Static bell jar welding method and welding robot

Abstract

The invention provides a standing bell jar welding method and a welding robot. The standing bell jar welding method comprises the steps that image collection is conducted on a to-be-welded workpiece; according to the single-frame welding line image, extracting welding line center line coordinates and welding line widths corresponding to all center line points; splicing a plurality of frames of images by adopting a weighted fusion mode to obtain a full-circle weld joint panoramic expansion view; welding seam areas with different widths are classified and positioned according to threshold values, corresponding area masks are generated, the masks are fused with the whole-circle welding seam panoramic expansion graph to achieve color coding visual marking, and a welding seam width marking graph is output; segmenting the welding seam width marking graph according to a preset welding seam width threshold value to obtain segmented graphs; and the image coordinate system in the segmented graph is compared with the workpiece coordinate system of the workpiece to be welded, and the welding speed is controlled according to the welding seam width corresponding to the segmented graph. According to the method, the influence of image distortion is reduced, the welding quality is improved by combining the welding seam width, and the wide application prospect is achieved.

Description

Technical Field

[0001] This disclosure pertains to the field of automated welding, and more specifically relates to a method for welding a stationary bell jar and a welding robot. Background Technology

[0002] The isobaric bell-type electrolyte injection step is an important step in lithium battery production. By using vacuum and isobaric injection methods, the wetting and absorption rate of electrolyte can be accelerated, the settling time can be shortened, and the injection efficiency can be greatly improved. Therefore, the settling bell is an essential piece of equipment in this step.

[0003] A stationary bell jar is a cylindrical workpiece whose weld quality directly determines its structural strength, sealing performance, and service life. Weld width, as a core indicator for evaluating welding quality, requires precise detection and calibration to identify defects such as excessive width or narrowness, and to pinpoint the specific location of these defects on the workpiece, providing a basis for subsequent rework or quality traceability.

[0004] However, in the current process of welding robots, the acquisition of weld seam images for rotating cylindrical workpieces is easily affected by the curvature of the cylindrical surface and perspective effect, resulting in severe distortion. This directly leads to low accuracy of weld seam feature extraction and large width measurement error. Moreover, under the premise of generating large measurement errors, the impact of the weld seam on the welding quality is not fully considered. Instead, attention is focused on reference factors such as welding current, resulting in unstable welding quality.

[0005] Therefore, a welding method that can eliminate image distortion, achieve full-circle weld reconstruction, and fully consider weld width is urgently needed for development. Summary of the Invention

[0006] To address the problems in the prior art, this disclosure provides a static bell jar welding method and a welding robot.

[0007] The technical solution adopted by this disclosure to solve the above-mentioned technical problems is as follows: A method for welding a stationary bell jar includes: Image acquisition is performed on the workpiece to be welded to obtain a single-frame weld image, and each single-frame weld image corresponds to the rotation angle of the workpiece to be welded. The coordinates of the weld centerline and the weld width corresponding to each centerline point are extracted from the single-frame weld image. Based on the workpiece rotation angle corresponding to each single frame image, the corrected single frame unfolded image is aligned to the whole circumference coordinate system, and multiple frames are stitched together using a weighted fusion method to obtain the whole circumference weld panoramic unfolded image. Preset calibration threshold, classify and locate weld seam areas of different widths according to the threshold, generate corresponding area masks, merge the masks with the full-circle weld panoramic unfolded map to achieve color-coded visual annotation, and output weld seam width annotation map; The weld width annotation diagram is segmented according to a preset weld width threshold to obtain a segmented diagram; The image coordinate system in the segmented diagram is compared with the workpiece coordinate system of the workpiece to be welded, and the welding speed is controlled according to the weld width corresponding to the segmented diagram.

[0008] The step of controlling the welding speed according to the weld width corresponding to the segmented diagram includes: The welding speed is obtained using the following formula: ; in, v For welding speed; b The average weld width in any segmented diagram; K This refers to the process coefficient; I This refers to the welding current. η For welding efficiency; d w The diameter of the welding wire; h This refers to the effective thickness of the weld. ρ This refers to the metal density of the welding wire.

[0009] The step of extracting the weld centerline coordinates and the weld width corresponding to each centerline point from the single-frame weld image includes: Geometric distortion correction is performed on the acquired single-frame images to eliminate perspective and cylindrical surface curvature distortion; Obtain a binary image of the weld, and extract the coordinates of the weld centerline and the weld width corresponding to each centerline point from the binary image by quantization. The step of quantizing and extracting the coordinates of the weld centerline and the weld width corresponding to each centerline point from the binary image includes: To obtain the center line of a single pixel, the core condition is 2 ≤ N ( p )≤6 and S ( p =1, output the set of centerline coordinates C ={( x ' i , y ' i )| i =1,2,..., N}; The edge points are searched along the vertical direction from each point on the centerline, and the width is calculated as follows: , get the pixel width; where, ( x 1i , y 1i ), ( x 2i , y 2i ) is the firsti The coordinates of the left and right edge points corresponding to the center line point; W i : No. i The physical width of the weld corresponding to each centerline point; Based on the workpiece rotation angle corresponding to each single-frame image, the corrected single-frame unfolded image is aligned to the whole-circumference coordinate system, and multiple frames are stitched together using a weighted fusion method to obtain a whole-circumference panoramic unfolded image of the weld seam, including: The single-frame unfolded image alignment formula is: x ' align = R ·( α + α i ),in x ' align To align the horizontal axis of the unfolded graph, α For the inner circumference angle of a single frame, α For the first i The frame corresponds to the workpiece rotation angle; image registration involves extracting centerline matching points and calculating the affine transformation matrix. M Achieve alignment; The weighted fusion formula is as follows: ; in w 1. w 2 represents the weight of the overlapping region. w 1= d 2 / ( d 1+ d 2), d 1. d 2 represents the distance from one pixel to the next frame; I 1( x ', y ') represents the first input source to be fused at coordinates ( x ', y The original data value at the ') position; I 2( x ', y ') represents the first input source to be fused at coordinates ( x ', y The original data value at ').

