Thermal battery online assembly defect detection system based on machine vision

Abstract

The invention discloses a thermal battery online assembly defect detection system based on machine vision, which comprises an online detection platform for thermal battery defect detection, a built-in thermal battery defect detection algorithm and a thermal battery software interaction interface, the online detection platform is of a box structure and comprises a thermal battery clamp group, a light source illumination system, an optical imaging system, an image acquisition system and a computer processing system; the built-in thermal battery defect detection algorithm comprises a correct thermal battery detection template acquisition algorithm and a to-be-detected electric pile detection method; the thermal battery software interaction interface comprises four interfaces, namely a visual window interface, a data recording interface, an operation bar interface and an assembly list output interface. The problems of radiation, complex data acquisition and the like can be avoided; the image data acquisition is convenient, and if thermal battery assembly defects are detected, the thermal battery assembly sequence can be changed in time, so that the assembly defects are avoided, the cost of waste products is reduced, and the overall production cost is reduced.

Description

Technical Field

[0001] The invention belongs to thermal battery detection technology, and in particular relates to a thermal battery online assembly defect detection system based on machine vision. Background Art

[0002] In the field of power supply, primary batteries (such as thermal batteries) are the first choice for missile power supply both at home and abroad because of their advantages such as high power, long life and maintenance-free. However, since primary batteries are not rechargeable, their electrical performance can only be verified by sampling, which makes it impossible to reduce the risk to zero. Therefore, before the weapons and equipment are put into use, it is essential to effectively evaluate the safety and reliability of the assembled thermal batteries and reserve batteries. At present, there are three main methods used at home and abroad to detect thermal battery defects: electrochemical performance detection, battery internal material structure morphology detection and X-ray non-destructive testing. However, electrochemical performance detection and battery internal material structure morphology detection can no longer meet the military's testing requirements; X-ray detection methods are limited by the detection environment and cannot meet the needs of real-time detection of thermal battery production lines. In addition, X-ray detection is usually carried out after the thermal battery is packaged, and assembly defects cannot be discovered in time, thereby effectively reducing production costs.

[0003] With the continuous advancement of machine vision technology, its application in product quality inspection on industrial production lines is becoming more and more popular, and the use of machine vision to replace quality inspectors for defect inspection has gradually become a trend. At present, the application of machine vision in surface defect inspection is mainly concentrated in industrial production lines with large-scale production, such as parts, cloth, printed materials, integrated circuits, wheels, glass and other processing and manufacturing fields.

[0004] There have been a lot of studies on the application of machine vision in product surface defect detection at home and abroad, and good results have been achieved. For example, reference [1] designed a weld defect detection and classification algorithm for thin-walled metal cans based on machine vision technology. The accuracy of weld defect recognition and classification reached more than 95%, which can meet the requirements of real-time and continuous weld defect detection. Reference [2] proposed a method for detecting fruit skin defects using a machine vision system, which helps fruit packaging manufacturers to quickly and efficiently evaluate fruit quality and distinguish between good and bad fruits. The system's false detection rate is only 1.7%. Reference [3] proposed an online monitoring system for ceramic can surface defects based on machine vision. Experiments show that the system can quickly and accurately detect whether the target object has defects.

[0005] At present, the existing method for detecting thermal battery assembly defects is to detect the packaged thermal battery, that is, to use CT to scan the battery stack to obtain X-ray images, and then perform defect analysis based on the images. Reference [4] proposed a method for using CT to scan the thermal battery stack to obtain X-ray images, and to identify thermal battery assembly defects based on these image analysis. By comparing and analyzing the defective single thermal battery with the correct template single thermal battery, and using the self-comparison of the internal structural features of the thermal battery, the overall inversion of the thermal battery single battery, the missing internal structure of the single battery, and the wrong structural order of the single battery were detected, and a non-destructive testing system was established.

[0006] The above methods have the following problems in applying thermal battery assembly defect detection:

[0007] 1) The current thermal battery detection method is to use X-ray non-destructive testing technology. To scan the packaged thermal battery through CT, the finished thermal battery needs to be sent to the CT scanning equipment separately to collect the internal assembly image data of the thermal battery, and perform defect analysis based on this image. Testing thermal batteries in this way increases the complexity and time cost of the thermal battery manufacturing process. In addition, this detection method is greatly constrained by the detection environment and has high requirements for the detection environment.

[0008] 2) Due to the special properties of thermal battery monomer materials, thermal battery monomers must be stored in a vacuum environment, otherwise the battery materials will deteriorate. Therefore, the use of X-ray non-destructive testing technology to detect thermal battery assembly defects must be carried out after the thermal battery packaging is completed. Failure to promptly discover and correct defects during the assembly process may result in assembly defects flowing into the next production link or market.

[0009] 3) Using X-rays to detect thermal battery assembly defects cannot be directly used in the battery assembly line, and requires special X-ray

[0010] X-ray equipment and operators are complex to operate. X-ray scanning equipment has certain radiation, which requires strict safety protection measures and certain requirements for operators.

[0011] References:

[0012] [1]Sun J, Li C, Wu XJ, et al. An effective method of weld defect detection and classification based on machine vision [J]. IEEE Transactions on Industrial Informatics, 2019, 15(12): 6322-6333.

[0013] [2]Wang L, Li A, Tian X.Detection of fruit skin defects using machinevision system[C] / / 2013Sixth International Conference on Business Intelligence and Financial Engineering.IEEE,2013:44-48.

