] In one possible implementation, the method includes: Generate regular expression query commands based on character sequences; regular expression query commands are used to indicate the factory identification codes that match the character order and arbitrary length character intervals in the character sequence. The database is traversed and matched according to the regular expression query command to filter out at least one candidate factory identification code from the database. For each candidate factory identification code, the encoder converts the candidate factory identification code into the corresponding QR code digital bitmap.

[0008] In one possible implementation, the method includes: A standard grid structure diagram is constructed for each QR code digital bitmap, and an actual grid structure diagram is constructed based on the set of etched point coordinates. The standard grid structure diagram is used to represent each etched point in the QR code digital bitmap and the edges that are connected to each etched point. The actual grid structure diagram is used to represent each etched point in the set of etched point coordinates and the edges that are connected to each etched point. The actual mesh structure diagram is matched with each standard mesh structure diagram in a fault-tolerant topology, and the target factory identification code is obtained by tracing from at least one candidate factory identification code based on the matching result.

[0009] In one possible implementation, the method includes: For each candidate factory identification code, the QR code digital bitmap is mapped to coordinates to generate a standard coordinate set that is consistent with the scale of the etched point coordinate set. Based on the standard center distance of the factory identification code, establish the connection relationship between each etching point in each standard coordinate set and construct the corresponding standard mesh structure diagram. The actual adjacency determination radius is determined based on the standard center distance and the upper limit of the material deformation rate of the battery tabs. The connection relationship between each etching point in the etching point coordinate set is established based on the actual adjacency determination radius, and the actual mesh structure diagram is constructed.

[0010] In one possible implementation, the method includes: For each standard mesh structure diagram, a fault-tolerant topology matching is performed between the standard mesh structure diagram and the actual mesh structure diagram to determine the matching result of the standard mesh structure diagram; the matching result includes the etched points and edges that match in the corresponding standard mesh structure diagram and the actual mesh structure diagram; The number of matching etch points and the deformation ratio of each matching edge are determined based on the matching results, and the target factory identification code is determined from at least one candidate factory identification code based on the number of matching etch points and the deformation ratio of each matching edge.

[0011] In one possible implementation, the method includes: The maximum deformation ratio of each standard mesh structure diagram is determined based on the matching results; the maximum deformation ratio is the maximum value of the deformation ratio of the matching edges in the standard mesh structure diagram and the actual mesh structure diagram. If there is a minimum maximum deformation ratio among the maximum deformation ratios of each standard grid structure diagram, the candidate factory identification code corresponding to the minimum maximum deformation ratio shall be used as the target factory identification code. If there are multiple minimum maximum deformation ratios in each standard mesh structure diagram, the candidate factory identification code with the most matching etching points in the standard mesh structure diagram corresponding to the multiple minimum maximum deformation ratios is taken as the target factory identification code.

[0012] In one possible implementation, the method further includes: Select the standard mesh structure diagram from each standard mesh structure diagram, where the maximum deformation ratio is less than or equal to the upper limit of the material deformation rate of the battery tab; If there is no standard mesh structure diagram in each standard mesh structure diagram whose maximum deformation ratio is less than or equal to the upper limit of the material deformation rate of the battery tab, then the current traceability anomaly is determined.

[0013] This application provides a battery combustion traceability system based on heterogeneous dual codes, including: The feature extraction module is used to acquire a microscopic depth image of the battery tab surface, and perform image depth analysis based on the microscopic depth image to determine the character sequence corresponding to the plaintext character distribution area on the battery tab surface and the set of etching point coordinates corresponding to the QR code distribution area; the set of etching point coordinates includes the three-dimensional absolute coordinates of each etching point within the QR code distribution area. The candidate matching unit is used to perform database traversal matching based on the character sequence to obtain at least one candidate factory identification code and the QR code bitmap corresponding to each candidate factory identification code. The traceability verification unit is used to perform structural comparison and analysis on the QR code digital bitmap corresponding to at least one candidate factory identification code based on the set of etched point coordinates, and to trace the source of the target factory identification code from at least one candidate factory identification code.

[0014] The present invention has the following beneficial effects: Based on the above technical solution, this application effectively removes the interference of carbonized coking and oxide layers by acquiring microscopic depth images of the battery tab surface and performing image depth analysis. Simultaneously, it determines the character sequence corresponding to the plaintext character distribution area and the set of etched point coordinates corresponding to the QR code distribution area, providing accurate feature basis for subsequent traceability. Then, database traversal matching based on the character sequence can quickly narrow down the traceability range and filter out candidate factory identification codes that match the character features, improving traceability efficiency. Finally, structural comparison analysis is performed based on the etched point coordinate set and the QR code bitmap corresponding to each candidate factory identification code to trace the source and obtain the target factory identification code. This effectively solves the technical problem that conventional two-dimensional image registration algorithms fail due to coordinate origin drift and nonlinear distortion when the tab surface is covered with coking material after a fire and undergoes heat-induced wrinkling, significantly improving the success rate of accident analysis for severely damaged batteries. Attached Figure Description

[0015] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating a battery combustion tracing method based on heterogeneous dual codes, provided in one embodiment of the present invention. Figure 2 This is a system architecture diagram of a battery combustion traceability system based on heterogeneous dual codes, provided as an embodiment of the present invention. Detailed Implementation

[0017] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the battery combustion tracing method and system based on heterogeneous dual codes proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0019] In all division and logarithmic operations covered in this application, a smoothing mechanism is employed to prevent computer program crashes or invalid values ​​from being generated due to a zero denominator or a zero input. Specifically, a positive correction factor is superimposed on the denominator term of the division operation or the argument term of the logarithmic function. For example, the value is This ensures the robustness and feasibility of the algorithm under extreme conditions.