[0010] The calibration threshold is determined using an adaptive method, expressed as follows:

[0011] in, μ This represents the average width of the weld over the entire circumference. σ The standard deviation of width, T low For a threshold that is too narrow,T high The threshold is too wide; N This represents the total number of points along the centerline of the entire weld circumference. W i For the first i The weld width corresponding to each centerline point.

[0012] A welding robot, comprising: Column assembly; An active component, which can be vertically movable, is mounted on the image acquisition component; A welding execution component is disposed on the movable component and is used to drive the welding execution component to move vertically along the column assembly via the movable component; A support component is disposed below the column assembly to support the workpiece to be welded, thereby positioning the workpiece below the welding execution component. An image acquisition component is mounted on the column assembly and positioned above the workpiece to be welded, for acquiring images of the workpiece to be welded; A control component, electrically connected to the active component, image acquisition component, and welding execution component, is used to control the active component, image acquisition component, and welding execution component to perform welding of the bell jar stationary cavity using the stationary bell jar welding method described above.

[0013] The active components include: The movable component is movably mounted on the column assembly; A drive unit is disposed on the movable part and drives the movable part to move vertically on the column assembly; A support unit is disposed on the movable part; one end of the support unit is connected to the welding execution assembly to adjust the height between the welding execution assembly and the weld seam on the workpiece to be welded. A wire feeding mechanism, disposed on the support unit, is used to provide welding wire for the welding execution assembly.

[0014] The wire feeding mechanism includes: A rotating shaft mechanism is located at the other end of the support unit; The welding wire spool is mounted on the rotating shaft mechanism and can rotate along the rotating shaft mechanism; A wire feeding drive unit is disposed at one end of the support unit near the welding execution component, and is used to provide welding wire to the welding execution component.

[0015] The welding execution component includes: A slider is provided at the end of the support unit, which can move laterally; A lateral drive mechanism is provided on the support unit for driving the slider part; The welding torch is mounted on the slider and is used to move laterally along the support unit under the drive of the slider to adjust the lateral relative position of the welding torch and the weld seam on the workpiece to be welded.

[0016] The carrier component includes: The base is set on the ground; At least two rotating shaft mechanisms are mounted on the base; At least two support members are provided on the rotating shaft mechanism to support the workpiece to be welded; A rotary drive unit is connected to any of the aforementioned rotating shaft mechanisms for driving at least one rotating shaft mechanism to rotate, thereby causing the workpiece to be welded to rotate. The rotary drive unit is electrically connected to the control component and is used to control the welding speed of the workpiece to be welded by controlling the rotation speed of the rotating shaft mechanism according to the size of the weld.

[0017] Beneficial effects: The welding method described in this disclosure achieves geometric distortion correction through cylindrical surface unfolding, establishes multi-coordinate system mapping relationships, effectively eliminates image distortion caused by perspective and cylindrical surface curvature, and significantly improves the accuracy of weld feature extraction. Based on the workpiece rotation angle, it achieves single-frame unfolded image alignment, and combines feature point registration and weighted fusion techniques to solve the problem of multi-frame splicing misalignment, successfully constructing a 360° full-circumference weld panorama that completely covers all weld areas of the workpiece. Furthermore, it segments the full-circumference weld according to the weld width, applying different welding speeds to different segments to ensure welding quality. This method reduces the impact of image distortion and, combined with weld width, improves welding quality, showing broad application prospects. Attached Figure Description

[0018] Figure 1 This is a flowchart of the static bell jar welding method described in this disclosure; Figure 2 This is a schematic diagram of the structure of the welding robot described in this disclosure; Figure 3 for Figure 2 Enlarged view of point E in the image; Figure 4 for Figure 3 One embodiment of the slider part; Figure 5 for Figure 2 A schematic diagram of the structure of the load-bearing components; Figure 6 for Figure 5 A schematic diagram of the load-bearing components in the middle.

[0019] exist Figures 2-6 middle: 1. Column assembly; 2. Moving assembly; 3. Welding execution assembly; 4. Bearing assembly; 5. Image acquisition assembly; 6. Control assembly; 201. Moving part; 202. Drive unit; 203. Support unit; 204. Wire feeding mechanism; 301. Slider part; 302. Lateral drive mechanism; 303. Welding torch part; 401. Base; 402. Rotating shaft mechanism; 403. Support part; 404. Rotary drive unit; 2041. Rotary shaft mechanism; 2042. Welding wire spool; 2043. Wire feeding drive unit; 203 A. Groove; 3011. Guide rail; 3012. Connecting strip; 3013. Fixing plate. Detailed Implementation

[0020] The technical solutions in the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings.