[0014] [3]Bao N, Ran

[0015] [4] Zhou Wei, Xiao Xin, Wang Rui, Liu Zhibo, Yi Jilu. Thermal battery assembly defect detection based on image recognition[J]. Manufacturing Automation, 2022, 44(6): 87-91. Summary of the invention

[0016] In view of the shortcomings of detecting thermal battery assembly defects, the present invention provides a thermal battery online assembly defect detection system based on machine vision.

[0017] A machine vision-based thermal battery online assembly defect detection system of the present invention comprises an online detection platform for thermal battery defect detection, a built-in thermal battery defect detection algorithm and a thermal battery software interaction interface.

[0018] The online inspection platform is a box structure, including a thermal battery fixture group, a light source lighting system, an optical imaging system, an image acquisition system and a computer processing system. The thermal battery fixture group is a drawer board set at the bottom of the box structure, and the thermal battery fixture is installed in the middle of the drawer board to fix the thermal battery. The light source lighting system includes a small ring light source and a large ring light source set up and down on the bracket inside the box, and the small ring light source and the large ring light source can be adjusted up and down in the vertical direction. The optical imaging system includes a high-pixel industrial camera and lens, and the industrial camera and the lens can be adjusted up and down in the vertical direction through an adjustment device. The industrial camera, the small ring light source and the large ring light source are arranged from top to bottom in sequence, with their centers in a line and vertically aligned with the thermal battery being photographed. The image acquisition system includes a screen, which is used to display the output results of the thermal battery assembly defects and the window for the operator to interact with the system. The computer processing system has a built-in thermal battery defect detection algorithm.

[0019] The built-in thermal battery defect detection algorithm includes the correct thermal battery detection template acquisition algorithm and the battery stack detection method to be detected.

[0020] The thermal battery software interactive interface includes four interfaces, namely, the visualization window interface, the data recording interface, the operation bar interface, and the assembly list output interface.

[0021] Furthermore, the drawer plate adopts an equidistant multi-circular hole design, the thermal battery clamp is designed as a V-shaped groove, and the drawer plate and the thermal battery clamp are made of plastic.

[0022] Furthermore, the industrial camera is a MV-CH310-10TC-M58-NN industrial camera produced by Hikvision Robotics, the lens model is MVL-LF3528-F, and the focal length is 35mm.

[0023] Furthermore, the correct hot battery detection template collection algorithm is specifically as follows:

[0024] The thermal battery image is captured by a high-pixel industrial camera, and the image of the thermal battery stack is manually selected and saved. Next, the battery stack image is grayed out, and the highlight part of the single heater is identified to obtain an image containing the highlight part. Then, the OTSU binarization algorithm is used for adaptive binarization processing, and the part of the image with pixels greater than the threshold is set to white, and the part with pixels less than the threshold is set to black. Finally, the noise in the binary image is removed through opening and closing operations such as corrosion and expansion.

[0025] Traverse each row of the binary image, obtain the width of the highlight part of each row and the number of the leftmost columns; calculate the median of the width of the highlight part and the number of the leftmost columns of all rows, and use them as the width and starting point of the highlight part respectively; then, extend the width of the highlight part to the left, and cut out the area containing the highlight part and the non-highlight part from the image.

[0026] The non-highlight part of the intercepted image is processed to achieve the segmentation of the first pile, the last pile and the middle pile; then the middle pile is processed and the structures of the first and last piles are identified.

[0027] Furthermore, the specific steps for segmenting the first pile, the tail pile, and the middle pile are as follows:

[0028] S1.1: Grayscale the non-highlight part of the battery stack, then select a suitable threshold based on the white pad for binarization, and then perform opening and closing operations to remove noise.

[0029] S1.2: Identification of the last gasket of the first pile: perform local processing on the first one-sixth of the rows of the binary image, delineate each connected domain and calculate the area, and retain the three connected domains with the largest area; since the first pile contains two regional gaskets and an ignition plate and electric ignition head assembly, the thickness difference with the white negative electrode sheet is quite large, and the last connected domain is the last gasket of the first pile.

[0030] S1.3: Identification of the first gasket of the tail pile: The last one-sixth of the rows of the binary image are processed in a similar way to the identification of the last gasket of the first pile. The tail pile has three regional gaskets, which also have a large difference in thickness from the white negative electrode sheet. The three connected domains with the largest area are retained, and the first connected domain is the first gasket of the tail pile.

[0031] S1.4: Division of the head, tail and middle piles: The part above the bottom of the connected domain of the last gasket of the first pile is the first pile; the part below the bottom of the connected domain of the last gasket of the first pile and the part above the top of the connected domain of the tail pile are the middle pile; the part below the top of the connected domain of the tail pile is the tail pile.

[0032] Furthermore, the specific steps of the intermediate heap processing are:

[0033] S2.1: Identification of the middle lead-out plate and segmentation of the upper and lower piles: Due to the structure of the middle lead-out plate, the visual feature is yellow, and the lead-out plate is located in the middle of the entire middle pile. One-tenth of the middle of the middle pile is cut out for local area analysis; first convert the local area into an HSV image, and then identify the location of the lead-out plate based on the specific characteristics of the three HSV channels expressed by the yellow color; calculate the average pixel value of each row, and the row with the largest average value is the row for segmenting the upper and lower piles; the part above the row with the largest average value is the upper pile, and the part below is the lower pile.