[0020] The normalization function mentioned in this application Unless otherwise specified, all values ​​are normalized using maximum and minimum values. The maximum and minimum values ​​are preset empirical extreme values ​​derived from a large amount of historical experimental data. If the calculated result exceeds the [0,1] interval, it is restricted to the [0,1] range by a truncation function (i.e., if the result is less than 0, it is taken as 0, and if it is greater than 1, it is taken as 1) to eliminate the influence of outliers on the evaluation index.

[0021] The specific scheme of the battery combustion traceability method and system based on heterogeneous dual codes provided by the present invention will be described in detail below with reference to the accompanying drawings.

[0022] Please see Figure 1 The diagram illustrates a flowchart of a battery combustion tracing method based on heterogeneous dual codes according to an embodiment of the present invention. The method includes the following steps: Step 101: Obtain a microscopic depth image of the battery tab surface, and perform image depth analysis based on the microscopic depth image to determine the character sequence corresponding to the plaintext character distribution area on the battery tab surface and the set of etching point coordinates corresponding to the QR code distribution area.

[0023] The set of etched point coordinates includes the three-dimensional absolute coordinates of each etched point within the QR code distribution area.

[0024] It should be noted that because the surface of the battery tabs is covered with black carbonized coke after the fire, conventional two-dimensional vision devices cannot identify the surface etching patterns by extracting color and grayscale contrast. Furthermore, the heat-induced wrinkling of the tabs causes complex undulations on the metal surface, meaning that the distance between each etching pit and the actual object distance to the camera is necessarily different. Therefore, this application can extract the true pit morphology hidden beneath by acquiring microscopic depth images and performing depth analysis, thus removing the color and grayscale interference from the surface coke.

[0025] Among them, the micro-depth image is an image that can characterize the depth information of various locations on the surface of the battery tab. Unlike traditional two-dimensional grayscale images, the micro-depth image can effectively remove the interference of carbonized coke and oxide layers on the tab surface, and capture the depth features of the etching pits formed by laser etching. It is suitable for extracting the features of the tab surface after battery combustion. The plaintext character distribution area is the area of ​​low-density plaintext characters laser-etched on the tab surface, and the QR code distribution area is the area of ​​high-density QR code laser-etched on the tab surface. The etching point is the pit feature point formed by laser etching within the QR code area. Its three-dimensional absolute coordinates are the absolute coordinate points of the etching point in physical space constructed with a fixed dimension, which can objectively reflect the actual spatial position of the etching point. For example, it can be represented by three-dimensional coordinates with millimeters as the unit.

[0026] Step 102: Perform database traversal matching based on the character sequence to obtain at least one candidate factory identification code and the QR code bitmap corresponding to each candidate factory identification code.

[0027] The character sequence is formed by arranging the characters identified in the plaintext character distribution area according to spatial order. Due to surface damage caused by the fire, the character sequence may contain placeholder marks representing the covered area, so it may not be able to completely represent the factory identification code and can only be used as the basis for database matching.

[0028] The factory database stores the factory identification code and corresponding QR code digital bitmap of all battery cells produced during the battery manufacturing process. The QR code digital bitmap is a two-dimensional digital image obtained after the factory identification code is converted by the encoder, and contains the information of the etching point arrangement of the QR code.

[0029] In one possible implementation, this application can generate regular expression query instructions based on character sequences.

[0030] The regular expression query command is used to indicate the factory identification code that matches the character order and arbitrary length character spacing in the character sequence.

[0031] Because surface wrinkling caused by a fire causes macroscopic area shrinkage of the tab metal, the actual physical spacing between the remaining plaintext characters is reduced non-linearly, making it impossible for conventional fixed-length string matching to retrieve the original record in the factory database.

[0032] Therefore, the regular expression query instruction constructed in this application can preserve the absolute order of recognizable characters in a character sequence and remove placeholders (such as...) Replace `or ?` with a variable-length regular expression wildcard that allows arbitrary scaling of the spacing between preceding and following characters (e.g., `?`). This allows the database retrieval to tolerate irregular variations in character spacing, ensuring that factory identification codes that match the character characteristics can be selected.

[0033] For example, if the character sequence is "12?AB", then the placeholder wildcard "?" will be replaced with a variable-length regular expression wildcard ". The generated regular expression query can match all factory identification codes that start with "12", end with "AB", and contain any characters in between.

[0034] Subsequently, this application can perform traversal matching in the database according to regular expression query instructions to filter out at least one candidate factory identification code from the database.

[0035] This database can serve as a production traceability database for battery manufacturers, storing information such as the factory identification code, production batch, and production time of all battery cells.

[0036] In some embodiments, this application can further limit the traversal and matching range by combining the manufacturing time window. This application can obtain the production batch information of the vehicle where the currently inspected fault battery is located or other traceable components of the vehicle (such as energy storage cabinets), and limit the query range to the manufacturing time range corresponding to that batch. For example, extract the batch's warehousing registration date and extend it before and after by 3 months to form a time constraint condition, avoiding the problems of excessive computing power consumption and too many candidates caused by unlimited traversal. The database engine will compare each manufacturing identification code within the manufacturing time range with the matching rules of the regular expression query command, and filter out all manufacturing identification codes that meet the rules as candidate manufacturing identification codes.

[0037] Thus, this application can convert each candidate factory identification code into a corresponding QR code digital bitmap using an encoder.

[0038] The encoder used is the standard QR code encoder employed when this batch of tabs left the factory. It maintains the same encoding rules as the laser etching process, accurately converting the candidate factory identification code (plain text) into a two-dimensional digital bitmap. This QR code bitmap contains etching point arrangement information that is completely identical to the laser-etched QR code at the factory, serving as the theoretical basis for subsequent structural comparison and analysis. During the conversion process, it is necessary to adhere to the factory QR code encoding standard to ensure that the etching point structure of the generated QR code bitmap is consistent with the actual factory etching structure.

[0039] Step 103: Based on the set of etched point coordinates, perform structural comparison analysis with the QR code bitmap corresponding to at least one candidate factory identification code, and trace the source of the target factory identification code from at least one candidate factory identification code.