[0021] To address the problems in the prior art, this disclosure provides the following embodiments. Specific Implementation Example 1: This disclosure provides an embodiment: like Figure 1 A method for welding a stationary bell jar includes the following steps: S 1. Preliminary preparation and image acquisition: A data acquisition system consisting of a coaxially arranged camera, a rotary table, and an encoder was constructed. The parameters of the bell jar stationary cavity, camera parameters, and image parameters were defined. The workpiece was controlled to rotate around the axis, and the camera synchronously acquired single-frame weld images. The workpiece rotation angle corresponding to each frame image was obtained in real time by the encoder. The bell jar stationary cavity was regarded as a cylinder and referred to as the workpiece.

[0023] Further, steps S The workpiece parameters mentioned in section 1 include the workpiece radius. R axial length L and rotational angular velocity ω The camera parameters include intrinsic and extrinsic parameters, with the intrinsic parameter being the pixel focal length. f x , f and principal point coordinates ( u 0, v 0), of which f x = f · s x , f = f · s ,f The physical focal length of the camera. s x , s The pixel size is the extrinsic parameter, and the extrinsic parameter is the height of the camera from the top surface of the workpiece. H Rotation matrix R and translation vector T Because the camera and the workpiece are arranged coaxially, the rotation matrix R Approximately the identity matrix I Translation vector T =[0,0, H The image parameters include image resolution. W × H , Horizontal × Vertical, Unit: pixels, Pixel Density k and the time interval of data collection t , where pixel density k = W / (2 πR ), characterizing 1 mm The number of pixels corresponding to the physical length.

[0024] Further, steps S In step 1, three types of coordinate systems are established to achieve coordinate mapping and ensure the accuracy of subsequent image processing. Specifically, these include: (1) Image coordinate system u , v ): represents pixel plane coordinates. u The axis is along the horizontal direction of the image. v The axis is perpendicular to the image, with the origin located at the top left corner of the image; (2) Camera coordinate system X , Y , Z (): With the optical center of the camera as the origin, Z The axis points perpendicularly to the upper surface of the workpiece. X Axis and Image Coordinate System u The axes are parallel. Y Axis and Image Coordinate System v The axes are parallel; (3) Cylindrical world coordinate system α , z ): α The workpiece circumferential angle, ranging from 0 to 2. π (unit: rad ), z The axial height of the workpiece is defined by the upper end face of the workpiece as the reference plane. z =0, downward along the workpiece axis z The value is negative, at the lower end face. z =- L ,unit: mm.

[0025] S 2. Weld feature extraction: Geometric distortion correction is performed on the acquired single-frame image to eliminate perspective and cylindrical surface curvature distortion. Then, after noise reduction, enhancement and segmentation processing, a binary image of the weld is obtained. The coordinates of the weld centerline and the weld width corresponding to each centerline point are extracted from the binary image by quantization. Further, steps S The geometric distortion correction described in section 2 adopts a cylindrical surface unfolding method and achieves distortion elimination through a three-step coordinate mapping, specifically including: S 2.1.1 Image and Camera Coordinates: Based on the inverse projection of the pinhole camera model, the image pixel coordinates ( u , v Mapped to camera 3D coordinates X , Y , Z The mapping formula is: ; in,( X c , Y c , Z c () represents the three-dimensional coordinates in the camera coordinate system, indicating the spatial position of a pixel with the camera's optical center as the origin; u , v ): Pixel coordinates in the image coordinate system. u Along the horizontal direction of the image, v Along the vertical direction of the image; u 0, v 0): The principal point coordinates in the camera's intrinsic parameters, i.e., the center pixel position of the image coordinate system; f x , f y ): Pixel focal length in camera intrinsic parameters; s x , s y () represents the pixel size; Z w The axial height coordinates are in the cylindrical world coordinate system, and are relative to the camera coordinate system. Z c Inverse correlation ( Z = - z ), z (Height of the cylindrical world coordinate system) H The height of the camera from the top surface of the workpiece is the external parameter of the camera, which is the vertical distance from the optical center of the camera to the top surface of the workpiece.

[0026] S 2.1.2 Camera → World Coordinates: Utilizing the constraint conditions of points on the surface of the cylindrical workpiece. X ²+ Y ²= R ², where X = X , Y = Y Since the camera and workpiece are coaxially arranged, and using trigonometric relationships, the mapping from camera coordinates to cylindrical world coordinates is derived as follows: ; in, R The workpiece radius is the cross-sectional radius of the rotating cylindrical workpiece in the workpiece parameters. α : Circumferential angle in cylindrical world coordinate system, representing the circumferential position of the workpiece; z w This is the axial height in the cylindrical world coordinate system, with the upper end face of the workpiece as the reference plane, and the value is negative as it is taken downwards along the axis.

