A method, system, device and storage medium for detecting welding quality of sealing pins

Abstract

The present invention discloses a sealing pin welding quality detection method, system, equipment and storage medium, the method comprising: obtaining a two-dimensional image of a power battery, performing image recognition on the two-dimensional image to determine whether the power battery has surface defects; when it is determined that the power battery has surface defects, commanding a tilted 3D camera to scan the surface of the power battery to obtain a three-dimensional image; locating the weld area of ​​the power battery multiple times according to the depth information of the three-dimensional image, performing partition detection on the weld area to determine whether the weld area has depth defects, and displaying the defect results. The present invention uses a combination of 2D and 3D machine vision technology to analyze the welding condition of the sealing pins on the power battery, improve the comprehensiveness of the sealing pin welding quality detection, and reduce the occurrence of missed detection.

Description

Technical Field

[0001] The present invention relates to the field of power battery welding quality detection, and in particular to a sealing pin welding quality detection method, system, equipment and storage medium. Background Art

[0002] At present, in the field of power lithium batteries, the existing sealing nail weld quality inspection methods are mainly the following:

[0003] (1) Manually observe the real-time magnified image of the weld on the screen of the microscope camera. When the product passes directly under the microscope, the camera captures the weld image and transmits it to the display screen. The worker observes the weld image on the display screen to determine whether there are defects in the weld. If there are defects, the product is placed in the NG position. If there are no defects, the product is moved and the image of the part that has not yet been imaged is continued. This method is affected by factors such as the worker's work experience, fatigue level, and worker mood, and is prone to missed kills and over-kills. The detection process is difficult to control, the cost is high, and it is difficult to achieve automated monitoring of the production process;

[0004] (2) 2D machine vision automated judgment, that is, the product moves to the bottom of the camera, the machine vision light source is turned on, the product weld is imaged, and then the computer displays, processes, calculates and judges whether there are defects in the collected image. If there are defects, the product is placed at the defective position, otherwise it continues to the next station. This method is easily affected by the instability of the light source and is limited by the inability to detect height information characteristics. For example, defects such as poor flatness of sealing pins and high welds cannot be solved by 2D machine vision, resulting in high miss kill and over kill rates, and quality cannot be guaranteed;

[0005] (3) 3D machine vision automated judgment: When the product moves under the laser line of the 3D camera, the camera is triggered to collect the product weld image and send the image to the host computer. The computer displays the collected point cloud data, converts it into a depth image, processes, calculates and analyzes the image, and then determines whether there are defects in the weld. This method can detect defects with certain depth information, but it cannot judge defects that are not sensitive to depth information, such as cracks and weld black. In addition, when the weld is just completed, the weld surface is relatively smooth. The laser of the 3D sensor is irradiated to the weld surface, which is easy to produce diffuse reflection, thereby forming small pits on the weld surface, causing misjudgment; and the 3D sensor is easily affected by signal interference in the direction of movement, resulting in certain errors in weld detection and unable to improve the detection accuracy. Summary of the invention

[0006] In order to overcome the deficiencies of the prior art, one of the objectives of the present invention is to provide a sealing pin welding quality detection method, which can improve the accuracy of sealing pin welding quality detection.

[0007] A second object of the present invention is to provide a sealing pin welding quality detection system.

[0008] A third object of the present invention is to provide an electronic device.

[0009] A fourth object of the present invention is to provide a computer-readable storage medium.

[0010] One of the purposes of the present invention is achieved by the following technical solution:

[0011] A sealing nail welding quality detection method, comprising:

[0012] Acquire a two-dimensional image of a power battery, and perform image recognition on the two-dimensional image to determine whether the power battery has surface defects;

[0013] When it is determined that the power battery has a surface defect, an order is given to control the tilted 3D camera to scan the surface of the power battery to obtain a three-dimensional image;

[0014] The weld area of ​​the power battery is located multiple times according to the depth information of the three-dimensional image, the weld area is partitioned and detected to determine whether there is a deep defect in the weld area, and the defect results are displayed.

[0015] Furthermore, the 3D camera is tilted downward at an angle of 5° to 10°.