[0040] It should be noted that due to the nonlinear three-dimensional deformation of the electrode surface after a fire, feature registration algorithms relying on a two-dimensional global coordinate system will encounter large-scale misalignment and fail. Therefore, this application can match and analyze the spatial structure of the etched points in the QR code area of ​​the damaged electrode with the etched point arrangement structure of the QR code bitmap corresponding to the candidate factory identification code, thereby adapting to the spatial position changes of the etched points, and finally tracing back to obtain the target factory identification code that matches the actual factory identity of the damaged electrode, which can be used for subsequent failure analysis and liability determination in battery accidents.

[0041] Based on the above technical solution, this application effectively removes the interference of carbonized coking and oxide layers by acquiring microscopic depth images of the battery tab surface and performing image depth analysis. Simultaneously, it determines the character sequence corresponding to the plaintext character distribution area and the set of etched point coordinates corresponding to the QR code distribution area, providing accurate feature basis for subsequent traceability. Then, database traversal matching based on the character sequence can quickly narrow down the traceability range and filter out candidate factory identification codes that match the character features, improving traceability efficiency. Finally, structural comparison analysis is performed based on the etched point coordinate set and the QR code bitmap corresponding to each candidate factory identification code to trace the source and obtain the target factory identification code. This effectively solves the technical problem that conventional two-dimensional image registration algorithms fail due to coordinate origin drift and nonlinear distortion when the tab surface is covered with coking material after a fire and undergoes heat-induced wrinkling, significantly improving the success rate of accident analysis for severely damaged batteries.

[0042] As a possible embodiment of this application, step 101 above can be implemented through the following steps: Step 201: Perform multi-angle illumination scanning on the battery tabs and collect shadow image sequences under different illumination directions.

[0043] Because the surface of the battery tabs is covered with black carbonized coke after a fire, conventional two-dimensional vision devices cannot identify the surface etching patterns by extracting color and grayscale contrast. Therefore, this application can use light sources at various specific angles and combine them with three-dimensional calculation methods to remove the color and grayscale interference of the surface coke, thereby obtaining a shadow image sequence and realizing the extraction of the true pit morphology hidden underneath.

[0044] Among them, the shadow image sequence is a collection of images that reflect the changes in brightness and darkness on the surface of the electrode, which are simultaneously acquired by an industrial camera under different lighting directions. The images under different lighting directions can show the depth features of different positions on the surface of the electrode, especially the shadow differences of the pits formed by laser etching under different lighting conditions, which provides basic data for the subsequent construction of the depth field.

[0045] For example, this application can retrieve the industrial camera intrinsic parameter matrix and lens distortion coefficient pre-stored in memory during the equipment calibration stage, as well as the factory-rated etching depth parameters of the tab material during batch processing. Then, it controls a multi-segmented ring-shaped light-emitting diode (LED) light source arranged above the inspection stage to perform multi-angle illumination scanning on the tab, controlling the light source sectors to illuminate sequentially in a circumferential order, while simultaneously controlling an orthogonally downward-placed industrial camera to synchronously trigger exposure (e.g., synchronously triggering exposure with an exposure time of 5 milliseconds), thereby acquiring and generating the corresponding shadow image sequence. The LED light source used in this application can be a high-brightness, high-power synchronous flash light source, ensuring that a shadow image sequence meeting the signal-to-noise ratio requirements is obtained under low-reflectivity surface (black surface) conditions.

[0046] Step 202: Construct the relative height field of the battery tab surface based on the shadow image sequence, and extract the etching foreground binary image based on the relative height field.

[0047] The relative height field is used to characterize the depth value of each pixel on the surface of the battery tab relative to the camera's field of view reference plane.

[0048] Because the coking residue from the fire forms irregular protrusions on the surface of the electrode tab, and the electrode tab itself may have macroscopic undulations due to heat wrinkling, it is impossible to use a globally uniform absolute height threshold for filtering. Therefore, this application can dynamically calculate the local reference plane on the surface of the battery electrode tab and combine it with the factory-set etching depth as a dynamic depth lower limit threshold, thereby accurately filtering upwardly protruding contaminants and flat substrates that have not been broken by the laser, achieving the separation of the target etched pits.

[0049] For example, this application can use a photometric stereo algorithm to process the shadow image sequence. This photometric stereo algorithm analyzes the changes in shadows and highlights on an object's surface under different light sources, and can deduce the tilt angle of the object's surface. Thus, this application can calculate the surface normal vector corresponding to each pixel on the battery tab surface. Then, based on the calculated surface normal vector, a two-dimensional gradient integral operation is performed along the X and Y axes of the pixel plane to reconstruct the microscopic relative height field of the tab surface. ,in For pixel positions in the image coordinate system, the function value This value represents the depth of the pixel relative to the current camera field of view reference plane. The smaller the value (i.e., the larger the absolute value of the negative value), the farther the pixel is from the camera lens in physical space, meaning the deeper the etch pit.

[0050] Because the surface of the electrode exhibits macroscopic undulations caused by heating and wrinkling, as well as irregular local protrusions caused by carbonized coke, this application can obtain a local dynamic reference surface for subsequent binary processing after generating the microscopic relative height field. For example, for each pixel in the microscopic relative height field, a sliding window is set centered on that pixel (the window size can be determined according to the relative scale of the laser etching point, for example, set to 2 to 3 times the pixel diameter occupied by the etching point). Based on the depth value distribution of all pixels within the sliding window, the corresponding local reference surface depth is calculated (for example, it can be the Gaussian weighted average or mode of each depth value within the sliding window). Then, the factory-rated etching depth parameter is subtracted from the local reference surface depth to obtain the lower limit threshold of the depth corresponding to that pixel. Iterateing through each pixel in the microscopic relative height field, the depth value of that pixel is compared numerically with the corresponding lower limit threshold. Based on the comparison results, a binary mapping is performed to obtain the etching foreground binary image.

