Welding defect detection method and apparatus, and storage medium and program product

Abstract

The present application relates to the field of batteries. Provided are a welding defect detection method and apparatus, and a storage medium and a program product. The method comprises: obtaining a welding-seam area and a non-welding-seam area by means of a first image having structure information; obtaining a plurality of detection areas of a welding seam on the basis of the welding-seam area and the non-welding-seam area; and when preset attribute information of at least one detection area among the plurality of detection areas of the welding seam meets a preset condition, determining that there is a defect in the welding seam.

Description

Welding defect detection method, device, storage medium and program product

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This disclosure is based on the Chinese patent application with application number 202311543281.0, application date November 20, 2023, and application name “Welding Defect Detection Method, Device and Storage Medium”, and claims the priority of the Chinese patent application. The entire content of the Chinese patent application is hereby introduced into this disclosure as a reference. Technical Field

[0003] The present application relates to, but is not limited to, the field of batteries, and in particular to a welding defect detection method, device, storage medium, and program product. Background Art

[0004] During the production of power lithium batteries, the top cover welds can exhibit a certain amount of defects due to the different material compositions of the top cover and aluminum shell, as well as the divergence of the welding laser. To ensure production yield, delivery rate, and battery cell safety, top cover weld inspection has become an essential step.

[0005] In related technologies, the detection of weld defects is achieved by directly locating the defect position and outputting the results, but the misjudgment rate is high.

[0006] The above statements are only used to provide background information related to the present application and do not necessarily constitute prior art.

[0007] Summary of the Invention

[0008] A technical problem to be solved by the present application is to provide a welding defect detection method, device, storage medium and program product that can reduce the misjudgment rate of weld defect detection.

[0009] In a first aspect, the present application provides a welding defect detection method, comprising: obtaining a first image of the welding between a battery top cover and a shell, the first image including structural information; detecting the weld area in the first image to obtain a weld area and a non-weld area; determining the effective width and center width of the weld area in the weld width direction; determining, based on the effective width, a first predetermined distance and a second predetermined distance for the weld area to extend toward the non-weld area along the weld width direction; dividing the weld between the battery top cover and the shell based on the first predetermined distance, the second predetermined distance and the center width to obtain multiple detection areas; determining that a defect exists in the weld when the preset attribute information of at least one detection area among the multiple detection areas of the weld meets a preset condition.

[0010] In the technical solution of the embodiment of the present application, the image is divided into regions, the weld area and non-weld area of ​​interest are clarified, and the weld structure is then refined to obtain multiple detection areas, which facilitates subsequent independent judgment of defects in each area and improves the accuracy of weld defect identification. When the preset attribute information of at least one detection area meets the preset conditions, it is determined that the weld has defects. Through regional analysis and specification control, the misjudgment rate of weld defects can be reduced while achieving effective interception of weld defects, and the detection accuracy can be improved, thereby increasing the battery yield and shipment volume.

[0011] In some embodiments, determining the effective width of the weld region in the weld bead width direction includes: determining a first edge position and a second edge position of the weld region in the weld bead width direction; and obtaining the effective width of the weld region in the weld bead width direction based on the difference between the second edge position and the first edge position. Determining the effective width of the weld region facilitates the division of the weld structure.

[0012] In some embodiments, the multiple inspection areas include an upper cover weld area, a center weld area, and a side cover weld area, wherein the center weld area is located between the upper cover weld area and the side cover weld area along the weld width direction, wherein the weld between the battery top cover and the shell is divided into regions based on the first predetermined distance, the second predetermined distance, and the center width, and obtaining the multiple inspection areas includes: determining a third edge position and a fourth edge position of the center weld area in the weld width direction according to the first edge position and the second edge position of the weld area in the weld width direction, as well as the effective width and the center width; The first edge position is extended along the weld width direction by a first predetermined distance toward the non-weld area away from the center weld area to obtain the fifth edge position of the upper cover weld area in the weld width direction, wherein the third edge position is the edge position of the upper cover weld area close to the center weld area in the weld width direction; and the second edge position is extended along the weld width direction by a second predetermined distance toward the non-weld area away from the center weld area to obtain the sixth edge position of the side cover weld area in the weld width direction, wherein the fourth edge position is the edge position of the side cover weld area close to the center weld area in the weld width direction. Dividing the weld into the upper cover weld area, the center weld area, and the side cover weld area facilitates the production line to develop independent defect measurement standards for different weld areas, thereby reducing the misjudgment rate of weld defects.

[0013] In some embodiments, based on the first and second edge positions of the weld area in the weld bead width direction, as well as the effective width and the center width, determining the third and fourth edge positions of the central weld area in the weld bead width direction includes: determining the third and fourth edge positions of the central weld area in the weld bead width direction includes: moving the first edge position toward the second edge position by a third predetermined distance to obtain the third edge position, wherein the third predetermined distance is half of the difference between the effective width and the center width; and moving the second edge position toward the first edge position by the third predetermined distance to obtain the fourth edge position. Identifying the central weld area and determining the edge position of the central weld area can improve the accuracy of weld segmentation.

[0014] In some embodiments, the multiple inspection areas further include a first inspection area and a second inspection area, wherein the first inspection area is located along the weld bead extension direction, on a first side of the upper roof weld area, the center weld bead area, and the side roof weld area, and the second inspection area is located along the weld bead extension direction, on a second side of the upper roof weld area, the center weld bead area, and the side roof weld area. Because the weld has a curvature, the R-angle area of ​​the weld can be identified, facilitating independent determination of defects in the R-angle area.

[0015] In some embodiments, obtaining the multiple inspection areas further includes: extending a seventh edge position of the weld area in the weld bead extension direction by a fourth predetermined distance toward the upper cover weld area, the center weld bead area, and the side cover weld area to obtain an eighth edge position of the first inspection area, wherein the seventh edge position is an edge position of the first inspection area away from the upper cover weld area, the center weld bead area, and the side cover weld area. By identifying the edge position of the first inspection area, the accuracy of weld delineation is improved.

[0016] In some embodiments, obtaining the multiple inspection areas further includes: extending a ninth edge position of the weld area in the weld bead extension direction by a fifth predetermined distance toward the upper cover weld area, the center weld bead area, and the side cover weld area to obtain a tenth edge position of the second inspection area, wherein the ninth edge position is an edge position of the second inspection area away from the upper cover weld area, the center weld bead area, and the side cover weld area. By identifying the edge position of the second inspection area, the accuracy of weld delineation is improved.

[0017] In some embodiments, the method further includes: obtaining at least one first detection frame containing a detection target based on the first image, wherein the first detection frame has a category; obtaining a detection frame to be identified based on the at least one first detection frame; determining a detection area corresponding to the detection frame to be identified based on the position of the detection frame to be identified in the weld; and determining at least one attribute information of the detection target in the detection frame to be identified, wherein if the preset attribute information of at least one detection area in the multiple detection areas of the weld meets a preset condition, determining that the weld is defective includes: determining that the weld is defective if at least one attribute information of the detection target in at least one detection frame to be identified in at least one detection area is greater than or equal to a measurement threshold corresponding to the category of the detection frame to be identified and the detection area where the detection target is located. By locating the detection frames and determining the detection area where each detection frame is located, it is convenient to detect the target in each detection frame according to the standard of the detection area, thereby identifying whether the target is a defect, and then determining whether the weld is defective, thereby improving detection accuracy and efficiency.

[0018] In some embodiments, the method further includes: obtaining a second image of the weld between the battery top cover and the battery shell before determining the detection area corresponding to the detection frame to be identified based on the position of the detection frame to be identified at the weld, wherein the second image includes brightness information; obtaining at least one second detection frame containing the detection target based on the second image, wherein the second detection frame has a category, wherein obtaining the detection frame to be identified based on at least one first detection frame includes: obtaining the detection frame to be identified based on at least one first detection frame and at least one second detection frame. By comprehensively analyzing images with brightness information and images with structural information, the probability of missed detection of the detection frame can be reduced.