[0027] S 2.1.3. Cylindrical Surface Unfolding: Cut the cylindrical surface along the generatrix and unwrap the circumferential angles. α Convert to unfolded length to obtain the coordinates of the unfolded planar diagram. x ', y To map the cylindrical weld seam to a plane, the formula is: ; in, x 'These are the coordinates along the circumference of the unfolded diagram. y 'The axis coordinates of the unfolded diagram are used. After correction, the spiral weld is converted into a straight line, and the annular weld is converted into a horizontal line segment, which facilitates subsequent feature extraction.

[0028] Further, steps S The preprocessing described in section 2 includes noise reduction and enhancement. Noise reduction employs a 3×3 kernel Gaussian filter to eliminate Gaussian noise introduced by the image sensor. The kernel function is: ; in, The Gaussian filter kernel in coordinates ( x , y The weight value at () is used to smooth the image and eliminate Gaussian noise; σ The standard deviation of the Gaussian kernel is used, with a value ranging from 1.0 to 2.0, balancing noise reduction and edge preservation; enhancement employs adaptive histogram equalization. CLAHE This method addresses the issue of low contrast between the weld seam and background caused by uneven lighting in welding scenes. Specifically, it divides the image into 8×8 sub-blocks and calculates the cumulative distribution function for each sub-block. CDF ( k ): ; in, For sub-blocks T medium grayscale value k The cumulative distribution function value at time is used to adjust the image grayscale distribution and enhance contrast; the enhanced pixel value is: ; T The sub-blocks are identified after the image is segmented, using an 8×8 sub-block division; k This represents the gray level within the sub-block, with a value range of 0-255; For sub-blocks T medium grayscale value i The number of pixels, i.e., the grayscale histogram of the sub-block; N T For sub-blocks T Total number of pixels, in 8×8 sub-blocks N T = 64.

[0029] In the steps S The segmentation described in section 2 uses adaptive threshold segmentation, where the threshold is dynamically adjusted according to local illumination conditions. The segmentation formula is as follows: ; Among them, adaptive threshold ; in, For the segmented binary image at pixel (( u , v The pixel value at () T ( u , v () is an adaptive threshold. T ( u , v )= μ ( u , v )- k · σ ( u , v After segmentation, a binary image is obtained with the weld seam as the foreground pixel value of 255 and the background as the workpiece substrate with a pixel value of 0. for CLAHE Enhanced grayscale value; T ( u , v ): pixel ( u , v The adaptive threshold at point () is dynamically adjusted according to local lighting conditions; μ ( u , v) is pixels ( u , v The average gray value of a 3×3 local neighborhood; For pixels ( u , v ) 3×3 local neighborhood grayscale standard deviation; k The threshold adjustment coefficient, ranging from 0.8 to 1.2, controls the tightness of the segmentation. Morphological closing operations are then used to complete the weld fracture region, resulting in a complete binary image of the weld. Morphological closing operations include dilation and erosion.

[0030] Further, steps S The weld feature quantification extraction described in section 2 includes centerline extraction and width calculation, specifically: using... Zhang - Suen The algorithm performs skeleton thinning on the binary image of the weld, preserving the weld centerline with a width of one pixel. The core condition for thinning is: 2 ≤ N ( p )≤6, N ( p (pixels) p The number of foreground pixels in the 8-neighborhood, and S ( p )=1, S ( p (pixels) p The neighborhood connectivity is used to ensure that the points are edge points. The final output is the set of centerline coordinates. C ={( x ' i , y ' i )| i =1,2,..., N}, N The total number of points on the centerline; along the direction perpendicular to each point on the centerline, i.e., the weld width direction, and perpendicular to the tangent direction of the centerline, search for the left and right edge points of the weld, and calculate the weld width at that point using the distance formula between the two points: Pixel width conversion formula: ;in,( x 1i , y 1i ), ( x 2i , y 2i ) is the first i The coordinates of the left and right edge points corresponding to the center line point; W i : No. i The physical width of the weld corresponding to each centerline point, that is, the straight-line distance between the left and right edge points; w i : No.i The weld pixel width corresponding to each centerline point; k Pixel density in the image parameters; i The index of the centerline point. i =1, 2, ..., N ; N This represents the total number of points along the center line.

[0031] S 3. Reconstruction of the complete circumference panoramic view of the weld: Based on the workpiece rotation angle corresponding to each single frame image, the corrected single frame unfolded image is aligned to a unified circumference coordinate system. The misalignment caused by mechanical error is eliminated by image registration. Multiple frames are stitched together using a weighted fusion method, and a 360° full circumference panoramic unfolded image of the weld is obtained after post-processing. S 4. Width-based differential calibration of weld areas: Based on the whole-circle weld width data, the calibration threshold is determined, weld areas of different widths are classified and located according to the threshold, corresponding area masks are generated, and the masks are integrated with the panoramic image to achieve color-coded visual annotation, and structured data and calibration reports are exported simultaneously.