[0051] For example, the etched foreground binary image satisfies the following formula: in, Represents the pixels in the etched foreground binary image. Binarized numerical values, For pixels in the microscopic relative height field The depth value, For pixels The corresponding depth lower limit threshold.

[0052] When the depth value is less than the corresponding lower depth threshold, the matrix position corresponding to the pixel is assigned a value of 1, indicating that the position is a real etch pit; when the depth value is greater than or equal to the corresponding lower depth threshold, the matrix position corresponding to the pixel is assigned a value of 0, indicating that the position is the background (including surface coking protrusions and flat metal substrate).

[0053] Morphological opening operations are performed on the obtained etching foreground binary image to remove isolated noise points caused by scratches or tiny particles of coking material on the metal surface, thus obtaining the final etching foreground binary image.

[0054] Step 203: Determine the character sequence corresponding to the plaintext character distribution area and the set of etching point coordinates corresponding to the QR code distribution area based on the etching foreground binary image.

[0055] In one possible implementation, this application can perform optical character recognition on the plaintext character distribution area in the etched foreground binary image, and map the characters that fail to be recognized as placeholders to generate the character sequence corresponding to the plaintext character distribution area.

[0056] Among them, Optical Character Recognition (OCR) can extract characters from the plaintext character distribution area in the etched foreground binary image. Since some strokes of the plaintext characters on the electrode surface may be melted or obscured by carbides after battery combustion, resulting in some characters being unrecognizable, the unrecognizable character areas can be mapped to placeholder wildcards. These placeholder wildcards represent missing characters, and the appropriate wildcard can be selected based on actual needs. Symbols such as "?" and "?" are used. When generating character sequences, the relative arrangement order of recognizable characters on the X-axis of the image is strictly preserved. Recognizable characters and placeholder wildcards are concatenated according to spatial order to ensure that the generated character sequence can truly reflect the incomplete features of plaintext characters.

[0057] Meanwhile, the geometric centroids of each etching point in the QR code distribution area of ​​the etching foreground binary image are extracted, and the three-dimensional absolute coordinates of each etching point are determined by combining the camera intrinsic parameter matrix to generate a set of etching point coordinates.

[0058] For example, this application can traverse all foreground connected regions with a pixel value of 1 within the QR code distribution area, with each foreground connected region corresponding to an etch point, and calculate the geometric centroid pixel coordinates of each foreground connected region, denoted as... ,in Indicates the first One etched point was detected.

[0059] Then, by combining the observation of the relative height field, the depth values ​​corresponding to each etching point were retrieved. Distance from the reference object calibrated by the equipment The actual object distance of the etching point in three-dimensional space is updated to the dynamic object distance by adding the vertical distances (i.e., the vertical distance from the camera lens to the inspection stage plane) together. (Right now It then retrieves the pre-loaded camera intrinsic parameter matrix and uses matrix multiplication to convert the pixel coordinates, which originally lacked a realistic scale, into three-dimensional absolute coordinates containing depth information.

[0060] For example, the calculation process of the conversion operation satisfies the following formula: in, For the first The three-dimensional absolute coordinates of the geometric centroid of each etched point. For the first Dynamic object distance at each etching point It is the inverse of the camera intrinsic parameter matrix. , They represent the first The geometric centroid of each etch point is located at the horizontal and vertical pixel coordinates on the image plane, with constant 1 serving as a placeholder for homogeneous coordinates.

[0061] After conversion, this application can obtain the three-dimensional absolute coordinates of the geometric centroid of each etching point in the actual testing stage space, thus forming a set of etching point coordinates.

[0062] Based on the above technical solution, this application, by scanning the battery tabs under multiple angles of illumination and acquiring shadow image sequences, can fully capture the depth features of the etching pits on the tab surface under different illuminations, providing comprehensive and accurate basic data for the subsequent construction of the depth field. Based on the shadow image sequence, a relative height field is constructed, converting the brightness changes of the image into quantified depth values, realizing the digital representation of the depth features of the tab surface. Then, based on the relative height field, an etching foreground binary map is extracted, effectively removing background interference such as carbonized coke and oxide layers, retaining only the effective features of laser etching. Finally, based on the etching foreground binary map, the character sequence and the set of etching point coordinates are determined, realizing the accurate conversion from depth image to traceability feature information, improving the accuracy and reliability of feature extraction of damaged tab surface, avoiding extraction errors caused by background interference and feature loss, and laying a solid feature foundation for subsequent traceability work.

[0063] As a possible embodiment of this application, step 103 above can be implemented through the following steps: Step 301: Construct a standard grid structure diagram for each QR code digital bitmap, and construct an actual grid structure diagram based on the set of etched point coordinates.

[0064] The standard mesh structure diagram is used to represent each etched point in the QR code digital bitmap and the edges connecting these etched points. The actual mesh structure diagram is used to represent each etched point in the set of etched point coordinates and the edges connecting these etched points. In other words, the mesh structure diagram is a topological network constructed with etched points as nodes and the adjacency relationships between etched points as edges, providing a direct representation of the spatial structure of the etched points.

[0065] The aforementioned standard mesh structure diagram is constructed based on the QR code digital bitmap corresponding to the candidate factory identification code. It reflects the undeformed theoretical topological structure of the QR code etching points when the battery cell leaves the factory. Its nodes are the etching points in the QR code digital bitmap, and the edges are the inherent adjacent connections between the etching points. The aforementioned actual mesh structure diagram is constructed based on the coordinate set of the etching points of the damaged tab. It reflects the actual topological structure of the QR code etching points after the tab is deformed following battery combustion. Its nodes are the actual etching points on the surface of the damaged tab, and the edges are the adjacent connections between the actual etching points.

[0066] In one possible implementation, this application can perform coordinate mapping on the QR code bitmap corresponding to each candidate factory identification code to generate a standard coordinate set that is consistent with the scale of the etched point coordinate set.