[0019] In some embodiments, at least one attribute information includes a measured distance. Determining at least one attribute information of a target within a detection frame to be identified includes: using a cross-section of an area adjacent to the detection frame, but not within the detection frame to be identified, as a measurement reference plane for the detection frame to be identified; and determining the measured distance of the target relative to the measurement reference plane within the detection frame to be identified. By finding a more accurate measurement reference point, the true measurement value of the defect can be accurately and effectively reflected, thereby effectively improving the accuracy of the specification measurement value.

[0020] In some embodiments, at least one attribute information includes a measurement distance, and the multiple detection areas include an upper cover weld area, a central weld area, and a side cover weld area. The central weld area is located between the upper cover weld area and the side cover weld area along the weld width direction. wherein, determining at least one attribute information of the detection target in the detection frame to be identified includes: when the detection frame to be identified where the detection target is located is located in the upper cover weld area, based on the pixel distribution difference in the first image, a cross section of the non-weld area closest to the upper cover weld area along the weld width direction is used as the measurement distance of the detection frame to be identified where the detection target is located. The measurement reference surface is used; when the detection frame to be identified where the detection target is located is located in the side top cover weld area, based on the pixel distribution difference in the first image, the cross section of the non-weld area closest to the side top cover weld area along the weld width direction is used as the measurement reference surface of the detection frame to be identified where the detection target is located; when the detection frame to be identified where the detection target is located is located in the central weld area, the cross section of the area in the central weld area excluding the detection target is used as the measurement reference surface of the detection frame to be identified where the detection target is located; and the measurement distance of the detection target relative to the measurement reference surface of the detection frame to be identified where the detection target is located is determined. For different detection areas, selecting more accurate measurement reference points can truly and effectively reflect the actual measurement value of the defect, thereby effectively improving the accuracy of the specification measurement value.

[0021] In some embodiments, the plurality of inspection areas further include a first inspection area and a second inspection area, wherein the first inspection area is located along the weld extension direction at a first side of the upper top cover weld area, the center weld area, and the side top cover weld area, and the second inspection area is located along the weld extension direction at a second side of the upper top cover weld area, the center weld area, and the side top cover weld area, wherein determining at least one attribute information of the inspection target in the to-be-identified inspection frame includes: when the to-be-identified inspection frame where the inspection target is located is located in the first inspection area, using the cross-section of the area in the first inspection area excluding the inspection target as the measurement reference surface of the to-be-identified inspection frame where the inspection target is located; when the to-be-identified inspection frame where the inspection target is located is located in the second inspection area, using the cross-section of the area in the second inspection area excluding the inspection target as the measurement reference surface of the to-be-identified inspection frame where the inspection target is located; and determining the measurement distance of the inspection target relative to the measurement reference surface of the to-be-identified inspection frame where the inspection target is located. For the target located in the R angle area, selecting a corresponding measurement reference point improves the actual measurement value of the defect in the area, thereby improving the measurement accuracy.

[0022] In some embodiments, the at least one attribute information also includes area information. Determining the at least one attribute information of the detection target in the detection frame to be identified further includes: determining the pixel corresponding to the detection target based on the depth information of the pixel in the first image; and determining the area information of the detection target based on the pixel corresponding to the detection target in the first image. By detecting the area information of each detection target, it is convenient to compare the area information with a threshold, and to determine defects based on the area size.

[0023] In some embodiments, obtaining at least one first detection frame containing a detection target includes: acquiring at least one first suspected detection frame in a first image, wherein the first suspected detection frame has a category and a probability confidence level; and filtering the at least one first suspected detection frame based on a probability confidence level threshold to obtain the at least one first detection frame. By filtering out invalid detection frames, the accuracy of subsequent defect detection can be improved.

[0024] In some embodiments, obtaining at least one second detection frame containing a detection target includes: acquiring at least one second suspected detection frame in a second image, wherein the second suspected detection frame has a category and a probability confidence level; and filtering the at least one second suspected detection frame based on a probability confidence level threshold to obtain the at least one second detection frame. By filtering out invalid detection frames, the accuracy of subsequent defect detection can be improved.

[0025] In some embodiments, a measurement threshold for the attribute information of the detection target within the detection frame to be identified is set based on the category and detection area of ​​the detection frame to be identified. By setting a corresponding measurement threshold for each detection frame based on its category and detection area, the accuracy of defect identification within each detection frame can be improved.

[0026] In a second aspect, a welding defect detection device includes: an image acquisition module, configured to acquire a first image of the welding between a battery top cover and a shell, the first image including structural information; a first division module, configured to detect the weld area in the first image to obtain a weld area and a non-weld area; a second division module, configured to confirm multiple detection areas of the weld between the battery top cover and the shell based on the weld area and the non-weld area; and a defect identification module, configured to determine that there is a defect in the weld when preset attribute information of at least one detection area in the multiple detection areas of the weld meets preset conditions.

[0027] In the technical solution of the embodiment of the present application, the image is divided into regions to identify the weld area and non-weld area of ​​interest, and then the weld structure is refined to obtain multiple detection areas, which facilitates independent judgment of defects in the detection areas. When the preset attribute information of at least one detection area meets the preset conditions, it is determined that the weld has defects. Through regional analysis and specification control, the misjudgment rate of weld defects can be reduced while achieving effective interception of weld defects, thereby improving the battery yield and shipment volume.

[0028] In a third aspect, a welding defect detection device includes: a memory; and a processor coupled to the memory, wherein the processor is configured to execute the welding defect detection method as described above based on instructions stored in the memory.

[0029] In a fourth aspect, a computer-readable storage medium stores computer program instructions, which, when executed by a processor, implement the welding defect detection method as described above.

[0030] In a fifth aspect, a computer program product is provided, which includes a computer program or instructions. When the computer program or instructions are run on a computer, the computer is caused to execute the welding defect detection method as described above.

[0031] Other features and advantages of the present application will become apparent from the following detailed description of exemplary embodiments of the present application with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the drawings without creative work.

[0033] FIG1 is a flow chart of a welding defect detection method according to one or more embodiments;

[0034] FIG2 is a schematic diagram illustrating the relative positions of a camera and a battery cell according to one or more embodiments;

[0035] FIG3 is a second flow chart of a welding defect detection method according to one or more embodiments;

[0036] FIG4 is an example diagram of weld structured partitioning according to one or more embodiments;

[0037] FIG5 is a third flow chart of a welding defect detection method according to one or more embodiments;

[0038] FIG6 is a schematic diagram of defect attribute detection according to one or more embodiments;

[0039] FIG7 is a fourth flow chart of a welding defect detection method according to one or more embodiments;

[0040] FIG8 is a first structural diagram of a welding defect detection device according to one or more embodiments;

[0041] FIG9 is a second structural diagram of a welding defect detection device according to one or more embodiments;

[0042] FIG10 is a third structural diagram of a welding defect detection device according to one or more embodiments. DETAILED DESCRIPTION

[0043] The following detailed description of the embodiments of the present application is provided in conjunction with the accompanying drawings and examples. The following detailed description of the embodiments and the accompanying drawings are used to illustrate the principles of the present application, but are not intended to limit the scope of the present application, that is, the present application is not limited to the described embodiments.

[0044] In the description of this application, it should be noted that, unless otherwise specified, "multiple" means more than two; the terms "upper", "lower", "left", "right", "inside", "outside", etc., indicating directions or positional relationships, are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting this application. In addition, the terms "first", "second", "third", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. "Vertical" is not strictly perpendicular, but is within the allowable error range. "Parallel" is not strictly parallel, but is within the allowable error range.

[0045] At the same time, it should be understood that for the convenience of description, the sizes of the various parts shown in the drawings are not drawn according to the actual proportional relationship.

[0046] The following description of at least one exemplary embodiment is merely illustrative in nature and is in no way intended to limit the present disclosure, its application, or uses.

[0047] Technologies, methods, and equipment known to ordinary technicians in the relevant art may not be discussed in detail, but where appropriate, the technologies, methods, and equipment should be considered part of the specification.