[0016] Furthermore, before obtaining the three-dimensional image, the method further includes:

[0017] The point cloud data obtained by scanning with the 3D camera is collected, position compensation is performed on the point cloud data according to the tilt angle of the 3D camera, and then the position-compensated point cloud data is spliced ​​to generate the three-dimensional image.

[0018] Further, the method for locating the weld area includes:

[0019] Based on the depth information of the three-dimensional image, the average height values ​​of the sealing pins and the outer shell on the power battery are calculated respectively, and the flatness of the sealing pins and the outer shell are evaluated to generate an evaluation result; and based on the average height of the outer shell, the height of the sealing pin is extracted to determine the sealing pin area, and it is converted into a binary image and then the main body is separated to obtain a preliminary weld area.

[0020] Furthermore, the method for locating the weld area further comprises:

[0021] A profile contour line is generated in the preliminary weld area, and the inner and outer edge points of the weld are determined according to the height information of the profile contour line to complete the weld positioning.

[0022] Furthermore, the method for determining whether there is a deep defect in the weld area includes:

[0023] The characteristic information of the profile contour line is identified, and the characteristic information of the profile contour line is analyzed to determine whether there is a deep defect in the weld area.

[0024] Furthermore, the method for determining whether there is a deep defect in the weld area also includes:

[0025] Defect judgment is performed on the designated position of the weld based on the pre-built weld defect model to identify whether there is a deep defect in the weld area.

[0026] The second object of the present invention is achieved by adopting the following technical solution:

[0027] A sealing pin welding quality detection system is provided, which performs the sealing pin welding quality detection method as described above; the system comprises:

[0028] A two-dimensional vision module, used to photograph the power battery using a 2D camera to obtain a two-dimensional image;

[0029] A three-dimensional vision module, used for controlling a tilted 3D camera to scan the surface of the power battery to obtain a three-dimensional image;

[0030] The defect analysis module is used to perform image recognition on the two-dimensional image to determine whether the power battery has surface defects; at the same time, the weld area of ​​the power battery is repeatedly located according to the depth information of the three-dimensional image, the weld area is partitioned and detected to determine whether the weld area has deep defects, and the defect results are displayed.

[0031] The third object of the present invention is achieved by adopting the following technical solution:

[0032] An electronic device comprises a processor, a memory and a computer program stored in the memory and executable on the processor, wherein the processor implements the sealing pin welding quality detection method as described above when executing the computer program.

[0033] The fourth object of the present invention is achieved by adopting the following technical solution:

[0034] A computer-readable storage medium stores a computer program, which, when executed, implements the sealing pin welding quality detection method as described above.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] The present invention adopts a combination of 2D and 3D machine vision technology to analyze the welding conditions of the sealing pins on the power battery, thereby improving the comprehensiveness of the sealing pin welding quality inspection and reducing the occurrence of missed inspections. At the same time, the 3D camera receiving end is installed in a downward tilted manner, which can reduce the raised interference signal generated by the 3D camera in motion and improve the detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a schematic diagram of the overall process of the battery quality detection method of the present invention;

[0038] Figure 2 is the original image of the sealing nail of the present invention;

[0039] Figure 3 This is a schematic diagram of the binarization result of the sealing nail image of the present invention;

[0040] Figure 4 It is a schematic diagram of the contour extraction result of the present invention;

[0041] Figure 5 This is a schematic diagram of the weld positioning of the present invention;

[0042] Figure 6 This is a schematic diagram of the arc ring straight line of the present invention;

[0043] Figure 7 It is a schematic diagram of a single weld contour line of the present invention;

[0044] Figure 8 for Figure 7 Schematic diagram of the cross-sectional distribution of the weld contour line;

[0045] Fig. 9 It is a schematic diagram of the flow of the sealing pin welding quality detection method of the present invention. DETAILED DESCRIPTION

[0046] The present invention is further described below in conjunction with the accompanying drawings and specific implementation methods. It should be noted that, under the premise of no conflict, the various embodiments or technical features described below can be arbitrarily combined to form a new embodiment.

[0047] Embodiment 1

[0048] The present embodiment provides a method for inspecting the welding quality of sealing pins. The method combines 2D machine vision and 3D machine vision to jointly judge the welding quality of sealing pins on a power battery, thereby reducing the occurrence of over-killing and under-killing and improving the accuracy of quality inspection.