[0067] Coordinate mapping refers to the process of converting the pixel coordinates of etched points in the QR code digital bitmap into three-dimensional absolute coordinates with the same dimensions as the etched point coordinate set. Taking the dimension of the etched point coordinate set as millimeters as an example, it is necessary to retrieve the laser marking machine's factory printing scale when the batch of tabs leaves the factory. This scale represents the physical millimeter width corresponding to a unit pixel. The two-dimensional pixel coordinates of the etched points in the QR code digital bitmap are multiplied by this scale to obtain two-dimensional millimeter coordinates. Then, a zero reference depth coordinate is uniformly added to each etched point, and finally a standard coordinate set containing three-dimensional absolute coordinates is generated to ensure that the dimensions and spatial scale of the standard coordinate set are consistent with the etched point coordinate set, laying the foundation for subsequent comparison of the mesh structure.

[0068] Subsequently, based on the standard center distance of the factory identification code, the connection relationship between each etching point in each standard coordinate set is established, and the corresponding standard mesh structure diagram is constructed.

[0069] The standard center distance, set at the factory for this batch of QR codes, is the theoretical distance between two adjacent etched points and is the core basis for determining the adjacency relationship between etched points. First, this application determines the adjacency determination radius based on the standard center distance. The adjacency determination radius is a fixed multiple of the standard center distance, ensuring that adjacent etched points in the diagonal direction can also establish a connection. For example, the adjacency determination radius can be set to 1.8 times the standard center distance. Then, the three-dimensional spatial linear Euclidean distance between each pair of etched points in the standard coordinate set is calculated. If the distance between two etched points is less than or equal to the adjacency determination radius, an undirected connecting edge is established between them, representing the existence of a connection relationship. Finally, a standard mesh structure diagram is constructed using etched points as nodes and connecting edges as adjacency relationships.

[0070] Subsequently, the actual adjacency determination radius is determined based on the standard center distance and the upper limit of the material deformation rate of the battery tabs. Then, the connection relationship between each etching point in the etching point coordinate set is established according to the actual adjacency determination radius, and the actual mesh structure diagram is constructed.

[0071] Among them, the upper limit of material deformation rate is the maximum relative elongation that the metal foil of the battery tab can withstand before macroscopic tearing at high temperature. It is determined by the material property report or laboratory test report and is a physical parameter characterizing the deformation limit of the tab material.

[0072] This application can obtain the actual adjacency determination radius by amplifying it based on the standard center distance corresponding to the adjacency determination radius and combining it with the upper limit of the material deformation rate. For example, the actual adjacency determination radius satisfies... ,in, This is the actual adjacency determination radius. The adjacency determination radius corresponding to the standard center distance. This represents the upper limit of the material's deformation rate. This actual adjacency determination radius can accommodate the increased spacing between etched points caused by the thermal stretching of the electrode tabs after battery combustion, preventing the connection between etched points from breaking due to deformation. Thus, the three-dimensional spatial linear Euclidean distance between all pairs of etched points in the etched point coordinate set can be calculated using the method described above. Connection relationships are then established based on the actual adjacency determination radius, ultimately constructing the actual mesh structure.

[0073] Step 302: Perform fault-tolerant topology matching between the actual mesh structure diagram and each standard mesh structure diagram, and trace the target factory identification code from at least one candidate factory identification code based on the matching results.

[0074] Among them, fault-tolerant topology matching refers to subgraph isomorphic matching that allows for missing nodes, which is different from precise matching that requires the topology to be completely identical. This method is suitable for scenarios where some etched points on the electrode surface are missing due to melting after the battery burns. It can find the largest topological overlap area between the incomplete actual mesh structure diagram and the complete standard mesh structure diagram and establish the mapping relationship between nodes.

[0075] In some embodiments, the matching result includes etched points and edges that match the standard mesh structure diagram and the actual mesh structure diagram. Based on the matching result, candidate factory identification codes can be screened, and the candidate factory identification code with the highest matching degree with the actual mesh structure diagram can be determined as the target factory identification code.

[0076] For example, fault-tolerant topology matching can employ the Maximum Common Subgraph (MCS) matching algorithm or an equivalent fault-tolerant subgraph isomorphism algorithm to ensure that the matching results accurately reflect the degree of deviation between the actual structure and the theoretical structure.

[0077] Based on the above technical solution, this application can transform the etching point arrangement structure of the QR code into an intuitive topological network structure by constructing a standard grid structure diagram and an actual grid structure diagram. This achieves modular representation of the spatial structure of the etching points. By using a fault-tolerant topology matching method to compare the actual grid structure diagram with each standard grid structure diagram, it adapts to the actual scenario of missing etching points on the electrode surface after battery combustion. It can establish an effective node mapping relationship between the incomplete structure and the complete structure, accurately reflecting the matching situation between the actual structure and the theoretical structure. Thus, by selecting the target factory identification code based on the matching results, it achieves accurate comparison between the actual structure of the damaged electrode and the theoretical structure of the candidate factory identification code, improving the accuracy and reliability of structural comparison analysis.

[0078] As a possible embodiment of this application, step 302 above can be implemented through the following steps: Step 401: For each standard mesh structure diagram, perform fault-tolerant topology matching between the standard mesh structure diagram and the actual mesh structure diagram to determine the matching result of the standard mesh structure diagram.

[0079] The matching results include the etched points and edges that match the standard mesh structure diagram and the actual mesh structure diagram.

[0080] In some embodiments, when performing fault-tolerant topology matching on each standard mesh structure graph, this application uses the actual mesh structure graph as the query graph and the standard mesh structure graph as the target graph. A fault-tolerant subgraph isomorphism algorithm is used to find the largest common subgraph between the query and target graphs. Nodes in this common subgraph are the matching etch points, and edges in the common subgraph are the matching edges. Finally, the correspondence between each pair of matching etch points and edges is output, while retaining information on unmatched nodes and edges. This matching result can intuitively reflect the degree of overlap between the standard mesh structure graph and the actual mesh structure graph.