[0048] In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not limiting. Therefore, other examples of the exemplary embodiments may have different values.

[0049] It should be noted that like reference numerals and letters refer to like items in the following figures, and therefore, once an item is defined in one figure, it need not be further discussed in subsequent figures.

[0050] In order to make the objectives, technical solutions and advantages of this application more clear, this application is further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.

[0051] In related technologies, after scanning the weld bead with a camera, defect features are directly extracted and an artificial intelligence (AI) model is built. During normal production, the AI ​​model directly locates and identifies defects. While this model can detect defects, the lack of regional and specification-based card control filtering leads to a high rate of misjudgment, which affects shipment volume.

[0052] In an embodiment of the present application, by identifying weld areas of interest and non-weld areas of no interest, and dividing the welds into sub-areas, each area can be inspected separately, and when the preset attribute information of at least one inspection area meets the preset conditions, it is determined that the weld has defects, which can reduce misjudgments, improve the accuracy of weld defect detection, and thereby improve battery yield and shipment volume.

[0053] Figure 1 is a flow chart of a welding defect detection method according to one or more embodiments. This embodiment is performed by a welding defect detection device.

[0054] In step 110 , a first image of the welding between the battery top cover and the shell is acquired, where the first image includes structural information.

[0055] In some embodiments, the first image is a depth image, which contains structural information, such as contour features and depth features. By analyzing the contour features, different areas in the image can be identified.

[0056] In some embodiments, as shown in FIG2 , an industrial camera 210 is deployed (two industrial cameras 210 are shown in FIG2 ). The industrial camera is, for example, a 3D (Three Dimensional) camera. The 3D camera emits a laser at a certain angle and uses the laser to scan the top cover weld of the battery cell 220 to obtain an image of the top cover weld. The 3D camera is, for example, a line scan camera. The laser emitted by the camera forms an angle of 20° to 40° with the axis of the battery cell, for example, 30°. The distance between the camera and the weld is 45mm to 75mm, for example, 60mm. Those skilled in the art should understand that the limitations on angles and distances here are only for example purposes, and the corresponding angles and distances can be set according to the specific circumstances of the production line. The 3D camera can simultaneously generate a grayscale image containing brightness information and a depth map containing structural information. The grayscale image has grayscale features. Digital modeling can be performed using the obtained images.

[0057] In step 120 , the weld area in the first image is detected to obtain a weld area and a non-weld area.

[0058] The laser light emitted by a 3D camera has a certain width. When scanning a roof weld, the image often contains a lot of redundant areas. Without structured segmentation, this will result in many false positives in areas of non-interest, impacting detection efficiency. In some embodiments, by analyzing the contour features in the depth image, it is possible to identify weld areas of interest and non-weld areas of non-interest.

[0059] In step 130 , a plurality of inspection areas of the weld between the battery top cover and the battery shell are identified based on the weld area and the non-weld area.

[0060] In some embodiments, the weld structure is refined and divided into an upper cover weld area, a side cover weld area, and a center weld area, so as to facilitate the production line to formulate independent defect measurement standards for different weld areas.

[0061] For example, the difference in aluminum thickness between the top cover and the side cover leads to different defect specification metrics. For example, the aluminum shell of the top cover is generally thicker than that of the side cover. Therefore, a pinhole defect of the same depth is more likely to occur in the weld of the side cover than in the weld of the top cover. Therefore, in this step, the weld is divided into multiple inspection areas.

[0062] In step 140 , when preset attribute information of at least one detection area among the multiple detection areas of the weld meets a preset condition, it is determined that the weld has a defect.

[0063] In some embodiments, for each inspection area, a measurement threshold for each detection attribute of the inspection target in that inspection area can be adaptively set based on the actual production process and defect specification standards. For example, the depth information, height information, or area information of the inspection target in each inspection area is measured, and the obtained data is compared with the attribute threshold of the inspection target in the inspection area to determine whether the inspection target in the inspection area is a defect. If at least one inspection target in the inspection area is a defect, the weld is determined to be defective. The inspection targets may include protrusions, pits, pinholes, cracks, polarization, broken welds, etc.

[0064] In the above embodiment, the first image containing structural information is used to identify the weld area of ​​interest and the non-weld area of ​​no interest, and further obtain multiple detection areas of the weld, so that the detection target of each detection area can be treated differently according to the process differences and the defect standards of different areas, and when the preset attribute information of at least one detection area meets the preset conditions, it is determined that the weld has defects, and the false positive rate is reduced while achieving effective interception of weld defects, thereby improving the accuracy of weld defect detection, thereby improving the yield rate and shipment volume.

[0065] FIG3 is a second flow chart of a welding defect detection method according to one or more embodiments. As shown in FIG3 , the above step 130 can be implemented through steps 310 to 330:

[0066] At step 310 , the effective width and center width of the weld region in the weld bead width direction are determined.

[0067] In some embodiments, a first edge position and a second edge position of the weld area in the weld width direction are determined; and an effective width of the weld area in the weld width direction is obtained based on the difference between the second edge position and the first edge position.

[0068] As shown in Figure 4 , the weld region is between lines 420 and 450, with line 450 representing the first edge of the weld region and line 420 representing the second edge of the weld region. If line 420 is the left starting point of the weld, with its coordinate in the weld bead width direction as L, and line 450 is the right ending point of the weld, with its coordinate in the weld bead width direction as R, then the effective width of the weld region, W, equals RL.

[0069] In some embodiments, the center width is determined based on the effective width of the weld region. For example, the center width is obtained by multiplying the effective width by a certain proportionality factor, where the proportionality factor ρ is set to 0 < ρ < 1, and the specific value depends on the degree of protrusion of the weld. In some embodiments, the center width can also be directly preset.

[0070] In step 320 , a first predetermined distance and a second predetermined distance for the weld region to expand toward the non-weld region along the weld width direction are determined based on the effective width.

[0071] In some embodiments, although the weld area has been identified through the image, in actual applications, weld defects may spread. Therefore, the weld area is expanded to the non-weld areas on the left and right sides of Figure 4. For example, line 420 is extended to the left by a second predetermined distance to obtain line 410, and line 450 is extended to the right by a first predetermined distance to obtain line 460. The area between lines 410 and 460 is the area to be analyzed. The first and second predetermined distances can be the same or different, and a value of 2 mm is taken as an example.

[0072] In step 330 , the weld is divided into regions based on the first predetermined distance, the second predetermined distance, and the center width to obtain a plurality of inspection regions.

[0073] In some embodiments, according to the difference in the thickness of the aluminum shell material of the upper top cover and the side top cover, the weld is divided into an upper top cover weld area, a center weld area and a side top cover weld area. The center weld area is located between the upper top cover weld area and the side top cover weld area along the weld width direction.

[0074] Compared with treating the weld as a whole without distinction, in the above embodiment, the weld is divided into multiple detection areas, which facilitates the subsequent setting of detection target thresholds for each detection area and the judgment of each detection target, which can realize weld defect judgment more accurately and flexibly.

[0075] In some embodiments, the third edge position and the fourth edge position of the central weld area in the weld width direction are determined based on the first edge position and the second edge position of the weld area in the weld width direction, the effective width, and the center width.

[0076] For example, the first edge position is moved toward the second edge position by a third predetermined distance to obtain a third edge position, wherein the third predetermined distance is half of the difference between the effective width and the center width; and the second edge position is moved toward the first edge position by a third predetermined distance to obtain a fourth edge position.

[0077] Taking Figure 4 as an example, line 450 represents the first edge of the weld area, and line 420 represents the second edge of the weld area. Moving line 450 to the left by a third predetermined distance yields line 440, the third edge. Moving line 420 to the right by a third predetermined distance yields line 430, the fourth edge. This determines the left starting position and right ending position of the center weld area.

[0078] In some embodiments, the coordinates of the left starting position of the central weld bead area in the weld bead width direction are L+(1-ρ)*W / 2, and the coordinates of the right ending position of the central weld bead area in the weld bead width direction are R-(1-ρ)*W / 2. (1-ρ)*W / 2 is a third predetermined distance. In some embodiments, the third predetermined distance can also be a preset value.