[0034] S2.2: Single-body segmentation: perform segmentation on each single body of the upper and lower piles segmented above; firstly capture the highlight parts of the upper and lower piles, as the single heater will show reflective visual characteristics under the illumination of the annular light source; firstly perform highlight recognition on the single heater, and perform OTSU binarization, i.e., adaptive binarization, on the image based on highlight recognition; count the number of pixels with a grayscale value of 255 in each row, and when the number of pixels with a grayscale value of 255 in the first five rows is greater than 0, and the number of pixels with a grayscale value of 255 in the last three rows is equal to 0, it is a segmentation row for each single body; then segment the single body based on the segmentation row of each single body.

[0035] S2.3: Internal segmentation of the monomer, including single heater identification, composite sheet identification and negative electrode sheet identification: Since the single heater has reflective characteristics, each monomer image is first subjected to highlight processing, and then adaptive binarization is performed; then, the number of pixels with a grayscale value of 255 in each row is counted. If the number is greater than 0, it is determined that the row contains a single heater; the grayscale value of the row where the single heater corresponds to each column in the previously intercepted monomer image is set to 0, that is, it is set to black; the image is binarized, and then the grayscale of each pixel is nonlinearly adjusted. The grayscale nonlinear adjustment method is:

[0036] a. For pixels whose grayscale value is greater than 0 and less than 60, set it to 0.5 times the current pixel grayscale value;

[0037] b. For pixels whose grayscale value is greater than or equal to 60 and less than 80, set it to 1.25 times the current pixel grayscale value;

[0038] c. For pixels whose grayscale value is greater than or equal to 80 and less than 120, set it to 1.5 times the current pixel grayscale value;

[0039] d. For pixels whose grayscale value is greater than or equal to 150 and less than 250, set it to twice the current pixel grayscale value;

[0040] After adjustment, the average grayscale value of each row of pixels is calculated. If the average grayscale value of the first four rows is 0 and the average grayscale value of the last four rows is greater than 0, it is identified as a single heating sheet; if the average grayscale value of the first four rows is less than 100 and the average grayscale value of the last four rows is greater than 100, it is identified as a composite sheet; if the average grayscale value of the first four rows is not 0 and the average grayscale value of the last four rows is greater than 0, it is identified as a negative electrode sheet.

[0041] S2.4: Output the assembly list based on the categories and quantities of each monomer structure identified.

[0042] Furthermore, the identification of each structure of the head and tail pile is specifically as follows:

[0043] For the structural recognition of the first and last piles, first capture the images of the non-highlight parts, convert them to grayscale, and then binarize them based on the white features of the gasket. Due to camera distortion, the images of cylindrical objects taken will appear tilted: the structure of the first pile image is slightly tilted from the upper left to the lower right, while the structure of the tail pile image is slightly tilted from the lower left to the upper right. Therefore, only the average grayscale value of the pixels in half of the column of the non-highlight part of each row is calculated.

[0044] For the first pile, the structure includes the end heating plate, the upper fixing band, the gasket, the ignition plate and the electric ignition head assembly; the visual features of the gasket, the ignition plate and the electric ignition head assembly are white, while the visual features of the end heating plate and the upper fixing band are black; these color features are used to distinguish the gasket from the end heating plate and the upper fixing band; the ignition plate is tightly combined with the electric ignition head assembly, and the thickness of the ignition plate is much smaller than the thickness of the electric ignition head assembly, and the combined thickness of the two is also much greater than the thickness of the gasket; therefore, the ignition plate and the electric ignition head assembly, as well as the gasket, are identified by these thickness differences.

[0045] For the tail pile, the identification method is similar to that of the upper pile; the tail pile includes a lower fixing belt, a gasket and an end heating plate; the visual feature of the gasket is white, while the visual features of the end heating plate and the lower fixing belt are black; the gasket, the end heating plate and the lower fixing belt are identified based on this feature.

[0046] Furthermore, the detection method of the battery stack to be detected is implemented by multi-angle template matching. The specific steps of multi-angle template matching are:

[0047] S3.1: Perform pyramid downsampling twice on the template battery stack image and the image of the battery to be matched with the template.

[0048] S3.2: Perform the first template matching between the unrotated template battery stack image and the template battery matching image, and record the suitability of the current match, which is regarded as the most suitable match.

[0049] S3.3: First, rotate the template battery stack image 5° to the left, and then perform template matching. Compare the obtained matching fitness with the most suitable template matching fitness. If it is greater than the most suitable template matching stack, record the current angle and let the obtained matching fitness replace the most suitable template matching fitness. Rotate right 0.1° in sequence, and then perform the above comparison process to obtain the maximum matching fitness and record the angle of rotation of the template battery.

[0050] After obtaining the battery stack, carry out the correct thermal battery detection template detection process, and compare the detection data with the correct template battery stack detection data. If any inconsistency is found, it will be marked in red in the assembly list output table.

[0051] Furthermore, the thermal battery software interaction interface is based on the Visual C++ language on the Windows 11 system and is completed on the QT platform, including four interfaces:

[0052] The visualization window interface is used for real-time display of hot batteries photographed by high-pixel industrial cameras, display of the battery stack in the hot battery, and selection of the shooting template to obtain the correct template battery stack.

[0053] The data recording interface is used to record the thermal battery detection log, record the thermal battery stack detection time, detect the thermal battery stack model, and the detection results.

[0054] The assembly list output interface is used to display the order and quantity of each structure of the thermal battery.

[0055] The operation bar interface is used to adjust various parameters and control image capture.

[0056] Furthermore, the interactive software operation process of the online detection platform is as follows:

[0057] S4.1: In the thermal battery production line, place the assembled but unpackaged thermal battery semi-finished products on the drawer board fixture of the equipment.