[0081] For example, this application first uses the currently constructed actual mesh structure diagram as the input query graph and the standard mesh structure diagram corresponding to the currently processed candidate factory identification code as the input target graph. Through the depth-first search mechanism, it reads a certain etch point in the actual mesh structure diagram and counts the connectivity of the etch point (i.e. the number of directly connected edges around it) and the secondary connection of these neighboring etch points.

[0082] Next, the algorithm iterates through the standard mesh structure graph to find candidate standard node clusters with similar topologies. Fault-tolerant topology matching allows skipping isolated etched points missing in the actual mesh structure graph, focusing instead on the largest connected regions that can overlap between the two. When the algorithm, under the fault-tolerant rules, identifies a local network structure in the actual mesh structure graph that has the maximum number of node correspondences with a theoretical local network structure in the standard mesh structure graph, it locks the one-to-one correspondence between each etched point in these two network clusters, outputting a candidate mapping set containing all the aforementioned pairwise etched point associations. Thus, by sequentially performing fault-tolerant matching between the actual mesh structure graph and each standard mesh structure graph, the matching result for each standard mesh structure graph can be obtained.

[0083] Step 402: Determine the number of matching etch points and the deformation ratio of each matching edge based on the matching results, and determine the target factory identification code from at least one candidate factory identification code based on the number of matching etch points and the deformation ratio of each matching edge.

[0084] The number of matching etching points is the total number of matching etching points in the matching results of the corresponding standard mesh structure diagram. The larger the value, the higher the degree of topological overlap between the standard mesh structure diagram and the actual mesh structure diagram, and the higher the matching degree of the candidate factory identification code.

[0085] The deformation ratio of each matched edge is the relative deviation ratio between the actual length of the matched edge in the actual mesh structure diagram and the standard length of the corresponding edge in the standard mesh structure diagram. This ratio quantifies the structural deviation caused by the deformation of the tabs on each connecting edge. The smaller the deformation ratio, the smaller the deviation between the actual and theoretical structures, and the higher the matching degree of the candidate factory identification code. When selecting target factory identification codes, this application can comprehensively consider the number of matching etch points and the deformation ratio of the edges, prioritizing candidate factory identification codes with a large number of etch points and a small deformation ratio as the target factory identification code.

[0086] For example, the deformation ratio satisfies the following formula: in, For the first The deformation ratio of the strip edge, For the first The actual length of the connecting line of the edge can be determined from the coordinate set of the etched points. The three-dimensional absolute coordinates of the two etched points connected by the strip edge were calculated. For the first The standard line length of a strip can be determined based on the first element in the standard coordinate set. The three-dimensional absolute coordinates of the two etched points connected by the strip edge are calculated. It is a safety parameter used to correct fractions where the denominator is 0, and its dimensions are the same as... The same applies; the specific value can be determined based on... The value of the value determines the outcome, such as .

[0087] The larger the value, the more severe the spatial stretching or compression distortion the metal foil must withstand if the edge is taken as the actual corresponding connection, and the higher the probability that the corresponding connection is false.

[0088] In one possible implementation, this application can determine the maximum deformation ratio of each standard mesh structure diagram based on the matching results.

[0089] The maximum deformation ratio is the maximum value of the deformation ratio of the matching edges in the standard mesh structure diagram and the actual mesh structure diagram. The maximum deformation ratio characterizes the limit of structural deviation that the standard mesh corresponding to the candidate identification code needs to withstand when matching with the actual mesh. Its value can intuitively reflect the overall deviation between the candidate identification code and the damaged electrode tab. The smaller the maximum deformation ratio, the smaller the overall deviation between the theoretical structure of the candidate identification code and the actual structure of the damaged electrode tab, and the higher the matching degree. This application can first traverse all matching edges in the standard mesh structure diagram and the actual mesh structure diagram, calculate the deformation ratio of each edge, and then extract the maximum value from all deformation ratios as the maximum deformation ratio of the standard mesh structure diagram.

[0090] If there is a minimum maximum deformation ratio among the maximum deformation ratios of each standard mesh structure diagram, the candidate factory identification code corresponding to the minimum maximum deformation ratio shall be used as the target factory identification code.

[0091] If there are multiple minimum maximum deformation ratios in each standard mesh structure diagram, the candidate factory identification code with the most matching etching points in the standard mesh structure diagram corresponding to the multiple minimum maximum deformation ratios is taken as the target factory identification code.

[0092] When a unique minimum value exists among the maximum deformation ratios corresponding to all candidate factory identification codes, it indicates that the theoretical structure of the candidate identification code corresponding to this minimum value has the smallest overall deviation from the actual structure of the damaged electrode, and the highest matching degree. Therefore, this application can directly determine it as the target factory identification code. This rule ensures the objectivity and uniqueness of the screening results. For multiple candidate factory identification codes with the same maximum deformation ratio, all of which are minimum values, a secondary screening can be performed based on the number of matching etching points. The more matching etching points, the higher the topological overlap between the theoretical structure of the candidate identification code and the actual structure of the damaged electrode, and the higher the matching degree. Therefore, the candidate factory identification code with the largest number of matching etching points is determined as the target factory identification code. This rule solves the screening problem of multiple candidate identification codes with the same limit deviation, ensuring the unique determination of the target factory identification code.

[0093] In some embodiments, after determining the maximum deformation ratio of each standard mesh structure diagram, this application may also perform preliminary screening based on the upper limit of material deformation rate.

[0094] This application can filter standard mesh structure diagrams whose maximum deformation ratio is less than or equal to the upper limit of the material deformation rate of the battery tab in each standard mesh structure diagram. If there is no standard mesh structure diagram in each standard mesh structure diagram whose maximum deformation ratio is less than or equal to the upper limit of the material deformation rate of the battery tab, then the current traceability is determined to be abnormal.