[0079] In some embodiments, the first edge position is extended along the weld width direction to the non-weld area away from the center weld area by a first predetermined distance to obtain the fifth edge position of the upper cover weld area in the weld width direction, wherein the third edge position is the edge position of the upper cover weld area close to the center weld area in the weld width direction.

[0080] Taking Figure 4 as an example, the weld extension width of the upper cover is set to determine the boundary between the upper cover weld area and the non-weld inspection area. Line 450 is extended to the right by a first predetermined distance to obtain line 460, the fifth edge position. This line 460 defines the boundary between the upper cover weld area and the non-weld inspection area, while line 440 is the boundary between the upper cover weld area and the center weld area.

[0081] In some embodiments, the first predetermined distance is determined based on the effective width of the weld. For example, the first predetermined distance is obtained by multiplying the effective width by a certain proportionality factor, where the proportionality factor α is set to 0 < α < 1, and the specific value depends on the extent of the weld. The coordinates of line 460 in the weld width direction are R + α * W. In some embodiments, the first predetermined distance can also be directly preset.

[0082] In some embodiments, the second edge position is extended along the weld width direction toward the non-weld area away from the center weld area by a second predetermined distance to obtain the sixth edge position of the side top cover weld area in the weld width direction, wherein the fourth edge position is the edge position of the side top cover weld area close to the center weld area in the weld width direction.

[0083] Taking Figure 4 as an example, the weld extension width of the side roof cover is set to determine the boundary between the side roof cover weld area and the non-weld inspection area. Line 420 is extended to the left by a second predetermined distance to obtain line 410, the sixth edge position. Line 410 defines the boundary between the side roof cover weld area and the non-weld inspection area. Line 430 delineates the boundary between the side roof cover weld area and the center weld area.

[0084] In some embodiments, the second predetermined distance is determined based on the effective width of the weld. For example, the second predetermined distance is obtained by multiplying the effective width by a certain proportionality factor, where the proportionality factor α is set to 0 < α < 1, and the specific value depends on the extent of the weld. The coordinates of line 410 in the weld width direction are L - α * W. In some embodiments, the first predetermined distance can also be directly preset.

[0085] In the above embodiment, the welding processes of the top cover and the side cover of the battery cell are different, and the weld area is divided into the top cover weld area, the center weld area and the side cover weld area, thereby realizing the structured partitioning of the weld, which is conducive to the production line to formulate independent defect measurement standards for different weld areas.

[0086] In some embodiments, the weld further includes a first inspection area and a second inspection area, wherein the first inspection area is located along the direction of extension of the weld bead on the first side of the upper cover weld bead area, the center weld bead area, and the side cover weld bead area, and the second inspection area is located along the direction of extension of the weld bead on the second side of the upper cover weld bead area, the center weld bead area, and the side cover weld bead area. The first inspection area and the second inspection area are R-angle areas, i.e., the connection position of two adjacent edges of the top cover. The size of the R-angle area is related to the angle between the laser and the battery cell. The smaller the horizontal angle between the laser and the battery cell, the longer the R-angle imaging area.

[0087] In some embodiments, the seventh edge position of the weld area in the weld extension direction is extended a fourth predetermined distance toward the upper cover weld area, the center weld area, and the side cover weld area to obtain the eighth edge position of the first inspection area, wherein the seventh edge position is the edge position of the first inspection area away from the upper cover weld area, the center weld area, and the side cover weld area.

[0088] Taking Figure 4 as an example, line 490 is moved downward by a fourth predetermined distance to obtain line 470, i.e., the eighth edge position. Line 490 corresponds to the seventh edge position. The fourth predetermined distance is, for example, 10 mm. The value of the fourth predetermined distance is related to the horizontal angle between the laser and the battery cell.

[0089] In some embodiments, the ninth edge position of the weld area in the weld extension direction is extended a fifth predetermined distance toward the upper cover weld area, the center weld area, and the side cover weld area to obtain the tenth edge position of the second inspection area, wherein the ninth edge position is the edge position of the second inspection area away from the upper cover weld area, the center weld area, and the side cover weld area.

[0090] Taking Figure 4 as an example, line 4100 is moved upward by the fifth predetermined distance to obtain line 480, i.e., the tenth edge position. Line 4100 corresponds to the ninth edge position. The fifth predetermined distance is, for example, 10 mm, and its value is related to the horizontal angle between the laser and the battery cell.

[0091] In the above embodiment, the weld is not only divided into the upper cover weld area, the center weld area and the side cover weld area, but also the R angle area can be divided, achieving a more refined division, thereby further improving the accuracy of subsequent detection.

[0092] FIG5 is a third flow chart of a welding defect detection method according to one or more embodiments.

[0093] In step 510, at least one first detection frame containing a detection target is obtained according to the first image.

[0094] Among them, the first detection box has a category.

[0095] In some embodiments, the first image is input into a fully trained deep learning algorithm model for defect detection. The deep learning algorithm model is, for example, an MMDetection model based on target detection. Compared with other detection modules, this model covers more state-of-the-art methods in the field of target detection and can achieve very high detection accuracy. The output of the model is a series of detection boxes with category information and probability confidence. Category information includes, for example, protrusion category, pit category, pinhole category, explosion point category, polarization category, broken weld category, etc. This application does not limit the model training process.

[0096] In some embodiments, at least one first suspected detection frame is obtained in the first image, wherein the first suspected detection frame has a category and a probability confidence; based on a probability confidence threshold, the at least one first suspected detection frame is filtered to obtain at least one first detection frame.

[0097] For example, applying a probability confidence threshold control to each suspected defect frame can filter out some false positive detection frames, thereby improving subsequent detection efficiency and accuracy.

[0098] In step 520, a detection frame to be identified is obtained according to at least one first detection frame.

[0099] For example, if the model outputs 5 detection boxes, 5 detection boxes to be identified are obtained.

[0100] In step 530 , a detection area corresponding to the detection frame to be identified is determined according to the position of the detection frame to be identified on the weld.

[0101] For example, it is identified whether the detection frame is located in the upper cover weld area, the center weld area, the side cover weld area or the R corner area.

[0102] In step 540 , at least one attribute information of the detection target in the detection frame to be identified is determined.

[0103] In some embodiments, the attribute information includes, for example, at least one of the following: a depth attribute, a height attribute, an area attribute, and the like.

[0104] In step 550, when at least one attribute information of the detection target of at least one to-be-identified detection frame of at least one detection area is greater than or equal to the measurement threshold corresponding to the category of the to-be-identified detection frame where the detection target is located and the detection area, it is determined that there is a defect in the weld.

[0105] In some embodiments, a measurement threshold for each attribute information of the detection target in the detection frame to be identified is set according to the category and detection area to which the detection frame to be identified belongs. That is, the measurement threshold is related to the defect category and the defect detection area. For example, for pinholes, the measurement thresholds taken for the pinholes located in the upper top cover weld area and the pinholes located in the side top cover weld area are different. For another example, different measurement thresholds can be set for pinholes and protrusions respectively. For another example, different measurement thresholds are set for pinholes located in the upper top cover weld area, pinholes located in the side top cover weld area, protrusions located in the upper top cover weld area, and protrusions located in the side top cover weld area. This can improve the accuracy of defect recognition in each detection frame.

[0106] In the above embodiment, by identifying each detection frame and its category, as well as the attribute information of the detection target within each detection frame and the detection area within each detection frame, a weld defect is determined if the attribute information of at least one detection target exceeds the measurement threshold corresponding to the category and detection area of ​​the detection frame within which the detection target is located. Combining visual texture information with defect structure information makes the detection results more accurate and can improve the yield rate of batteries.

[0107] In other embodiments of the present application, a second image of the battery top cover and the shell being welded is obtained, wherein the second image includes brightness information; based on the second image, at least one second detection frame containing a detection target is obtained, wherein the second detection frame has a category; based on at least one first detection frame and at least one second detection frame, a detection frame to be identified is obtained.