[0058] S4.2: Close the drawer panel, click "Template" and select "Photo Template".

[0059] S4.3: Select the hot battery stack in the visualization window and click the “Detect Button”.

[0060] S4.4: After the assembly list is generated, click “Save” to save the template data for inspection.

[0061] S4.5: Click the "Detect" button under "Battery Parameters" to start shooting.

[0062] The beneficial technical effects of the present invention are:

[0063] The present invention uses a high-pixel industrial camera instead of an X-ray scanning device to collect thermoelectric image data, which can avoid radiation and complex data collection. At the same time, the image data collection has low environmental requirements and can be directly used on the thermal battery production line. It is convenient to collect image data, and if a thermal battery assembly defect is detected, the thermal battery assembly order can be changed in time to avoid assembly defects and reduce the cost of waste, thereby reducing the overall production cost. At the same time, the detection system can make a thermal battery to detect a battery in real time, without the need to send it to the detection equipment separately for detection, which reduces the detection time and improves the overall production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 Schematic diagram of the structure of the online detection platform for thermal battery defects.

[0065] Figure 2 Schematic diagram of the thermal battery fixture group structure.

[0066] Figure 3 This is a flowchart of the head and tail heap segmentation.

[0067] Figure 4 Identify flow charts for each intermediate structure.

[0068] Figure 5 This is the basic process of thermal battery defect detection algorithm.

[0069] Figure 6 This is a schematic diagram of the thermal battery software interaction interface.

[0070] Figure 7 This is the operation flow chart of the interactive software of the online detection platform. DETAILED DESCRIPTION

[0071] The present invention is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0072] A machine vision-based thermal battery online assembly defect detection system of the present invention comprises an online detection platform for thermal battery defect detection, a built-in thermal battery defect detection algorithm and a thermal battery software interaction interface.

[0073] like Figure 1 As shown, the online detection platform is a box structure, including a thermal battery fixture group, a light source lighting system, an optical imaging system, an image acquisition system and a computer processing system.

[0074] The thermal battery fixture set is a drawer board 1 set at the bottom of the box structure, and a thermal battery fixture 2 is installed in the middle of the drawer board 1 to fix the thermal battery 3. When the worker needs to place the thermal battery 3, he only needs to open the drawer board 1, place the thermal battery 3 accurately, and then close the drawer. The drawer board 1 adopts an equidistant multi-circular hole design, such as Figure 2 As shown, the distance between the clamps can be adjusted according to the size of the thermal battery to properly place the thermal battery stack. The thermal battery clamp 2 is designed as a V-groove, which is suitable for fixing a cylindrical thermal battery stack. The V-groove can ensure that the thermal battery 3 remains fixed during the opening and closing of the drawer. Plastic rather than carbon steel is used to make the clamp material to prevent the influence of carbon steel on the performance of the thermal battery during the detection process.

[0075] The light source lighting system includes a small ring light source 5 and a large ring light source 8 which are arranged up and down on a bracket 4 inside the box, and the small ring light source 5 and the large ring light source 8 can be adjusted up and down in the vertical direction. These light sources are designed to provide sufficient lighting for the battery stack during camera shooting. The large ring light source 8 is used for overall fill light to ensure that every part of the battery stack can be illuminated and that the structural single heating plate in the thermal battery stack can fully reflect light. The small ring light source 5 is used to eliminate the shadow cast by the large ring light source on the battery stack. The reason for using two ring light sources is that if only one ring light source is used, a ring shadow will appear on the thermal battery stack. By adding a small ring light source that runs through the camera, the shadow area cast by the large ring light source can be effectively supplemented.

[0076] The optical imaging system includes a high-pixel industrial camera 6 and a lens, which can be adjusted up and down in the vertical direction through an adjustment device 7. The industrial camera 6 is a model MV-CH310-10TC-M58-NN industrial camera of Hikvision Robotics, and the lens model is MVL-LF3528-F with a focal length of 35 mm. The high-pixel industrial camera is selected because the internal structure of the thermal battery cell is a layered structure with very small thickness of each layer.

[0077] The industrial camera 6 , the small ring light source 5 and the large ring light source 8 are arranged sequentially from top to bottom, with their centers on a line and vertically aligned with the thermal battery 3 being photographed.

[0078] The image acquisition system comprises a screen 9, which is used to display the thermal battery assembly defect output results and a window for the operator to interact with the system.

[0079] The computer processing system has a built-in thermal battery defect detection algorithm. The built-in thermal battery defect detection algorithm includes a correct thermal battery detection template acquisition algorithm and a detection method for the battery stack to be detected.

[0080] The thermal battery software interactive interface includes four interfaces, namely, the visualization window interface, the data recording interface, the operation bar interface, and the assembly list output interface.

[0081] Furthermore, the correct hot battery detection template collection algorithm is specifically as follows:

[0082] The hot battery image is captured by a high-pixel industrial camera, and the image of the hot battery stack is manually selected and saved for subsequent multi-angle template matching of the battery to be inspected. The purpose of this step is to remove the external environment of the non-thermal battery stack, because the assembly defect detection of this system is mainly aimed at the hot battery stack. Next, the battery stack image is grayed, and the highlight part of the single heating plate is identified to obtain an image containing the highlight part; then, the OTSU binarization algorithm (Otsu method-maximum inter-class variance method) is used for adaptive binarization processing, and the part of the image pixel greater than the threshold is set to white (255), and the part less than the threshold is set to black (0); finally, the noise in the binary image is removed through opening and closing operations such as corrosion and expansion.