[0049] refer to Figure 1 , Fig. 9 As shown, the sealing pin welding quality detection method of this embodiment includes the following steps:

[0050] Step S1: acquiring a two-dimensional image of a power battery, and performing image recognition on the two-dimensional image to determine whether the power battery has surface defects;

[0051] Step S2: when it is determined that the power battery has surface defects, commanding a tilted 3D camera to scan the surface of the power battery to obtain a three-dimensional image;

[0052] Step S3: Locate the weld area of ​​the power battery multiple times according to the depth information of the three-dimensional image, perform partition detection on the weld area to determine whether there is a deep defect in the weld area, and display the defect results.

[0053] In this embodiment, the target object, i.e., the power battery with the sealing nails welded, is pre-transmitted to the position directly below the imaging position of the 2D camera. When the arrival information of the target object is received, a command is given to control the light source to light up, and the 2D camera is triggered to shoot the target object to obtain a two-dimensional image.

[0054] The collected two-dimensional images are classified using deep learning methods to determine whether there are surface defects. The surface defects refer to defects that can be directly identified through 2D vision, such as cracks, small pinholes, and blackening. The results are sent to the control system; if there are defects, the target object is directly transferred to the NG position; if no surface defects are detected, the target object can continue to be moved to the 3D camera, and the sealing nails on the power battery are photographed with the 3D camera to obtain a three-dimensional image, and the three-dimensional image with depth information is used to further perform quality inspection on the weld area.

[0055] The 3D camera starts shooting from the sealing nail position, and scans the sealing nail point cloud data at a speed of 20mm / s. The 3D camera collects the sealing nail point cloud data according to the pulse signal of the encoder, and sends each part of the data to the industrial computer. After the 3D camera scans, the point cloud data collected each time is spliced ​​together. Then the point cloud data is converted into a depth image and displayed.

[0056] In this embodiment, the 3D camera is installed at a downward tilt angle, which can reduce the convex interference signal generated by the 3D camera in motion and improve the accuracy of subsequent weld quality inspection. At the same time, before processing the point cloud data, the collected point cloud data must be compensated for the installation position. The point cloud data compensation is performed by using a standard gauge block according to the installation method, that is, the 3D camera is first installed parallel to the horizontal plane, and then the value of the specified position is measured; then the 3D camera is installed according to the actual tilt angle, and the value of the specified position is measured again. By measuring the same position value twice, the installation angle compensation value is calculated, and then written into the camera controller to achieve installation angle compensation; before software processing, the camera is first corrected for the installation angle to avoid errors caused by installation. Among them, the tilt angle of the 3D camera is 5°~10°. In this embodiment, the receiving end of the 3D camera is installed with a downward tilt of 7°.

[0057] This embodiment uses a 3D camera to collect complete weld point cloud data to avoid point cloud data loss due to blind spots and high reflectivity on the weld surface. At the same time, the camera is installed at an angle to reflect some stray light to avoid the contour sensor receiving the stray light, thereby improving the quality of the collected point cloud data and providing a strong guarantee for subsequent point cloud data processing.

[0058] After obtaining the three-dimensional image scanned by the 3D camera, select several areas on the surface of the sealing nail, calculate the average height value in each area, and take the maximum value Pin_Max and the minimum value Pin_Min of the height of these areas; also select several areas on the surface of the battery shell, calculate the average height value in these areas, and take the maximum value Top_Max and the minimum value Top_Min of the average height of each area. Afterwards, calculate Abs(Pin_Max-Top_Min) and Abs(Pin_Min-Top_Max) respectively, use their values ​​to evaluate the flatness of the sealing nail and the surface of the battery shell, and display the evaluation results.

[0059] like Figure 2 , Figure 3 As shown, the height of the sealing nail is extracted based on the average height of the upper surface of the shell, and the extracted image is converted into an 8-bit grayscale image. The converted grayscale image is threshold processed. According to the grayscale value distribution of the grayscale image, the Otsu method is selected for threshold extraction, in order to separate the main body of the object from the background. For the binary image, the connection between the main bodies is separated. This embodiment uses the opening operation morphological processing to separate some weakly connected objects, and then extracts the main body contour to obtain the main body contour, as shown in FIG. Figure 4 shown.