[0108] For example, four first detection frames are obtained using a depth image, and five second detection frames are obtained using a grayscale image. The pixel coordinates of the depth image and grayscale image are calibrated to obtain a mapping relationship between the pixel coordinates of the depth image and the grayscale image. This mapping relationship can be used to determine the positions of the five second detection frames in the depth image. When obtaining the detection frames to be identified, both the detection frames obtained from the depth image and the detection frames obtained from the grayscale image are considered to compensate for the missed detection frames caused by only considering the depth image.

[0109] In some embodiments, at least one second suspected detection frame is obtained in the second image, wherein the second suspected detection frame has a category and a probability confidence; and based on the probability confidence threshold, the at least one second suspected detection frame is filtered to obtain at least one second detection frame.

[0110] For example, applying a probability confidence threshold control to each suspected defect frame can filter out some false positive detection frames, thereby improving subsequent detection efficiency and accuracy.

[0111] In related art, when measuring the height or depth of a defect, the plane of the aluminum shell is used as the reference surface. However, since the weld seam itself has a certain thickness after the top cover is welded, the values ​​calculated by this method cannot reflect the actual depth and height of the defect. In addition, when multiple defects are close together, such as when there are both pinholes and protrusions, the measurement reference point is often inaccurate, resulting in measurement errors. In the embodiments of the present application, the measurement method using the plane of the aluminum shell as the reference surface is abandoned, and the cross-section of the weld seam after welding is selected as the reference. The specific implementation process is as follows.

[0112] In some embodiments of the present application, a cross-section of an area adjacent to the unidentified detection frame, but not part of the unidentified detection frame, is used as a measurement reference plane for the unidentified detection frame. The distance of the detection target relative to the measurement reference plane of the unidentified detection frame is determined. For protrusions, this measured distance is a height value; for pinholes, this measured distance is a depth value.

[0113] Taking Figure 6 as an example, to measure the depth of a concave pinhole, which is adjacent to a protrusion, a fully trained model is used to control a lower probability confidence threshold to reduce the possibility of missed detection. For example, if the probability confidence threshold is 0.05, all defect locations on the weld seam in the image can be obtained. For example, curve 610 is a defective area, and straight line 620 is a defect-free area. Using straight line 620 as the measurement reference surface, the accompanying slight protrusion point can be effectively avoided, and the distance from the concave point of the pinhole to the reference point can be calculated, thereby obtaining a more accurate depth measurement value of the pinhole. For example, the distance obtained by correctly measuring the plane is shown as label 630. Using related technologies, it is easy to select the wrong reference point, resulting in measurement errors, and the obtained measured distance is shown as label 640.

[0114] In this embodiment, all defect positions of the weld on the image are output in the form of a detection frame. The position of the detection frame constitutes the defective area, and the area outside the detection frame constitutes the defect-free area. Selecting the point in the defect-free area as the measurement reference point can effectively reduce the problem of inaccurate measurement caused by multiple defects close to each other.

[0115] In some embodiments, when the detection frame to be identified where the detection target is located is located in the upper top cover weld area, based on the pixel distribution difference in the first image, the cross-section of the non-weld area closest to the upper top cover weld area along the weld width direction is used as the measurement reference surface of the detection frame to be identified where the detection target is located, and the measurement distance of the detection target relative to the measurement reference surface of the detection frame to be identified where the detection target is located is determined.

[0116] Taking Figure 4 as an example, if the detection frame is located between line 440 and line 460, the cross section of the non-weld detection area to the right of line 460 is used as the measurement reference plane. Since there are no defects in the plane of the non-weld detection area, the measurement result is more accurate.

[0117] In some embodiments, when the detection frame to be identified where the detection target is located is located in the side top cover weld area, based on the pixel distribution difference in the first image, the cross-section of the non-weld area closest to the side top cover weld area along the weld width direction is used as the measurement reference surface of the detection frame to be identified where the detection target is located, and the measurement distance of the detection target relative to the measurement reference surface of the detection frame to be identified where the detection target is located is determined.

[0118] Taking Figure 4 as an example, if the inspection frame is located between line 410 and line 430, the cross section of the non-weld inspection area on the left side of line 410 is used as the measurement reference plane. Since there are no defects in the plane of the non-weld inspection area, the measurement result is more accurate.

[0119] In some embodiments, when the detection frame to be identified where the detection target is located is located in the central weld area, the cross-section of the area in the central weld area excluding the detection target is used as the measurement reference surface of the detection frame to be identified where the detection target is located, and the measurement distance of the detection target relative to the measurement reference surface of the detection frame to be identified where the detection target is located is determined.

[0120] Taking Figure 4 as an example, if the inspection frame is located between lines 430 and 440, since defects may exist above, below, left, and right in the central weld area, the cross-section of the area without the inspection frame is used as the measurement reference surface. This measurement reference surface is free of defects, thus also improving measurement accuracy.

[0121] In some embodiments, when the detection frame to be identified where the detection target is located is located in the first detection area, the cross-section of the area in the first detection area excluding the detection target is used as the measurement reference surface of the detection frame to be identified where the detection target is located; when the detection frame to be identified where the detection target is located is located in the second detection area, the cross-section of the area in the second detection area excluding the detection target is used as the measurement reference surface of the detection frame to be identified where the detection target is located; and the measurement distance of the detection target relative to the measurement reference surface of the detection frame to be identified where the detection target is located is determined.

[0122] For defects located in the R corner area, the cross section of the area without the inspection frame in this area is used as the measurement reference surface. This measurement reference surface does not contain defects, so it can also improve the accuracy of the measurement.

[0123] In the above embodiment, by finding a more accurate measurement reference point, the actual measurement value of the defect can be truly and effectively reflected, and even in the case of multiple defects close to each other, the accuracy of the specification measurement value can be effectively improved.

[0124] In some embodiments, the pixel points corresponding to the detection target are determined based on the depth information of the pixel points on the first image; and the area information of the detection target is determined based on the pixel points corresponding to the detection target on the first image.

[0125] For example, each pixel has three coordinates: x, y, and z. The z-axis represents image depth, the x-axis indicates weld width, and the y-axis indicates weld extension. A change in the z-axis coordinate indicates that the pixel is part of the detection target. By detecting pixels whose z-axis coordinates change continuously, the area corresponding to these pixels can be determined, which is the area of ​​the detection target. Subsequently, by determining whether the detection target's area exceeds a threshold, it can be determined whether the detection target is a defect.

[0126] FIG7 is a fourth flow chart of a welding defect detection method according to one or more embodiments.

[0127] In step 710 , a 3D camera scans the top cover weld to obtain a grayscale image containing brightness information and a depth image containing structure information.

[0128] In step 720, the grayscale image and the depth image are input into a trained deep learning model to obtain a series of suspected detection boxes containing category information and probability confidence.

[0129] In step 730 , false positive detection boxes are filtered.

[0130] For example, applying a probability confidence threshold control to each suspected defect frame can filter out some false positive detection frames, thereby improving subsequent detection accuracy and efficiency.

[0131] In step 740 , the weld is partitioned into structural areas, and a specific detection area of ​​each detection frame in the weld is determined.

[0132] In step 750 , attribute measurements are performed on the detection targets in each detection box.

[0133] In step 760 , based on the measurement thresholds corresponding to different detection areas and different types of defects, an attribute judgment is performed on each detection target to determine whether the detection target is a defect.

[0134] In step 770 , when a defect is identified, the battery is determined to be defective.

[0135] In the above-mentioned embodiment, a weld bead region segmentation algorithm is introduced to segment the inspection area according to actual needs. In combination with specification metrics, each region is independently controlled for specifications. This combination of visual texture information and defect structure information can reduce the misjudgment rate of good battery products, thereby increasing battery yield and shipment volume. This embodiment not only improves the accuracy of weld defect detection but also better conforms to the defect detection logic used in real-world scenarios, demonstrating its strong universality.