[0095] The upper limit of material deformation rate refers to the elongation at break of the tab metal foil at high temperature. This is an inherent physical parameter of the material, representing the maximum relative deformation ratio that the tab material can withstand before macroscopic tearing. If the maximum deformation ratio of a certain standard grid structure diagram exceeds the upper limit of the material deformation rate, it indicates that matching the theoretical structure of the candidate identification code with the actual structure of the damaged tab requires the tab material to withstand a structural deviation exceeding its deformation limit. This contradicts actual physical laws, therefore, the candidate identification code is more likely to be a false match and should be excluded. By comparing the maximum deformation ratio with the upper limit of the material deformation rate, this application achieves candidate object screening based on physical laws, ensuring that the screened candidate identification codes all conform to the deformation characteristics of the tab material.

[0096] However, after traversing all standard grid structure diagrams, if there is no standard grid structure diagram with a maximum deformation ratio less than or equal to the upper limit of the material deformation rate of the battery tab, that is, all candidate objects are eliminated because the deformation exceeds the fracture threshold, resulting in an empty screening set, it indicates that the remains of the damaged tab have undergone secondary crushing damage, or that the initially extracted character sequence has a fatal defect, causing all candidate factory identification codes to be unable to be reasonably matched with the actual structure of the damaged tab. If the screening continues at this time, the traceability result will not conform to the physical law. Therefore, this application can directly determine the traceability anomaly and stop the subsequent traceability process, while outputting anomaly prompt information to the human-computer interaction interface to facilitate staff to troubleshoot the problem in a timely manner.

[0097] Based on the above technical solution, this application can perform fault-tolerant topology matching for each standard mesh structure diagram and determine the matching result. It accurately extracts the matching etch points and edges between the actual mesh and the standard mesh. Then, based on the matching result, it determines the number of matching etch points and the deformation ratio of each edge, converting the matching degree of the mesh structure into a quantifiable numerical index. This avoids the error of subjective judgment in traditional matching and realizes the objective screening of candidate factory identification codes. Finally, it determines the target factory identification code by comprehensively considering the number of matching etch points and the deformation ratio of the edges. The screening is carried out from two dimensions: topological overlap and structural deviation, which improves the accuracy of target identification code screening and ensures that the selected candidate identification codes are highly matched with the actual structure of the damaged electrode.

[0098] Please see Figure 2 The diagram illustrates a system architecture of a battery combustion tracing system based on heterogeneous dual codes according to an embodiment of the present invention. The battery combustion tracing system 20 based on heterogeneous dual codes includes: The feature extraction module 21 is used to acquire a microscopic depth image of the surface of the battery tab, and perform image depth analysis based on the microscopic depth image to determine the character sequence corresponding to the plain text character distribution area on the surface of the battery tab and the set of etching point coordinates corresponding to the QR code distribution area; the set of etching point coordinates includes the three-dimensional absolute coordinates of each etching point within the QR code distribution area. The candidate matching unit 22 is used to perform database traversal matching based on the character sequence to obtain at least one candidate factory identification code and the QR code bitmap corresponding to each candidate factory identification code. The traceability verification unit 23 is used to perform structural comparison analysis based on the set of etched point coordinates and the QR code digital bitmaps corresponding to at least one candidate factory identification code, and to trace the target factory identification code from at least one candidate factory identification code.

[0099] It should be noted that the various embodiments of this application can be referenced or learned from each other. For example, the same or similar steps, method embodiments, system embodiments and device embodiments can be referenced from each other without limitation.

[0100] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0101] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for tracing battery combustion based on heterogeneous dual-code, characterized in that, include: A microscopic depth image of the battery tab surface is acquired, and image depth analysis is performed based on the microscopic depth image to determine the character sequence corresponding to the plaintext character distribution area on the battery tab surface and the set of etching point coordinates corresponding to the QR code distribution area; the set of etching point coordinates includes the three-dimensional absolute coordinates of each etching point within the QR code distribution area. Based on the character sequence, a database traversal and matching are performed to obtain at least one candidate factory identification code and the QR code bitmap corresponding to the at least one candidate factory identification code. Based on the set of etched point coordinates, a structural comparison analysis is performed on the QR code bitmaps corresponding to the at least one candidate factory identification code, and the target factory identification code is obtained by tracing the at least one candidate factory identification code. Based on the set of etched point coordinates, a structural comparison analysis is performed on the QR code bitmaps corresponding to the at least one candidate factory identification code, and the target factory identification code is obtained by tracing back from the at least one candidate factory identification code, including: A standard grid structure diagram is constructed based on each QR code digital bitmap, and an actual grid structure diagram is constructed based on the set of etched point coordinates. The standard grid structure diagram is used to represent each etched point in the QR code digital bitmap and the edges that are connected to each etched point. The actual grid structure diagram is used to represent each etched point in the set of etched point coordinates and the edges that are connected to each etched point. The actual mesh structure diagram is matched with each standard mesh structure diagram in a fault-tolerant topology, and the target factory identification code is obtained by tracing back from the at least one candidate factory identification code based on the matching result.

2. The battery combustion tracing method based on heterogeneous dual codes according to claim 1, characterized in that, A microscopic depth image of the battery tab surface is acquired, and image depth analysis is performed based on the microscopic depth image to determine the character sequence corresponding to the plaintext character distribution area and the set of etching point coordinates corresponding to the QR code distribution area on the battery tab surface, including: The battery tabs are scanned under multi-angle illumination to acquire shadow image sequences under different illumination directions; The relative height field of the battery tab surface is constructed based on the shadow image sequence, and the etching foreground binary image is extracted based on the relative height field; the relative height field is used to characterize the depth value of each pixel on the battery tab surface relative to the camera field of view reference plane; Based on the etching foreground binary image, determine the character sequence corresponding to the plaintext character distribution area and the set of etching point coordinates corresponding to the QR code distribution area.