[0060] The center of the main body contour is used as the sealing pin center, and the maximum ring width of the main body contour is used as the weld radius, such as Figure 5 As shown, an inner circular ring line and an outer circular ring line are generated according to the sealing pin center and the weld radius, and the area between the inner circular ring line and the outer circular ring line is marked as a preliminary weld area, thereby achieving the purpose of weld positioning.

[0061] like Figure 6 As shown, a number of arc ring straight lines are generated in the width direction of the preliminary weld area, and the gap between the straight lines is determined by the interval angle based on the detection defect accuracy. Therefore, before generating the arc ring straight lines, the user needs to customize the detection accuracy, obtain the customized detection accuracy parameters to determine the straight line gap, and evenly generate a number of the arc ring straight lines in the preliminary weld area according to the straight line gap.

[0062] The weld area depth image is divided into 4 areas, and the weld area analysis is performed separately. In each area, based on the height information of the weld, the depth data of the position of each arc ring straight line is assigned based on the depth information of the welding image to form the weld contour line; Figure 7 , Figure 8 As shown, according to the cross-sectional distribution of the weld contour line, the inside and outside of the weld are found, thereby achieving accurate positioning of the weld.

[0063] In the weld area, based on the mathematical model of the defect, the defect is searched along the weld direction, and the key feature points of the weld found during the search are recorded. Based on the feature information of the key feature points, it is judged whether the weld has a deep defect, that is, a weld defect that cannot be identified from the two-dimensional image, such as pits, bulges, surface flatness of sealing pins, broken welds, leaked welds, etc.; and the judgment results are displayed. Among them, the key feature points include the outermost point, innermost point, highest point and lowest point of the weld contour line; further, the feature information of the key feature points can also be obtained, including information such as the maximum height value of the weld, the minimum height value of the weld, and the width value of the weld.

[0064] Among them, the weld defect model can collect point cloud data of sample defects in advance according to customer inspection requirements, and perform mathematical modeling based on the specific form of the defects, so as to obtain the mathematical model of each sample defect, which provides the necessary conditions for weld defect search. The closer the mathematical model is to the actual defect form, the more accurate the subsequent defect search will be.

[0065] In areas where defects may occur, the key characteristic values ​​of the weld are used to process, calculate and judge according to the weld defect model. The length of the weld defect is calculated to determine whether it is a defect. Then a secondary judgment is made based on the characteristics of the weld to determine whether there is a defect.

[0066] Each divided area of ​​the weld area is processed in the above manner, and the defects in each area are counted separately, and then it is determined whether the entire weld has defects, and the judgment result is displayed on the display screen. If there are defects, an alarm is issued and the product is sent to the NG box; if it is a defect-free product, the product is sent to the next workstation.

[0067] This embodiment combines 2D machine vision with 3D machine vision to jointly judge weld defects. Since the weld has just been welded, the weld surface is relatively bright. When the laser is irradiated to the weld surface, the laser will produce diffuse reflection, which will cause loss when collecting point cloud data and form pits in the image, resulting in misjudgment. 2D machine vision can just make up for this defect. 2D imaging does not have such small pit defects, thereby improving the detection accuracy, making the defect judgment accuracy high and the misjudgment rate low. In addition, this embodiment directly searches for defects through the weld defect model. The entire process of finding defects uses 3D point cloud information, avoiding the misjudgment caused by converting 3D to 2D.

[0068] This embodiment uses 2D imaging for small pinholes, cracks and blackening defects, which improves the quality and resolution of image imaging; and adopts a deep learning method. The more samples are tested, the higher the accuracy of the judgment result.

[0069] Embodiment 2

[0070] This embodiment provides a sealing pin welding quality detection system, which executes the sealing pin welding quality detection method as described in the first embodiment; the system comprises:

[0071] A two-dimensional vision module, used to photograph the power battery using a 2D camera to obtain a two-dimensional image;

[0072] A three-dimensional vision module, used for controlling a tilted 3D camera to scan the surface of the power battery to obtain a three-dimensional image;

[0073] The defect analysis module is used to perform image recognition on the two-dimensional image to determine whether the power battery has surface defects; at the same time, the weld area of ​​the power battery is repeatedly located according to the depth information of the three-dimensional image, the weld area is partitioned and detected to determine whether the weld area has deep defects, and the defect results are displayed.