[0136] 8 is a first structural diagram of a welding defect detection device according to one or more embodiments. The welding defect detection device includes an image acquisition module 810 , a first segmentation module 820 , a second segmentation module 830 and a defect recognition module 840 .

[0137] The image acquisition module 810 is configured to acquire a first image of the welding between the battery top cover and the shell, where the first image includes structural information.

[0138] In some embodiments, the image acquisition module 810 is further configured to acquire a second image of the welding between the battery top cover and the shell, wherein the second image includes brightness information.

[0139] The first segmentation module 820 is configured to detect the weld area in the first image to obtain a weld area and a non-weld area.

[0140] In some embodiments, by analyzing contour features in the depth image, weld regions of interest and non-weld regions of no interest can be identified.

[0141] The second dividing module 830 is configured to identify a plurality of inspection areas of the weld between the battery top cover and the shell according to the weld area and the non-weld area.

[0142] In some embodiments, the second segmentation module 830 is configured to determine the effective width and center width of the weld region in the weld bead width direction; and, based on the effective width, determine a first predetermined distance and a second predetermined distance for the weld region to extend into the non-weld region along the weld bead width direction; and, based on the first predetermined distance, the second predetermined distance, and the center width, segment the weld region to obtain multiple detection regions. The effective width is determined based on the difference between the first edge position and the second edge position of the weld region in the weld bead width direction.

[0143] For example, the weld structure is refined and divided into the upper cover weld area, the side cover weld area and the center weld area, which is conducive to the production line to formulate independent defect measurement standards for different weld areas.

[0144] The defect identification module 840 is configured to determine that a defect exists in the weld when preset attribute information of at least one detection area among multiple detection areas of the weld meets a preset condition.

[0145] In the above embodiment, the first image containing structural information is used to identify the weld area of ​​interest and the non-weld area of ​​no interest, and further obtain multiple detection areas of the weld, so that the detection target of each detection area can be treated differently according to the process differences and the defect standards of different areas, and when the preset attribute information of at least one detection area meets the preset conditions, it is determined that the weld has defects, and the false positives are reduced while effectively intercepting the weld defects, the detection accuracy is improved, and the yield rate and shipment volume are thereby increased.

[0146] In some embodiments of the present application, the second division module 830 is configured to determine the third edge position and the fourth edge position of the central weld area in the weld width direction based on the first edge position and the second edge position of the weld area in the weld width direction, as well as the effective width and the center width; extend the first edge position along the weld width direction to the non-weld area away from the central weld area by a first predetermined distance to obtain the fifth edge position of the upper cover weld area in the weld width direction, wherein the third edge position is the edge position of the upper cover weld area close to the central weld area in the weld width direction; and extend the second edge position along the weld width direction to the non-weld area away from the central weld area by a second predetermined distance to obtain the sixth edge position of the side cover weld area in the weld width direction, wherein the fourth edge position is the edge position of the side cover weld area close to the central weld area in the weld width direction.

[0147] The third edge position is obtained by moving the first edge position toward the second edge position by a third predetermined distance, where the third predetermined distance is half of the difference between the effective width and the center width. The fourth edge position is obtained by moving the second edge position toward the first edge position by the third predetermined distance.

[0148] In the above embodiment, by dividing the weld into the upper cover weld area, the center weld area and the side cover weld area, it is convenient to subsequently perform targeted identification of detection targets in different areas, thereby improving the accuracy of weld defect detection.

[0149] In some embodiments, the multiple inspection areas further include a first inspection area and a second inspection area, wherein the first inspection area is located along the weld bead extension direction, on a first side of the upper roof weld area, the center weld bead area, and the side roof weld area, and the second inspection area is located along the weld bead extension direction, on a second side of the upper roof weld area, the center weld bead area, and the side roof weld area. By identifying the R-angle area, it is easier to perform targeted inspections on inspection targets located in the R-angle area, thereby improving the accuracy of weld defect detection.

[0150] The second dividing module 830 is configured to extend the seventh edge position of the weld area in the weld extension direction by a fourth predetermined distance toward the upper cover weld area, the center weld area, and the side cover weld area to obtain the eighth edge position of the first inspection area, wherein the seventh edge position is the edge position of the first inspection area away from the upper cover weld area, the center weld area, and the side cover weld area.

[0151] The second dividing module 830 is configured to extend the ninth edge position of the weld area in the weld extension direction by a fifth predetermined distance toward the upper cover weld area, the center weld area, and the side cover weld area to obtain the tenth edge position of the second inspection area, wherein the ninth edge position is the edge position of the second inspection area away from the upper cover weld area, the center weld area, and the side cover weld area.

[0152] In the above embodiment, by accurately dividing the R-angle area, the accuracy of the weld area division is improved, thereby facilitating the improvement of the accuracy of subsequent defect identification.

[0153] Figure 9 is a second schematic diagram of the structure of a welding defect detection device according to one or more embodiments. The welding defect detection device also includes a detection frame recognition module 910, a detection frame region determination module 920, and an attribute recognition module 930. The detection frame recognition module 910 is configured to obtain at least one first detection frame containing a detection target based on a first image, wherein the first detection frame has a category; and obtain a detection frame to be recognized based on the at least one first detection frame. The detection frame region determination module 920 is configured to determine the detection region corresponding to the detection frame to be recognized based on the position of the detection frame to be recognized in the weld. The attribute recognition module 930 is configured to determine at least one attribute information of the detection target in the detection frame to be recognized. The defect recognition module 840 is configured to determine the presence of a weld defect if at least one attribute information of the detection target in at least one detection frame to be recognized in at least one detection region is greater than or equal to the measurement threshold corresponding to the category and detection region of the detection frame to be recognized.

[0154] In this embodiment, a detection frame containing a detection target is identified through target detection, and the detection area where the detection frame is located is limited. The attribute information of each detection target is identified, and then the attribute information of detection targets in different detection areas and different categories is judged to identify whether there are defects in the weld. Through area suggestions and specification measurements, the occurrence of misjudgment of weld defects is reduced.

[0155] In some embodiments, the detection frame recognition module 910 is further configured to obtain at least one second detection frame containing the detection target based on the second image, wherein the second detection frame has a category, and obtain a detection frame to be recognized based on the at least one first detection frame and the at least one second detection frame. By combining the grayscale image and the depth image, the probability of missed detection of the detection frame can be reduced.

[0156] In some embodiments, the detection frame identification module 910 is further configured to obtain at least one first suspected detection frame in the first image, wherein the first suspected detection frame is associated with a category and a probability confidence level; and filter the at least one first suspected detection frame based on a probability confidence level threshold to obtain at least one first detection frame. This embodiment improves subsequent detection efficiency by filtering out false positive detection frames.

[0157] In some embodiments, the detection frame identification module 910 is further configured to obtain at least one second suspected detection frame in the second image, wherein the second suspected detection frame is associated with a category and a probability confidence level; and filter the at least one second suspected detection frame based on a probability confidence level threshold to obtain at least one second detection frame. This embodiment improves subsequent detection efficiency by filtering out false positive detection frames.

[0158] In some embodiments, the attribute recognition module 930 is further configured to use a cross-section of an area adjacent to the unidentified detection frame, but not within the unidentified detection frame, as a measurement reference plane for the unidentified detection frame, and to determine the distance of the detection target relative to the measurement reference plane of the unidentified detection frame. By selecting a more accurate measurement reference plane, the accuracy of the actual defect measurement can be improved.

[0159] In some embodiments, the attribute recognition module 930 is further configured to, when the detection frame to be identified where the detection target is located is located in the upper top cover weld area, use the cross-section of the non-weld area closest to the upper top cover weld area along the weld width direction as the measurement reference surface of the detection frame to be identified where the detection target is located, based on the pixel distribution difference in the first image; when the detection frame to be identified where the detection target is located is located in the side top cover weld area, use the cross-section of the non-weld area closest to the side top cover weld area along the weld width direction as the measurement reference surface of the detection frame to be identified where the detection target is located, based on the pixel distribution difference in the first image; when the detection frame to be identified where the detection target is located is located in the central weld area, use the cross-section of the area in the central weld area except the detection target as the measurement reference surface of the detection frame to be identified where the detection target is located; and determine the measurement distance of the detection target relative to the measurement reference surface of the detection frame to be identified where the detection target is located.