3. The battery combustion tracing method based on heterogeneous dual codes according to claim 2, characterized in that, Based on the etched foreground binary image, determine the character sequence corresponding to the plaintext character distribution area and the set of etched point coordinates corresponding to the QR code distribution area, including: Optical character recognition is performed on the plaintext character distribution area in the etched foreground binary image, and the characters that fail to be recognized are mapped as placeholders to generate the character sequence corresponding to the plaintext character distribution area. Geometric centroids are extracted from each etching point in the QR code distribution area of ​​the etching foreground binary image, and the three-dimensional absolute coordinates of each etching point are determined by combining the camera intrinsic parameter matrix to generate the etching point coordinate set.

4. The battery combustion tracing method based on heterogeneous dual codes according to claim 1, characterized in that, Based on the character sequence, a database traversal and matching process is performed to obtain at least one candidate factory identification code and the corresponding QR code bitmap for each of the at least one candidate factory identification code, including: A regular expression query instruction is generated based on the character sequence; the regular expression query instruction is used to indicate the query for factory identification codes that match the character order and arbitrary length character spacing in the character sequence. According to the regular expression query instruction, the database is traversed and matched to filter out at least one candidate factory identification code from the database. For each candidate factory identification code, the encoder converts the candidate factory identification code into a corresponding QR code digital bitmap.

5. The battery combustion tracing method based on heterogeneous dual codes according to claim 4, characterized in that, Each standard mesh structure diagram is constructed based on the bitmap of the QR code, and an actual mesh structure diagram is constructed based on the set of etched point coordinates, including: Coordinate mapping is performed on the QR code bitmap corresponding to each candidate factory identification code to generate a standard coordinate set with the same scale as the etched point coordinate set. Based on the standard center distance of the factory identification code, establish the connection relationship between each etching point in each standard coordinate set and construct the corresponding standard mesh structure diagram. The actual adjacency determination radius is determined based on the standard center distance and the upper limit of the material deformation rate of the battery tab, and the connection relationship between each etching point in the etching point coordinate set is established according to the actual adjacency determination radius to construct the actual mesh structure diagram.

6. The battery combustion tracing method based on heterogeneous dual codes according to claim 1, characterized in that, Perform fault-tolerant topology matching between the actual mesh structure diagram and each standard mesh structure diagram, and obtain the target factory identification code from the at least one candidate factory identification code based on the matching result, including: For each standard mesh structure diagram, a fault-tolerant topology matching is performed between the standard mesh structure diagram and the actual mesh structure diagram to determine the matching result of the standard mesh structure diagram; the matching result includes the etch points and edges that match in the corresponding standard mesh structure diagram and the actual mesh structure diagram; The number of matching etch points and the deformation ratio of each matching edge are determined based on the matching results, and the target factory identification code is determined from the at least one candidate factory identification code based on the number of matching etch points and the deformation ratio of each matching edge.

7. The battery combustion tracing method based on heterogeneous dual codes according to claim 6, characterized in that, Based on the matching results, the number of matching etching points and the deformation ratio of each matching edge are determined, and the target factory identification code is determined from the at least one candidate factory identification code based on the number of matching etching points and the deformation ratio of each matching edge, including: The maximum deformation ratio of each standard mesh structure diagram is determined based on the matching results; the maximum deformation ratio is the maximum value of the deformation ratio of the matching edges between the standard mesh structure diagram and the actual mesh structure diagram. If there is a minimum maximum deformation ratio among the maximum deformation ratios of each standard grid structure diagram, the candidate factory identification code corresponding to the minimum maximum deformation ratio shall be used as the target factory identification code. If there are multiple minimum maximum deformation ratios among the maximum deformation ratios of each standard mesh structure diagram, the candidate factory identification code with the largest number of matching etching points in the standard mesh structure diagram corresponding to the multiple minimum maximum deformation ratios shall be used as the target factory identification code.

8. The battery combustion traceability method based on heterogeneous dual codes according to claim 7, characterized in that, After determining the maximum deformation ratio for each standard mesh structure diagram, the method further includes: From each of the standard mesh structure diagrams, select the standard mesh structure diagram whose maximum deformation ratio is less than or equal to the upper limit of the material deformation rate of the battery tab; If there is no standard mesh structure diagram in each of the standard mesh structure diagrams whose maximum deformation ratio is less than or equal to the upper limit of the material deformation rate of the battery tab, then the current traceability is determined to be abnormal.

9. A battery combustion traceability system based on heterogeneous dual-code, characterized in that, include: The feature extraction module is used to acquire a microscopic depth image of the surface of the battery tab, and perform image depth analysis based on the microscopic depth image to determine the character sequence corresponding to the plaintext character distribution area on the surface of the battery tab and the set of etching point coordinates corresponding to the QR code distribution area; the set of etching point coordinates includes the three-dimensional absolute coordinates of each etching point in the QR code distribution area. The candidate matching unit is used to perform database traversal matching based on the character sequence to obtain at least one candidate factory identification code and the QR code bitmap corresponding to the at least one candidate factory identification code. The traceability verification unit is used to perform structural comparison and analysis on the set of etched point coordinates and the QR code digital bitmaps corresponding to the at least one candidate factory identification code, and to trace the target factory identification code from the at least one candidate factory identification code. Based on the set of etched point coordinates, a structural comparison analysis is performed on the QR code bitmaps corresponding to the at least one candidate factory identification code, and the target factory identification code is obtained by tracing back from the at least one candidate factory identification code, including: A standard grid structure diagram is constructed based on each QR code digital bitmap, and an actual grid structure diagram is constructed based on the set of etched point coordinates. The standard grid structure diagram is used to represent each etched point in the QR code digital bitmap and the edges that are connected to each etched point. The actual grid structure diagram is used to represent each etched point in the set of etched point coordinates and the edges that are connected to each etched point. The actual mesh structure diagram is matched with each standard mesh structure diagram in a fault-tolerant topology, and the target factory identification code is obtained by tracing back from the at least one candidate factory identification code based on the matching result.

Images