[0083] Traverse each row of the binary image, obtain the width of the highlight part of each row and the number of the leftmost columns; calculate the median of the width of the highlight part and the number of the leftmost columns of all rows, and use them as the width and starting point of the highlight part respectively; then, extend the width of the highlight part to the left, and cut out the area containing the highlight part and the non-highlight part from the image.

[0084] The non-highlight part of the intercepted image is processed to achieve the segmentation of the first pile, the last pile and the middle pile; then the middle pile is processed and the structures of the first and last piles are identified.

[0085] Furthermore, Figure 3 This is a flowchart of the head and tail heap segmentation. The specific steps are:

[0086] S1.1: Grayscale the non-highlight part of the battery stack, then select a suitable threshold based on the white pad for binarization, and then perform opening and closing operations to remove noise.

[0087] S1.2: Identification of the last gasket of the first pile: perform local processing on the first one-sixth of the rows of the binary image, delineate each connected domain and calculate the area, and retain the three connected domains with the largest area; since the first pile contains two regional gaskets and an ignition plate and electric ignition head assembly (with a larger area), the thickness difference with the white negative electrode sheet is quite large, and the last connected domain is the last gasket of the first pile.

[0088] S1.3: Identification of the first gasket of the tail pile: The last one-sixth of the rows of the binary image are processed in a similar way to the identification of the last gasket of the first pile. The tail pile has three regional gaskets, which also have a large difference in thickness from the white negative electrode sheet. The three connected domains with the largest area are retained, and the first connected domain is the first gasket of the tail pile.

[0089] S1.4: Division of the head, tail and middle piles: The part above the bottom of the connected domain of the last gasket of the first pile is the first pile; the part below the bottom of the connected domain of the last gasket of the first pile and the part above the top of the connected domain of the tail pile are the middle pile; the part below the top of the connected domain of the tail pile is the tail pile.

[0090] In this way, by processing the image of the non-highlight part, the effective segmentation of the first pile, the last pile and the middle pile can be achieved.

[0091] Furthermore, the intermediate heap processing flow is as follows Figure 4 As shown, the specific steps are:

[0092] S2.1: Identification of the middle lead-out plate and segmentation of the upper and lower piles: Due to the structure of the middle lead-out plate, the visual feature is yellow, and the lead-out plate is located in the middle of the entire middle pile. One-tenth of the middle of the middle pile is cut out for local area analysis; first convert the local area into an HSV image, and then identify the location of the lead-out plate based on the specific characteristics of the three HSV channels expressed by the yellow color; calculate the average pixel value of each row, and the row with the largest average value is the row for segmenting the upper and lower piles; the part above the row with the largest average value is the upper pile, and the part below is the lower pile.

[0093] S2.2: Single-body segmentation: perform segmentation on each single body of the upper and lower piles segmented above; firstly capture the highlight parts of the upper and lower piles, as the single heater will show reflective visual characteristics under the illumination of the annular light source; firstly perform highlight recognition on the single heater, and perform OTSU binarization, i.e., adaptive binarization, on the image based on highlight recognition; count the number of pixels with a grayscale value of 255 in each row, and when the number of pixels with a grayscale value of 255 in the first five rows is greater than 0, and the number of pixels with a grayscale value of 255 in the last three rows is equal to 0, it is a segmentation row for each single body; then segment the single body based on the segmentation row of each single body.

[0094] S2.3: Internal segmentation of the monomer, including single heater identification, composite sheet identification and negative electrode sheet identification: Since the single heater has reflective characteristics, each monomer image is first subjected to highlight processing, and then adaptive binarization is performed; then, the number of pixels with a grayscale value of 255 in each row is counted. If the number is greater than 0, it is determined that the row contains a single heater; the grayscale value of the row where the single heater corresponds to each column in the previously intercepted monomer image is set to 0, that is, it is set to black; the image is binarized, and then the grayscale of each pixel is nonlinearly adjusted. The grayscale nonlinear adjustment method is:

[0095] a. For pixels whose grayscale value is greater than 0 and less than 60, set it to 0.5 times the current pixel grayscale value;

[0096] b. For pixels whose grayscale value is greater than or equal to 60 and less than 80, set it to 1.25 times the current pixel grayscale value;

[0097] c. For pixels whose grayscale value is greater than or equal to 80 and less than 120, set it to 1.5 times the current pixel grayscale value;

[0098] d. For pixels whose grayscale value is greater than or equal to 150 and less than 250, set it to twice the current pixel grayscale value;

[0099] After adjustment, the average grayscale value of each row of pixels is calculated. If the average grayscale value of the first four rows is 0 and the average grayscale value of the last four rows is greater than 0, it is identified as a single heating sheet; if the average grayscale value of the first four rows is less than 100 and the average grayscale value of the last four rows is greater than 100, it is identified as a composite sheet; if the average grayscale value of the first four rows is not 0 and the average grayscale value of the last four rows is greater than 0, it is identified as a negative electrode sheet.

[0100] S2.4: Output the assembly list based on the categories and quantities of each monomer structure identified.

[0101] Furthermore, the identification of each structure of the head and tail pile is specifically as follows:

[0102] For the structural recognition of the first and last piles, first capture the images of the non-highlight parts, convert them to grayscale, and then binarize them based on the white features of the gasket. Due to camera distortion, the images of cylindrical objects taken will appear tilted: the structure of the first pile image is slightly tilted from the upper left to the lower right, while the structure of the tail pile image is slightly tilted from the lower left to the upper right. Therefore, only the average grayscale value of the pixels in half of the column of the non-highlight part of each row is calculated.