[0160] In this embodiment, different measurement reference surfaces are selected for detection frames of different detection areas. Since there are no defects in the measurement reference surfaces, the accuracy of the actual measurement values ​​of the defects is improved. Even in the case where multiple defects are close to each other, the accuracy of the specification measurement values ​​can be effectively improved.

[0161] In some embodiments, the attribute identification module 930 is also configured to, when the detection frame to be identified where the detection target is located is located in the first detection zone, use the cross-section of the area in the first detection zone excluding the detection target as the measurement reference surface of the detection frame to be identified where the detection target is located; when the detection frame to be identified where the detection target is located is located in the second detection zone, use the cross-section of the area in the second detection zone excluding the detection target as the measurement reference surface of the detection frame to be identified where the detection target is located; and determine the measurement distance of the detection target relative to the measurement reference surface of the detection frame to be identified where the detection target is located.

[0162] In this embodiment, by limiting the measurement reference surface of the detection frame of the R corner area, the detection accuracy of the actual measurement value of the defect in the R corner area can be improved.

[0163] In some embodiments, the attribute recognition module 930 is further configured to determine the pixel points corresponding to the detection target based on the depth information of the pixel points on the first image; and determine the area information of the detection target based on the pixel points corresponding to the detection target on the first image.

[0164] In this embodiment, by identifying the area information of the detection target, it is possible to determine whether the detection target is a defect from the perspective of area, thereby improving the comprehensiveness of attribute detection.

[0165] In some embodiments, the welding defect detection apparatus further includes a threshold setting module 940 configured to set a measurement threshold for each attribute information of a detection target within the detection frame to be identified, based on the category and detection area of ​​the detection frame to be identified. By setting corresponding measurement thresholds for each attribute information of detection targets of different types and in different detection areas, the accuracy of subsequent defect detection can be improved.

[0166] Figure 10 is a third schematic diagram of the structure of a welding defect detection device according to one or more embodiments. The welding defect detection device includes a memory 1010 and a processor 1020. Memory 1010 can be a disk, flash memory, or any other non-volatile storage medium. The memory is used to store the instructions described in the above embodiments. Processor 1020 is coupled to memory 1010 and can be implemented as one or more integrated circuits, such as a microprocessor or microcontroller. Processor 1020 is used to execute the instructions stored in the memory.

[0167] In some embodiments, the processor 1020 is coupled to the memory 1010 via a BUS 1030. The welding defect detection device can also be connected to an external storage device 1050 via a storage interface 1040 to access external data, and can also be connected to a network or another computer system (not shown) via a network interface 1060. Detailed descriptions are omitted here.

[0168] In this embodiment, the welding defect detection device stores data instructions in a memory and processes the instructions through a processor, which can reduce the misjudgment rate of weld defects and improve the accuracy of weld defect detection.

[0169] In other embodiments, the application provides a computer-readable storage medium having computer program instructions stored thereon, which implement the steps of the method in the above-described embodiment when the instructions are executed by a processor. Those skilled in the art will appreciate that the embodiments of the present application can be provided as methods, devices, or computer program products. Therefore, the application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the application can adopt the form of a computer program product implemented in one or more computer-usable non-transient storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0170] In some embodiments of the application, a computer program product is further provided, comprising computer program instructions, which, when executed by a processor, implement the welding defect detection method of any one of the above embodiments.

[0171] The technical solution of the present application can be applied to the detection of weld defects in the incoming material link, battery inspection sampling link, and shipment link.

[0172] Those skilled in the art will understand that in the above method of the specific embodiment, the writing order of each step does not mean a strict execution order and does not constitute any limitation on the real-time process. The specific execution order of each step should be determined by its function and possible internal logic.

[0173] The above description of the various embodiments tends to emphasize the differences between the various embodiments. The same or similar aspects can be referenced with each other and will not be repeated herein for the sake of brevity.

[0174] So far, the present application has been described in detail. In order to avoid obscuring the concept of the present application, some details well known in the art have not been described. Based on the above description, those skilled in the art can fully understand how to implement the technical solutions disclosed herein.

[0175] The methods and systems of the present application may be implemented in many ways. For example, the methods and systems of the present application may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above order of steps for the method is for illustration only, and the steps of the method of the present application are not limited to the order specifically described above, unless otherwise specifically stated. In addition, in some embodiments, the present application may also be implemented as programs recorded in a recording medium, which include machine-readable instructions for implementing the methods according to the present application. Therefore, the present application also covers recording media that store programs for executing the methods according to the present application.

[0176] Although some specific embodiments of the present application have been described in detail by way of example, it should be understood by those skilled in the art that the above examples are for illustration only and are not intended to limit the scope of the present application. It should be understood by those skilled in the art that the above examples may be modified without departing from the scope and spirit of the present application. The scope of the present application is defined by the appended claims. Industrial Applicability

[0177] The present disclosure provides a welding defect detection method, device, storage medium, and program product. The welding defect detection method includes: obtaining a first image of a battery top cover and a battery shell welded together, the first image including structural information; detecting a weld region in the first image to obtain a weld region and a non-weld region; identifying multiple detection regions of the weld between the battery top cover and the battery shell based on the weld region and the non-weld region; and determining that the weld is defective if preset attribute information of at least one detection region in the multiple detection regions meets preset conditions. In the embodiment of the present application, the image is divided into regions to identify the weld region and the non-weld region of interest, and then the weld structure is refined to obtain multiple detection regions. This facilitates independent defect determination for each detection region. If the preset attribute information of at least one detection region meets preset conditions, it is determined that the weld is defective. Through regional analysis and specification card control, the false positive rate of weld defects can be reduced while effectively intercepting weld defects, improving detection accuracy, and thereby increasing battery yield and shipment volume.

Claims

1. A welding defect detection method, comprising: Acquire a first image of the welding between the battery top cover and the shell, wherein the first image includes structural information; Detecting the weld area in the first image to obtain a weld area and a non-weld area; Determining the effective width and center width of the weld area in the weld width direction; Determine, according to the effective width, a first predetermined distance and a second predetermined distance by which the weld region is extended toward the non-weld region along the weld width direction; Based on the first predetermined distance, the second predetermined distance and the center width, the weld between the battery top cover and the shell is divided into regions to obtain a plurality of detection regions; as well as When preset attribute information of at least one detection area among multiple detection areas of the weld meets preset conditions, it is determined that the weld has a defect.

2. The welding defect detection method according to claim 1, wherein: Determining the effective width of the weld area in the direction of the weld width includes: Determining a first edge position and a second edge position of the weld area in a width direction of the weld bead; and According to the difference between the second edge position and the first edge position, the effective width of the weld area in the weld width direction is obtained.

3. The welding defect detection method according to claim 2, wherein: The multiple detection areas include an upper cover weld area, a center weld area and a side cover weld area, and the center weld area is located between the upper cover weld area and the side cover weld area along the weld width direction, wherein the weld between the battery top cover and the shell is divided into regions based on the first predetermined distance, the second predetermined distance and the center width to obtain multiple detection areas, including: Determine a third edge position and a fourth edge position of the central weld zone in the direction of the weld width according to a first edge position and a second edge position of the weld zone in the direction of the weld width, the effective width, and the center width; Extending the first edge position along the weld width direction to the non-weld area away from the central weld area by the first predetermined distance to obtain a fifth edge position of the upper cover weld area in the weld width direction, wherein the third edge position is an edge position of the upper cover weld area close to the central weld area in the weld width direction; and The second edge position is extended along the weld width direction to the non-weld area away from the central weld area by the second predetermined distance to obtain the sixth edge position of the side top cover weld area in the weld width direction, wherein the fourth edge position is the edge position of the side top cover weld area close to the central weld area in the weld width direction.