[0032] Further, steps S The single-frame unfolded image alignment described in section 3 is achieved based on the workpiece rotation angle, and the image acquisition time for each frame... t Corresponding workpiece rotation angle α = ω · t The alignment formula is: ; in, x ' align To align the coordinates of the circumference of the unfolded diagram after alignment. α For the circumferential angles within a single frame image, alignment maps the unfolded images of different frames to a unified integer coordinate system. x The range is 0~2 πR (corresponding to the 0~360° circumference of the workpiece). α i For the first i The workpiece rotation angle corresponding to the frame image; Image registration employs feature point matching to extract matching point pairs of weld centerlines from adjacent frames, and then calculates the affine transformation matrix using the least squares method. M : ; in, a 11 , a 12 , a 21 , a 22This is the rotation scaling factor. t x , t y The translation coefficient is used to achieve precise alignment of the unfolded images of adjacent frames, eliminating misalignment caused by mechanical rotation errors; x ' 2,align , y ' 2,align () represents the coordinates of the second frame image after registration in a unified coordinate system; x '2, y '2) The coordinates of the unfolded image before registration of the second frame image; M Here is the affine transformation matrix; t x , t y ) represents the translation coefficient in the matrix.

[0033] Multi-frame stitching uses a weighted fusion method, which calculates a weighted average of pixel values ​​in overlapping areas. The fusion formula is as follows: ; in, w 1. w 2 represents the weight of pixels in the overlapping region. w 1= d 2 / ( d 1+ d 2) w 2= d 1 / ( d 1+ d 2), d 1. d 2 represents the distance from the pixel to the edge of the two frames, and weighted fusion is used to eliminate the stitching seam; I pan ( x ', y '): The stitched panoramic image is located at coordinates ( x ', y The grayscale value at '); x ', y ') represents the coordinates of the unfolded diagram; I 1( x ', y '), I 2( x ', y ') represents the coordinates of two adjacent frames ( x ', y The grayscale value at '); w 1, w 2) Pixel weights in the overlapping region of two adjacent frames. After stitching, the image is smoothed using a Gaussian smoothing mechanism. σ =1.5, 3×3 cores, and CLAHEAfter equalization and post-processing, a 360° panoramic view of the weld seam with clear texture and no distortion is obtained.

[0034] Further, steps S The calibration threshold mentioned in section 4 is determined using an adaptive method, calculated based on the statistical characteristics of the entire weld width data, avoiding the subjectivity of manual parameter adjustment. The specific formula is as follows: ; in, μ This represents the average width of the weld over the entire circumference. σ The standard deviation of width, T low For a threshold that is too narrow, T high To avoid excessively wide thresholds, welds are classified into three categories based on these thresholds: excessively narrow regions ( W < T low ), normal area ( T low ≤ W ≤ T high ), excessively wide area ( W > T high ); N This represents the total number of points along the centerline of the entire weld circumference. W i For the first i The weld width corresponding to each centerline point.

[0035] Further, steps S The region localization described in section 4 is achieved by generating a category mask, filtering the centerline point set according to the width category, and obtaining the excessively narrow point set. C low , normal point set C w and wide point set C high For each point set, a binary mask is generated using polygon filling based on the corresponding edge point coordinates. After morphological closing operations are performed to fill in the holes, the resulting composite mask is obtained. Mask ,in Mask =1 indicates an excessively narrow region. Mask =2 indicates an excessively wide region. Mask =3 indicates the normal area. Mask =0 indicates background.

[0036] Further, steps S The visual annotations described in section 4 consist of three parts: (1) Color coding: The integrated mask is combined with the panoramic image. Red is used to mark areas that are too narrow, yellow is used to mark areas that are too wide, and green is used to mark areas that are normal, so as to intuitively distinguish different width categories. (2) Information labeling: Label the width value at the center of each area (e.g., "Too narrow: 2.3"). mm "Too wide: 6.1" mm The text includes the symbols and their physical locations, which are obtained through coordinate transformation of the unfolded diagram, and the circumferential angle. ;in, x ' represents the circumferential coordinates of the unfolded diagram.

[0037] (3) Auxiliary annotations: Category legends can be added to the corners of the panoramic image, such as color-category correspondence and length scale, to ensure that the annotation information is complete.

[0038] Further, steps S The structured data described in section 4 includes regions. ID Width category, average width, maximum width, minimum width, circumferential angle range, axial height range, and area; export format is as follows: CSV or Excel Simultaneously export 16-bit single-channel mask file and PDF The calibration report summarizes calibration parameters and statistical results, such as the percentage of normal areas, the number of abnormal areas, a visual panoramic view, and quality assessment conclusions, providing support for quality traceability.

[0039] S 5. According to the steps S The weld seam is segmented based on the excessively narrow, normal, and excessively wide regions obtained in step 4. The coordinate system of the segmented image is compared with the workpiece coordinate system of the workpiece to be welded, and the welding speed of the welding machine is controlled according to the segmentation result, as shown in the following formula: ; in, v For welding speed; b The weld width in any segmented diagram; K This refers to the process coefficient; I This refers to the welding current. η For welding efficiency; d w The diameter of the welding wire; h This refers to the effective thickness of the weld. ρ This refers to the metal density of the welding wire.