[0103] For the first pile, the structure includes the end heating plate, the upper fixing band, the gasket, the ignition plate and the electric ignition head assembly; the visual features of the gasket, the ignition plate and the electric ignition head assembly are white, while the visual features of the end heating plate and the upper fixing band are black; these color features are used to distinguish the gasket from the end heating plate and the upper fixing band; the ignition plate is tightly combined with the electric ignition head assembly, and the thickness of the ignition plate is much smaller than the thickness of the electric ignition head assembly, and the combined thickness of the two is also much greater than the thickness of the gasket; therefore, the ignition plate and the electric ignition head assembly, as well as the gasket, are identified by these thickness differences.

[0104] For the tail pile, the identification method is similar to that of the upper pile; the tail pile includes a lower fixing belt, a gasket and an end heating plate; the visual feature of the gasket is white, while the visual features of the end heating plate and the lower fixing belt are black; the gasket, the end heating plate and the lower fixing belt are identified based on this feature.

[0105] Furthermore, the detection method of the battery stack to be detected is implemented by multi-angle template matching. The specific steps of multi-angle template matching are:

[0106] S3.1: Perform pyramid downsampling twice on the template battery stack image and the image of the battery to be matched with the template.

[0107] S3.2: Perform the first template matching between the unrotated template battery stack image and the template battery matching image, and record the suitability of the current match, which is regarded as the most suitable match.

[0108] S3.3: First, rotate the template battery stack image 5° to the left, and then perform template matching. Compare the obtained matching fitness with the most suitable template matching fitness. If it is greater than the most suitable template matching stack, record the current angle and let the obtained matching fitness replace the most suitable template matching fitness. Rotate right 0.1° in sequence, and then perform the above comparison process to obtain the maximum matching fitness and record the angle of rotation of the template battery.

[0109] After obtaining the battery stack, carry out the correct thermal battery detection template detection process, and compare the detection data with the correct template battery stack detection data. If any inconsistency is found, it will be marked in red in the assembly list output table.

[0110] At this point, the basic process of the thermal battery defect detection algorithm proposed by this system can be summarized as follows: Figure 5 shown.

[0111] Furthermore, the thermal battery software interface is based on the Visual C++ language on the Windows 11 system and is completed on the QT platform. Figure 6 As shown, it includes four interfaces: visualization window interface, data recording interface, operation bar interface, and assembly list output interface.

[0112] The visualization window interface is used for real-time display of hot batteries photographed by high-pixel industrial cameras, display of the battery stack in the hot battery, and selection of the shooting template to obtain the correct template battery stack.

[0113] The data recording interface is used to record the thermal battery detection log, record the thermal battery stack detection time, detect the thermal battery stack model, and the detection results.

[0114] The assembly list output interface is used to display the order and quantity of each structure of the thermal battery.

[0115] The operation bar interface is used to adjust various parameters and control image capture.

[0116] Furthermore, the interactive software operation process of the online detection platform is as follows Figure 7 As shown, specifically:

[0117] S4.1: In the thermal battery production line, place the assembled but unpackaged thermal battery semi-finished products on the drawer board fixture of the equipment.

[0118] S4.2: Close the drawer panel, click "Template" and select "Photo Template".

[0119] S4.3: Select the hot battery stack in the visualization window and click the “Detect Button”.

[0120] S4.4: After the assembly list is generated, click “Save” to save the template data for inspection.

[0121] S4.5: Click the "Detect" button under "Battery Parameters" to start shooting.

[0122] You only need to save one template for each model, and you don't need to re-test the template every time you shoot. When the data of the battery to be tested does not match the data of the correct template, it will be displayed in red in the list.

[0123] The present invention has the following characteristics:

[0124] 1. The method for collecting thermal battery image data by the online thermal battery assembly defect detection system proposed in the present invention is to obtain it through a high-pixel industrial camera instead of a CT scanning device, which can largely avoid radiation and harm to the human body and save time.

[0125] 2. An online thermal battery assembly defect detection system is proposed, which can be directly used on the thermal battery assembly line. If a thermal battery assembly defect is found, the assembly order of the battery cells can be changed in time to make the battery correct, which can greatly reduce production costs and production time.

[0126] The present invention proposes a thermal battery online assembly defect detection system based on machine vision for thermal battery detection. It is a brand-new detection system that can be used on the thermal battery assembly line to directly detect the semi-finished products of the battery. Different from the thermal battery monomer detection method based on CT scanning proposed by others before, the finished product of the thermal battery is inspected. If the thermal battery assembly is detected to be defective, the thermal battery can only be scrapped and cannot be modified. The system proposed by the present invention can modify the thermal battery assembly defects in a timely manner if they are detected, thus saving production costs and improving economic efficiency. Then the detection platform designed by the present invention is specially designed for detecting thermal battery assembly defects. Since the appearance shape of the thermal battery to be inspected is a cylindrical columnar structure, the fixture is designed as a V-groove fixture to facilitate the flat placement of the thermal battery stack.