[0074] In some embodiments, an electronic device is also provided, which includes a processor, a memory, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the sealing pin welding quality detection method in Example 1 is implemented.

[0075] In some embodiments, a computer-readable storage medium is also provided, on which a computer program is stored. When the computer program is executed, the sealing pin welding quality detection method in Example 1 is implemented.

[0076] The system, device and storage medium in this embodiment and the method in the aforementioned embodiment are based on multiple aspects of the same inventive concept. The method implementation process has been described in detail above, so those skilled in the art can clearly understand the structure and implementation process of the system, device and storage medium in this embodiment based on the aforementioned description. For the sake of brevity, they will not be repeated here.

[0077] The above-mentioned embodiments are only preferred embodiments of the present invention and cannot be used to limit the scope of protection of the present invention. Any non-substantial changes and substitutions made by technicians in this field on the basis of the present invention shall fall within the scope of protection required by the present invention.

Claims

1. A sealing nail welding quality detection method, characterized in that: include: Acquire a two-dimensional image of a power battery, and perform image recognition on the two-dimensional image to determine whether the power battery has surface defects; When it is determined that the power battery has a surface defect, an order is given to control the tilted 3D camera to scan the surface of the power battery to obtain a three-dimensional image; Locating the weld area of ​​the power battery multiple times according to the depth information of the three-dimensional image, performing partition detection on the weld area to determine whether there is a deep defect in the weld area, and displaying the defect results; The method for locating the weld area of ​​the power battery includes: calculating the average height values ​​of the sealing pins and the outer shell on the power battery respectively based on the depth information of the three-dimensional image, and evaluating the flatness of the sealing pins and the outer shell to generate an evaluation result; and taking the average height of the outer shell as a reference, extracting the height of the sealing pin to determine the sealing pin area, and converting it into a binary image and performing main body separation to obtain a preliminary weld area; generating a profile contour line in the preliminary weld area, and determining the inner and outer edge points of the weld according to the height information of the profile contour line to complete the weld positioning.

2. The sealing nail welding quality detection method according to claim 1 is characterized in that: The 3D camera is tilted downward at an angle of 5° to 10°.

3. The sealing nail welding quality detection method according to claim 1 is characterized in that: Before obtaining the three-dimensional image, the method further includes: The point cloud data obtained by scanning with the 3D camera is collected, position compensation is performed on the point cloud data according to the tilt angle of the 3D camera, and then the position-compensated point cloud data is spliced ​​to generate the three-dimensional image.

4. The sealing nail welding quality detection method according to claim 1 is characterized in that: The method for determining whether there is a deep defect in the weld area includes: The characteristic information of the profile contour line is identified, and the characteristic information of the profile contour line is analyzed to determine whether there is a deep defect in the weld area.

5. The sealing pin welding quality detection method according to claim 1 is characterized in that: The method for determining whether there is a deep defect in the weld area also includes: Defect judgment is performed on the designated position of the weld based on the pre-built weld defect model to identify whether there is a deep defect in the weld area.

6. A sealing pin welding quality detection system, characterized in that: Execute the sealing pin welding quality detection method according to any one of claims 1 to 5; the system comprises: A two-dimensional vision module, used to photograph the power battery using a 2D camera to obtain a two-dimensional image; A three-dimensional vision module, used for controlling a tilted 3D camera to scan the surface of the power battery to obtain a three-dimensional image; The defect analysis module is used to perform image recognition on the two-dimensional image to determine whether the power battery has surface defects; at the same time, the weld area of ​​the power battery is repeatedly located according to the depth information of the three-dimensional image, the weld area is partitioned and detected to determine whether the weld area has deep defects, and the defect results are displayed.

7. An electronic device, characterized in that: It includes a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the sealing pin welding quality detection method described in any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed, the sealing pin welding quality detection method according to any one of claims 1 to 5 is implemented.

Images