[0040] The process of comparing the image coordinate system in the segmented diagram with the workpiece coordinate system of the workpiece to be welded is as follows: Depending on the dimension of the welding scenario (planar / spatial), homography matrix or... PnPThe algorithm establishes a mapping model between the image coordinate system and the workpiece coordinate system: This is suitable for scenarios where the plane of the weld is parallel to the camera's imaging plane (such as butt welding of flat plates). The core is to establish a 2D model through the homography matrix. D Image coordinates and 2 D Workpiece coordinate mapping, specific steps: Images of the workpiece with calibration marks are captured using a camera fixed to the end effector or frame of the welding robot, ensuring that all marker points of the calibration marks are visible and unobstructed. Adaptive threshold segmentation combined with morphological processing is used to enhance image contrast. cv 2. findChessboardCorners The `()` function extracts checkerboard marker points, or extracts the array of marker points using Hough circle detection, and then... cv 2. cornerSubPix The `()` function performs subpixel optimization and obtains the image coordinates of the marker points. u i , v i Positioning accuracy ≤ 0.1 pixels; establish the correspondence between the marker points and the workpiece coordinates. The workpiece coordinates of the marker points are known (…). X i , Y i The homography matrix is ​​solved using the least squares method. H (3×3), satisfying: ; For any weld feature point in the image ( u , v ), through the homography inverse matrix H - ¹Convert to workpiece coordinates ( X , Y The expression is: ; H It is a 3×3 matrix; A cubic calibration block is used, with at least 6 non-coplanar markers set on its surface. The 3D values ​​of each marker are accurately measured. D Workpiece coordinates ( X i , Y i , Z i ); Input marker 3 D Workpiece coordinates ( X i , Y i , Z i ) and 2 DImage coordinates ( u i , v i ),pass EPNP The algorithm solves for the camera's extrinsic parameters, satisfying the camera's imaging projection model: ;in, s As a scale factor, K This is the camera intrinsic parameter matrix; when mapping from image to workpiece, the equation of the plane containing the weld is known and can be obtained by fitting the calibration object's marker points, such as... Z = aX + bY + c Solving by using the inverse matrix of the extrinsic parameters ( X , Y ), Z The values ​​are determined by the plane equations; furthermore, at least three sets of images of the calibration object in different postures are captured, and the average value of the extrinsic parameters is calculated to reduce the single-posture calibration error. The process of comparing the image coordinate system in the segmented diagram with the workpiece coordinate system of the workpiece to be welded is completed. Specific Implementation Example 2: This disclosure also provides an embodiment: like Figure 2 A welding robot includes: a column assembly 1, a movable assembly 2, a welding execution assembly 3, a support assembly 4, an image acquisition assembly 5, and a control assembly 6. The movable assembly 2 is vertically movable and mounted on the image acquisition assembly 2. The welding execution assembly 3 is mounted on the movable assembly 2 and is used to move the welding execution assembly 3 vertically along the column assembly 1 via the movable assembly 2. The support assembly 4 is located below the column assembly 1 and is used to support the workpiece to be welded, placing the workpiece below the welding execution assembly 3. The image acquisition assembly 5 is mounted on the column assembly 1 and located above the workpiece to be welded, and is used to acquire an image of the workpiece. The control assembly 6 is electrically connected to the movable assembly 2, the image acquisition assembly 5, and the welding execution assembly 3, and is used to control the movable assembly 2, the image acquisition assembly 6, and the welding execution assembly 3 to perform welding of the stationary bell jar cavity using the stationary bell jar welding method described in Specific Embodiment 1. The control assembly 4 can be selected as needed, such as... PLC Control devices, etc. Figure 2 In s This indicates the location of the weld.

[0042] Preferably, the movable component 2 includes: a movable part 201, a driving unit 202, a supporting unit 203, and a wire feeding mechanism 204; wherein, the movable part 201 is flat, and its edge can be provided with a guide groove to form a guide rail slider structure with the edge of the column assembly 1, so that the movable part 201 can be vertically moved on the column assembly 1; the driving unit 202 is bolted to the movable part 201, and its output end is connected to a gear; the column assembly 1 is provided with a rack along the height direction, so that the gear and the rack mesh to achieve the effect of driving the movable part 201 to move vertically on the column assembly 1. The supporting unit 203 is horizontally arranged on the movable part 201, forming a cross structure with the column assembly 1; one end of the supporting unit 203 is connected to the welding execution component 3 to adjust the height between the welding execution component 3 and the weld seam on the workpiece to be welded; the wire feeding mechanism 204 is arranged on the supporting unit 203 and is used to provide welding wire to the welding execution component 3.

[0043] Preferably, the wire feeding mechanism 204 includes: a rotating shaft mechanism 2041, a wire spool 2042, and a wire feeding drive unit 2043; the rotating shaft mechanism 2041 is disposed at the other end of the support unit 203; the wire spool 2042 is disposed on the rotating shaft mechanism 2041 and can rotate along the rotating shaft mechanism 2041; the wire feeding drive unit 2043 is disposed at one end of the support unit 203 near the welding execution component 3 and is used to provide welding wire to the welding execution component 3.

[0044] like Figure 3 The welding execution component 3 includes: a slider part 301, a transverse drive mechanism 302, and a welding torch part 303; wherein, the slider part 301 is movably disposed at the end of the support unit 203; the transverse drive mechanism 302 is disposed on the support unit 203 and is used to drive the slider part 301; the welding torch part 303 is disposed on the slider part 301 and is used to move laterally along the support unit 203 under the drive of the slider part 301, so as to adjust the transverse relative position of the welding torch part 303 and the weld seam on the workpiece to be welded.