Claims

1. A thermal battery online assembly defect detection system based on machine vision, characterized in that: Includes an online testing platform for thermal battery defect detection, built-in thermal battery defect detection algorithm and thermal battery software interactive interface; The online detection platform is a box structure, including a thermal battery fixture group, a light source lighting system, an optical imaging system, an image acquisition system and a computer processing system; The thermal battery fixture group is a drawer board (1) arranged at the bottom of the box structure, and the thermal battery fixture (2) is installed in the middle of the drawer board (1) for fixing the thermal battery (3); the light source lighting system includes a small ring light source (5) and a large ring light source (8) arranged on a bracket (4) inside the box body, and the small ring light source (5) and the large ring light source (8) can be adjusted up and down in the vertical direction; the optical imaging system includes a high-pixel industrial camera (6) and a lens, and the industrial camera (6) and the lens can be adjusted up and down in the vertical direction through an adjustment device (7); the industrial camera (6), the small ring light source (5) and the large ring light source (8) are arranged in sequence from top to bottom, with their centers on a line and vertically aligned with the thermal battery (3) being photographed; the image acquisition system includes a screen (9), and the screen (9) is used to display the output results of the thermal battery assembly defects and a window for the operator to interact with the system; the computer processing system has a built-in thermal battery defect detection algorithm; The built-in thermal battery defect detection algorithm includes a correct thermal battery detection template acquisition algorithm and a battery stack detection method to be detected; The thermal battery software interactive interface includes four interfaces, namely a visualization window interface, a data recording interface, an operation bar interface, and an assembly list output interface.

2. According to the machine vision-based thermal battery online assembly defect detection system of claim 1, it is characterized in that: The drawer plate (1) is designed with equidistant multiple circular holes, the thermal battery clamp (2) is designed as a V-shaped groove, and the drawer plate (1) and the thermal battery clamp (2) are made of plastic.

3. According to the machine vision-based thermal battery online assembly defect detection system of claim 1, it is characterized in that: The industrial camera (6) is a model MV-CH310-10TC-M58-NN industrial camera produced by Hikvision Robotics, the lens model is MVL-LF3528-F, and the focal length is 35 mm.

4. According to the machine vision-based thermal battery online assembly defect detection system of claim 1, it is characterized in that: The correct hot battery detection template acquisition algorithm is specifically as follows: Use a high-pixel industrial camera to capture the thermal battery image, manually select and save the image of the thermal battery stack; next, grayscale the battery stack image, identify the highlight part of the single heater, and obtain an image containing the highlight part; Then, the OTSU binarization algorithm is used for adaptive binarization processing, and the image pixels greater than the threshold are set to white, and the pixels less than the threshold are set to black; finally, the noise in the binary image is removed through opening and closing operations such as corrosion and expansion; Traverse each row of the binary image and obtain the width of the highlight part of each row and the number of the leftmost columns; Calculate the width of the highlight part of all rows and the median of the leftmost column number, which are used as the width and starting point of the highlight part respectively; Then, extend the width of the highlight part to the left to cut out the area including the highlight part and the non-highlight part from the image; The non-highlight part of the intercepted image is processed to achieve the segmentation of the first pile, the last pile and the middle pile; then the middle pile is processed and the structures of the first and last piles are identified.

5. The machine vision-based thermal battery online assembly defect detection system according to claim 4 is characterized in that: The specific steps of dividing the first pile, the tail pile and the middle pile are: S1.1: Grayscale the non-highlight part of the battery stack, then select a suitable threshold based on the white pad for binarization, and then perform an opening and closing operation to remove noise; S1.2: Identification of the last gasket of the first pile: perform local processing on the first sixth of the rows of the binary image, delineate each connected domain and calculate the area, and retain the three connected domains with the largest area; since the first pile contains two regional gaskets and one ignition plate and electric ignition head assembly, the thickness difference with the white negative electrode sheet is relatively large, the last connected domain is the last gasket of the first pile; S1.3: Identification of the first gasket of the tail pile: The last one-sixth of the rows of the binary image are processed in a similar way to the identification of the last gasket of the first pile. The tail pile has three regional gaskets, which are also quite different in thickness from the white negative electrode sheet. The three connected domains with the largest areas are retained, and the first connected domain is the first gasket of the tail pile. S1.4: Division of the head, tail and middle piles: The part above the bottom of the connected domain of the last gasket of the first pile is the first pile; the part below the bottom of the connected domain of the last gasket of the first pile and the part above the top of the connected domain of the tail pile are the middle pile; the part below the top of the connected domain of the tail pile is the tail pile.