4. The welding defect detection method according to claim 3, wherein: The step of determining the third edge position and the fourth edge position of the central weld area in the weld width direction according to the first edge position and the second edge position of the weld area in the weld width direction, the effective width, and the center width comprises: moving the first edge position toward the second edge position by a third predetermined distance to obtain the third edge position, wherein the third predetermined distance is half of the difference between the effective width and the center width; and The second edge position is moved toward the first edge position by the third predetermined distance to obtain the fourth edge position.

5. The welding defect detection method according to claim 3 or 4, wherein: The multiple inspection areas also include a first inspection area and a second inspection area, wherein the first inspection area is located along the extension direction of the weld at a first side of the upper top cover weld area, the center weld area and the side top cover weld area, and the second inspection area is located along the extension direction of the weld at a second side of the upper top cover weld area, the center weld area and the side top cover weld area.

6. The welding defect detection method according to claim 5, wherein: The obtaining of multiple detection areas further includes: The seventh edge position of the weld area in the weld extension direction is extended a fourth predetermined distance toward the upper cover weld area, the center weld area and the side cover weld area to obtain the eighth edge position of the first inspection area, wherein the seventh edge position is the edge position of the first inspection area away from the upper cover weld area, the center weld area and the side cover weld area.

7. The welding defect detection method according to claim 5 or 6, wherein: The obtaining of multiple detection areas further includes: The ninth edge position of the weld area in the weld extension direction is extended a fifth predetermined distance toward the upper cover weld area, the center weld area and the side cover weld area to obtain the tenth edge position of the second inspection area, wherein the ninth edge position is the edge position of the second inspection area away from the upper cover weld area, the center weld area and the side cover weld area.

8. The welding defect detection method according to any one of claims 1 to 7, wherein: The method further comprises: Obtaining at least one first detection frame containing a detection target according to the first image, wherein the first detection frame has a category; Obtaining a detection frame to be identified according to the at least one first detection frame; Determine a detection area corresponding to the detection frame to be identified according to the position of the detection frame to be identified on the weld; Determine at least one attribute information of the detection target in the detection frame to be identified; When preset attribute information of at least one detection area among the multiple detection areas of the weld meets a preset condition, determining that the weld has a defect includes: When at least one attribute information of the detection target in at least one to-be-identified detection frame of at least one detection area is greater than or equal to the measurement threshold corresponding to the category of the to-be-identified detection frame where the detection target is located and the detection area, it is determined that the weld has a defect.

9. The welding defect detection method according to claim 8, wherein: The method further comprises: Before determining the detection area corresponding to the detection frame to be identified according to the position of the detection frame to be identified on the weld, obtaining a second image of the welding between the battery top cover and the shell, wherein the second image includes brightness information; Obtaining at least one second detection frame including the detection target according to the second image, wherein the second detection frame has a category; Obtaining a detection frame to be identified according to the at least one first detection frame includes: The to-be-recognized detection frame is obtained according to the at least one first detection frame and the at least one second detection frame.

10. The welding defect detection method according to claim 8 or 9, wherein: The at least one attribute information includes a measured distance, wherein the determining at least one attribute information of the detection target in the to-be-identified detection frame includes: Using a cross section of an area that does not belong to the detection frame to be identified and is adjacent to the detection frame to be identified as a measurement reference surface for the detection frame to be identified; and Determine a measurement distance of the detection target relative to a measurement reference plane of a detection frame to be identified where the detection target is located.

11. The welding defect detection method according to any one of claims 8 to 10, wherein: The at least one attribute information includes a measurement distance, the multiple detection areas include an upper cover weld area, a central weld area, and a side cover weld area, the central weld area is located between the upper cover weld area and the side cover weld area along the weld width direction, wherein the determining of at least one attribute information of the detection target in the detection frame to be identified includes: In the case where the detection frame to be identified where the detection target is located is located in the upper cover weld area, according to the pixel distribution difference in the first image, a cross section of a non-weld area closest to the upper cover weld area along the weld width direction is used as a measurement reference surface of the detection frame to be identified where the detection target is located; In the case where the detection frame to be identified where the detection target is located is located in the side top cover weld area, according to the pixel distribution difference in the first image, a cross section of a non-weld area closest to the side top cover weld area along the weld bead width direction is used as a measurement reference surface of the detection frame to be identified where the detection target is located; When the detection frame to be identified where the detection target is located is located in the central weld bead area, a cross section of an area in the central weld bead area excluding the detection target is used as a measurement reference surface of the detection frame to be identified where the detection target is located; and Determine a measurement distance of the detection target relative to a measurement reference plane of a detection frame to be identified where the detection target is located.

12. The welding defect detection method according to claim 11, wherein: The multiple detection areas further include a first detection area and a second detection area, wherein the first detection area is located along the extension direction of the weld bead at a first side of the upper top cover weld area, the central weld bead area and the side top cover weld area, and the second detection area is located along the extension direction of the weld bead at a second side of the upper top cover weld area, the central weld bead area and the side top cover weld area, wherein determining at least one attribute information of the detection target in the detection frame to be identified includes: When the detection frame to be identified where the detection target is located is located in the first detection area, a cross section of an area in the first detection area excluding the detection target is used as a measurement reference surface of the detection frame to be identified where the detection target is located; When the detection frame to be identified where the detection target is located is located in the second detection area, a cross section of an area in the second detection area excluding the detection target is used as a measurement reference surface of the detection frame to be identified where the detection target is located; and Determine a measurement distance of the detection target relative to a measurement reference plane of a detection frame to be identified where the detection target is located.

13. The welding defect detection method according to any one of claims 8 to 12, wherein: The at least one attribute information further includes area information, wherein the at least one attribute information of the detection target in the to-be-identified detection frame is determined further includes: Determining the pixel corresponding to the detection target according to the depth information of the pixel on the first image; and The area information of the detection target is determined according to the pixel points corresponding to the detection target on the first image.

14. The welding defect detection method according to any one of claims 8 to 13, wherein: The obtaining of at least one first detection frame containing the detection target comprises: Acquire at least one first suspected detection frame in the first image, wherein the first suspected detection frame has a category and a probability confidence; and Based on the probability confidence threshold, the at least one first suspected detection frame is filtered to obtain the at least one first detection frame.

15. The welding defect detection method according to claim 9, wherein: The obtaining, according to the second image, at least one second detection frame including the detection target comprises: Acquire at least one second suspected detection frame in the second image, wherein the second suspected detection frame has a category and a probability confidence; and Based on the probability confidence threshold, the at least one second suspected detection frame is filtered to obtain the at least one second detection frame.

16. The welding defect detection method according to claim 8, wherein: The method further comprises: According to the category and the detection area to which the to-be-identified detection frame belongs, a measurement threshold of the attribute information of the detection target in the to-be-identified detection frame is set.

17. A welding defect detection device, comprising: An image acquisition module is configured to acquire a first image of the welding between the battery top cover and the shell, wherein the first image includes structural information; A first segmentation module is configured to detect the weld area in the first image to obtain a weld area and a non-weld area; A second division module is configured to determine an effective width and a center width of the weld area in a weld width direction, determine a first predetermined distance and a second predetermined distance for the weld area to extend toward the non-weld area along the weld width direction according to the effective width, and divide the weld between the battery top cover and the shell into regions based on the first predetermined distance, the second predetermined distance and the center width to obtain a plurality of detection regions; as well as The defect recognition module is configured to determine that there is a defect in the weld when preset attribute information of at least one detection area among multiple detection areas of the weld meets preset conditions.

18. A welding defect detection device, comprising: Memory; as well as A processor coupled to the memory, wherein the processor is configured to execute the welding defect detection method according to any one of claims 1 to 16 based on instructions stored in the memory.

19. A computer-readable storage medium having computer program instructions stored thereon, wherein the instructions, when executed by a processor, implement the welding defect detection method according to any one of claims 1 to 16.

20. A computer program product, comprising a computer program or instructions, which, when executed on a computer, enables the computer to execute the welding defect detection method according to any one of claims 1 to 16.

Images