[0045] The slider part 301 can have the following structure: like Figure 4 A groove 203 is provided at the end of the support unit 203. A The groove 203 A An internal transverse drive mechanism 302, such as a cylinder, is provided; a groove 203 is also provided. ATwo guide rails 3011 are provided inside; connecting strips 3012 are provided on the guide rails 3011 through dovetail grooves; the ends of the two connecting strips 3012 are connected to the fixing plate 3013 by bolts; the end face of the fixing plate 3013 facing the transverse drive mechanism 302 is connected to the output end of the transverse drive mechanism 302, and the fixing plate 3013 can be driven to move laterally relative to the support unit 203 by the transverse drive mechanism 302; the welding gun part 303 is fixed to the fixing plate 3013 by bolts.

[0046] like Figure 5 and 6 The supporting component 4 includes: a base 401, two rotating shaft mechanisms 402, two support members 403, and a rotary drive unit 404; wherein, the base 401 is disposed on the ground; the rotating shaft mechanisms 402 are arranged in parallel on the base 401; the support members 403 are disposed on the rotating shaft mechanisms 402 and are used to support the workpiece to be welded, such as... Figure 6 The rotary drive unit 404 is connected to any of the aforementioned rotating shaft mechanisms 402 via a sprocket drive, and is used to drive the rotating shaft mechanism 402 to rotate as the active shaft, while the other rotates as an auxiliary shaft, so as to rotate the workpiece to be welded; the rotary drive unit 404 is electrically connected to the control component 4, and is used to control the welding speed of the workpiece to be welded by controlling the rotation speed of the rotating shaft mechanism 402 according to the size of the weld; the rotating shaft mechanism can be a combination structure of shaft and bearing, which is a common technology. Figure 2 and 6 In G The location represents the workpiece. Specific Implementation Example 3: This disclosure also provides an embodiment: An electronic device includes: a storage medium and a processing unit; wherein the storage medium is used to store a computer program; the processing unit exchanges data with the storage medium and is used to execute the computer program through the processing unit when controlling a welding robot to perform the steps of the static bell jar welding method as described in Specific Embodiment 1.

[0048] In the aforementioned electronic devices, the storage medium is preferably a portable hard drive or a solid-state drive or... U Storage devices such as disks; processing units, preferably CPU It exchanges data with the storage medium and, during branch merging, executes the computer program through the processing unit to perform the above-described data exchange based on... pattern The steps of a self-learning method to dynamically adjust sampling time parameters.

[0049] The above CPUThe electronic device can perform various appropriate actions and processes according to the program stored in the storage medium. It also includes peripherals such as input sections including a keyboard, mouse, etc., and may also include components such as a cathode ray tube (CRT). CRT ), LCD display ( LCD Output sections such as speakers, etc.; particularly, according to embodiments disclosed herein, such as Figure 1 Any of the processes described herein can be implemented as computer software programs.

[0050] This disclosure also provides an embodiment: A computer-readable storage medium storing a computer program; when the computer program is run, it executes the steps of the static bell jar welding method as described in Specific Embodiment 1.

[0051] In this embodiment, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system, apparatus, or device. In this disclosure, the computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can transmit, propagate, or transfer a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wireless, wireline, optical fiber, etc. RF And so on, or any suitable combination of the above.

[0052] The above are merely specific embodiments of this disclosure, but the scope of protection of this disclosure is not limited thereto. Any changes or substitutions made by those skilled in the art within the scope of the technology disclosed in this disclosure are all within the scope of protection of this disclosure. Therefore, the scope of protection of this disclosure is determined by the scope of the claims.

Claims

1. A stationary bell jar welding method characterized by, The method comprises the following steps: image acquisition is performed on the workpiece to be welded to obtain single-frame weld images, and each single-frame weld image corresponds to a rotation angle of the workpiece to be welded; weld centerline coordinates and weld widths corresponding to each centerline point are extracted from the single-frame weld images; based on the rotation angles of the workpieces corresponding to each single frame image, the corrected single frame unwinding image is aligned to the whole-circle coordinate system, multi-frame images are spliced by using a weighted fusion method, and a whole-circle weld panoramic unwinding image is obtained; a preset calibration threshold is set, different width weld regions are classified and positioned according to the threshold, a corresponding region mask is generated, the mask is fused with the whole-circle weld panoramic unwinding image to realize color-coded visual marking, and a weld width marking image is output; the weld width marking image is segmented according to a preset weld width threshold, and a segmented image is obtained; the image coordinate system in the segmented image is compared with the workpiece coordinate system of the workpiece to be welded, and the welding speed is controlled according to the weld width corresponding to the segmented image.

2. The stationary bell welding method of claim 1, wherein, The welding speed control according to the weld width corresponding to the segmented image comprises: the welding speed is obtained by the following formula: ； wherein, v is the welding speed; b is the average bead width in any segment plot; K is the process coefficient; I is the welding current; η is the deposition efficiency; d w is the wire diameter; h is the effective thickness of the weld; ρ is the wire metal density.