6. The machine vision-based thermal battery online assembly defect detection system according to claim 4 is characterized in that: The specific steps of the intermediate stack processing are: S2.1: Identification of the middle lead-out sheet and segmentation of the upper and lower piles: Due to the structure of the middle lead-out sheet, the visual feature is yellow, and the lead-out sheet is located in the middle of the entire middle pile. One tenth of the middle of the middle pile is intercepted for local area analysis. First, the local area is converted into an HSV image, and then the location of the lead-out sheet is identified based on the specific features of the three HSV channels shown by the yellow color. The average pixel value of each row is calculated, and the row with the largest average value is the upper and lower pile segmentation row; the part above the row with the largest average value is the upper pile, and the part below is the lower pile; S2.2: Single-unit segmentation: Segment each single unit of the upper and lower piles segmented above; first intercept the highlight part of the upper and lower piles, because the single heater sheet will show reflective visual characteristics under the illumination of the annular light source; first perform highlight recognition on the single heater sheet, and perform OTSU binarization, i.e. adaptive binarization, on the image based on highlight recognition; Count the number of pixels with a grayscale value of 255 in each row. When the number of pixels with a grayscale value of 255 in the first five rows is greater than 0, and the number of pixels with a grayscale value of 255 in the last three rows is equal to 0, it is a segmentation row for each monomer; then segment the monomers based on the segmentation row of each monomer; S2.3: Internal segmentation of the monomer, including identification of single heater, composite sheet and negative electrode sheet: Since the single heater has reflective characteristics, each monomer image is first subjected to highlight processing, and then adaptive binarization is performed; then, the number of pixels with a grayscale value of 255 in each row is counted, and if the number is greater than 0, it is determined that the row contains a single heater; the grayscale value of the row where the single heater corresponds to each column in the previously intercepted monomer image is set to 0, that is, it is set to black; the image is binarized, and then the grayscale of each pixel is nonlinearly adjusted; Grayscale nonlinear adjustment method is: a. For pixels whose grayscale value is greater than 0 and less than 60, set it to 0.5 times the current pixel grayscale value; b. For pixels whose grayscale value is greater than or equal to 60 and less than 80, set it to 1.25 times the current pixel grayscale value; c. For pixels whose grayscale value is greater than or equal to 80 and less than 120, set it to 1.5 times the current pixel grayscale value; d. For pixels whose grayscale value is greater than or equal to 150 and less than 250, set it to twice the current pixel grayscale value; After adjustment, the average grayscale value of each row of pixels is calculated. If the average grayscale value of the first four rows is 0 and the average grayscale value of the last four rows is greater than 0, it is identified as a single heating sheet; if the average grayscale value of the first four rows is less than 100 and the average grayscale value of the last four rows is greater than 100, it is identified as a composite sheet; if the average grayscale value of the first four rows is not 0 and the average grayscale value of the last four rows is greater than 0, it is identified as a negative electrode sheet; S2.4: Output the assembly list based on the categories and quantities of each monomer structure identified.

7. The machine vision-based thermal battery online assembly defect detection system according to claim 4 is characterized in that: The identification of each structure of the head and tail pile is specifically as follows: For the structure recognition of the first and last piles, firstly, the images of the non-highlight part are intercepted and grayscaled, and then binarized based on the white features of the gasket; due to camera distortion, the image of the cylindrical object will be tilted: the structure of the first pile image is slightly tilted from the upper left to the lower right, while the structure of the last pile image is slightly tilted from the lower left to the upper right; therefore, only the average grayscale value of the pixels in the half column of the non-highlight part of each row is calculated; For the first pile, the structure includes the end heating plate, the upper fixing belt, the gasket, the ignition plate and the electric ignition head assembly; the visual features of the gasket, the ignition plate and the electric ignition head assembly are white, while the visual features of the end heating plate and the upper fixing belt are black; these color features are used to distinguish the gasket from the end heating plate and the upper fixing belt; the ignition plate is tightly combined with the electric ignition head assembly, and the thickness of the ignition plate is much smaller than the thickness of the electric ignition head assembly, and the combined thickness of the two is also much greater than the thickness of the gasket; therefore, the ignition plate and the electric ignition head assembly, as well as the gasket, are identified by these thickness differences; For the tail pile, the identification method is similar to that of the upper pile; the tail pile includes a lower fixing belt, a gasket and an end heating plate; the visual feature of the gasket is white, while the visual features of the end heating plate and the lower fixing belt are black; the gasket, the end heating plate and the lower fixing belt are identified based on this feature.

8. The machine vision-based thermal battery online assembly defect detection system according to claim 1 is characterized in that: The battery stack detection method to be detected is implemented by multi-angle template matching, and the specific steps of multi-angle template matching are: S3.1: Perform pyramid downsampling twice on the template battery stack image and the image of the battery to be matched; S3.2: Perform the first template matching on the image of the battery stack without the rotation template and the image of the battery to be matched with the template, and record the suitability of the current matching, which is regarded as the most suitable matching degree; S3.3: First, the template battery stack image is rotated 5° to the left, and then template matching is performed. The obtained matching fitness is compared with the most suitable template matching fitness. If it is greater than the most suitable template matching stack, the current angle is recorded, and the obtained matching fitness replaces the most suitable template matching fitness. Rotate right by 0.1° in sequence, then perform the above comparison process to obtain the maximum match and record the rotation angle of the template battery; After obtaining the battery stack, carry out the correct thermal battery detection template detection process, and compare the detection data with the correct template battery stack detection data. If any inconsistency is found, it will be marked in red in the assembly list output table.

9. The machine vision-based thermal battery online assembly defect detection system according to claim 1, characterized in that: The thermal battery software interactive interface is based on Visual C++ language on Windows 11 system and completed on QT platform, including four interfaces: The visualization window interface is used for real-time display of hot batteries photographed by high-pixel industrial cameras, display of the battery stack in the hot battery, and selection of the shooting template to obtain the correct template battery stack; The data recording interface is used to record the thermal battery detection log, record the thermal battery stack detection time, detect the thermal battery stack model, and the detection results; Assembly list output interface, used to display the order and quantity of each structure of the thermal battery; The operation bar interface is used to adjust various parameters and control image capture.

10. The machine vision-based thermal battery online assembly defect detection system according to claim 9, characterized in that: The interactive software operation process of the online detection platform is as follows: S4.1: In the thermal battery production line, place the assembled but unpackaged thermal battery semi-finished products on the drawer board fixture of the equipment; S4.2: Close the drawer, click "Template" and select "Photo Template"; S4.3: Select the hot battery stack in the visualization window, and then click the "Detect button"; S4.4: After the assembly list is generated, click "Save" to save the template data for detection; S4.5: Click the "Detect" button under "Battery Parameters" to start shooting.

Images