3. The stationary bell welding method of claim 1, wherein, The weld centerline coordinates and the weld widths corresponding to each centerline point are extracted from the single-frame weld images, which comprises the following steps: geometric distortion correction is performed on the collected single-frame images to eliminate perspective and cylindrical curvature distortion; a weld binary image is obtained, and weld centerline coordinates and weld widths corresponding to each centerline point are quantitatively extracted from the binary image; The weld centerline coordinates and the weld widths corresponding to each centerline point are quantitatively extracted from the binary image, which comprises the following steps: The single-pixel center line is obtained, and the core condition is 2≤ N ( p )≤6 and S ( p )=1, and the center line coordinate set C ={( x ' j , y ' j )| j =1,2,..., N} is output; N (p) is the number of foreground pixels in the 8-neighborhood of the pixel p; S ( p )=1 represents the connected number of the 8-neighborhood foreground pixels; Search for edge points along the vertical direction of each point on the centerline. i The width is calculated as follows: , get the pixel width; where, ( x 1i , y 1i ), ( x 2i , y 2i ) is the first i The coordinates of the left and right edge points corresponding to the center line point; W i For the first i The physical width of the weld corresponding to each centerline point.

4. The stationary bell welding method of claim 1, wherein, The corrected single frame unwinding image is aligned to the whole-circle coordinate system based on the rotation angles of the workpieces corresponding to each single frame image, multi-frame images are spliced by using a weighted fusion method, and a whole-circle weld panoramic unwinding image is obtained, which comprises the following steps: The single-frame unfolded image alignment formula is: x ' align = R ·( α + α i ),in x ' align To align the horizontal axis of the unfolded graph, α For the inner circumference angle of a single frame, α i For the first i The frame corresponds to the workpiece rotation angle; image registration involves extracting centerline matching points and calculating the affine transformation matrix. M Achieve alignment; The weighted fusion formula is: ； wherein w 1、 w 2 is a weight for the overlap region, w 1= d 2 / ( d 1+ d 2), d 1、 d 2 is a distance of a pixel to two frames; I 1( x ', y ') is the original data value of the first input source to be fused at coordinates ( x ', y ') ; I 2( x ', y ') is the original data value of the first input source to be fused at coordinates ( x ', y ').

5. The static bell jar welding method according to claim 1, wherein: The calibration threshold is determined in an adaptive manner, and is represented as: ； wherein, μ is the average width of the full circumference weld, σ is the standard deviation of the width, T low is the under-width threshold, T high is the over-width threshold; N is the total number of points of the centerline of the full circumference weld; W i is the weld width corresponding to the i th centerline point.

6. A welding robot, characterized in that, It comprises: a column assembly; a movable assembly vertically movable and arranged on the image acquisition assembly; a welding execution assembly arranged on the movable assembly and used to move the welding execution assembly along the vertical direction of the column assembly through the movable assembly; a bearing assembly arranged below the column assembly and used to bear the workpiece to be welded, so that the workpiece to be welded is located below the welding execution assembly; an image acquisition assembly arranged on the column assembly and located above the workpiece to be welded, and used to acquire images of the workpiece to be welded; a control assembly electrically connected with the movable assembly, the image acquisition assembly and the welding execution assembly, and used to control the movable assembly, the image acquisition assembly and the welding execution assembly to perform welding in the static bell jar through the static bell jar welding method according to any one of claims 1-5.

7. The welding robot according to claim 6, characterized in that The movable assembly comprises: a movable member movably arranged on the column assembly; a driving unit arranged on the movable member and used to drive the movable member to move vertically on the column assembly; A support unit is arranged on the movable element; one end of the support unit is connected to the welding execution assembly to adjust the height between the welding execution assembly and the weld on the workpiece to be welded; A wire feeding mechanism is arranged on the support unit to provide the welding execution assembly with welding wire.

8. The welding robot according to claim 7, characterized in that The wire feeding mechanism comprises: A rotating shaft mechanism is arranged at the other end of the support unit; A welding wire reel is arranged on the rotating shaft mechanism and can rotate along the rotating shaft mechanism; A wire feeding driving unit is arranged at the end of the support unit close to the welding execution assembly to provide the welding execution assembly with welding wire.

9. The welding robot of claim 7, wherein, The welding execution assembly comprises: A slider part is arranged at the end of the support unit and can move laterally; A lateral driving mechanism is arranged on the support unit to drive the slider part; A welding gun part is arranged on the slider part and can move laterally along the support unit under the drive of the slider part to adjust the lateral relative position between the welding gun part and the weld on the workpiece to be welded.

10. The welding robot of claim 6, wherein, The bearing assembly comprises: A base is arranged on the ground; At least two rotating shaft mechanisms are arranged on the base; At least two support elements are arranged on the rotating shaft mechanisms to support the workpiece to be welded; A rotating driving unit is in driving connection with any one of the rotating shaft mechanisms to drive at least one rotating shaft mechanism to rotate so as to rotate the workpiece to be welded; The rotating driving unit is in electrical connection with the control assembly to control the welding speed of the workpiece to be welded by controlling the rotating speed of the rotating shaft mechanism according to the size of the weld.

Images