Vertical welding method based on structured light vision

Abstract

The invention provides a vertical welding method based on structured light vision, comprising the following steps: assembling workpieces to be welded into V-shaped grooves with roughly symmetrical two sides; continuously taking images of the V-shaped grooves by a CDD industrial camera to obtain a picture data stream; processing each frame of the image to obtain the size and geometric morphology of the V-shaped groove; marking the number of weld bead layers and the metal volume filled in each weld bead layer; obtaining the coordinates of characteristic points arranged in sequence on each frame of the image; a welding robot welding a first layer of weld bead according to a set welding trajectory; performing a second scan on the welded first layer of weld bead by the CDD industrial camera to obtain a new picture data stream and obtain the coordinates of the characteristic points; the welding robot welding a second layer of weld bead according to the set welding trajectory; repeating scanning and welding until welding is completed; completing effective planning of the triangular swing trajectory during the welding process, thereby improving the welding quality of the robot's vertical welding.

Description

Technical Field

[0001] The invention relates to an automated welding method, in particular to a vertical welding method based on structured light vision. Background Art

[0002] Large structural parts composed of thick plates are widely used in manufacturing industries such as pressure vessels, shipbuilding, and heavy machinery. Due to their large size, they are difficult to flip during the welding process or the flipping cost is too high. In the actual welding process, it is inevitable to use multi-layer and multi-pass vertical welding.

[0003] At present, multi-layer and multi-pass vertical welding can be divided into two types: manual welding and robot welding. The main problem in the vertical welding process is the molten pool dripping problem. Manual welding is easy to operate, and the triangular swing welding method is often used to overcome the molten pool dripping. The principle is that during the swinging process, the lower molten pool solidifies, thereby playing a certain role in supporting the upper molten pool.

[0004] When using a welding robot for vertical welding, the triangular swing trajectory is demanding, difficult to calculate and plan, and the welding parameters are difficult to plan, which makes it impossible to achieve automatic welding of the above trajectory. At present, the robot vertical welding mainly improves the automation efficiency through the teaching method. The straight line method is often used for welding, and the weld metal forming type is poor, which makes it difficult to ensure the welding quality. The manual teaching efficiency is low, and it is difficult to achieve full automation and intelligence.

[0005] In order to solve the above problems, people have been seeking an ideal technical solution. Summary of the invention

[0006] The purpose of the present invention is to address the deficiencies in the prior art and thereby provide a vertical welding method based on structured light vision that has scientific design, high degree of automation, good welding quality and high welding efficiency.

[0007] In order to achieve the above object, the technical solution adopted by the present invention is: a vertical welding method based on structured light vision, comprising the following steps:

[0008] Step S1, assembling the workpieces to be welded into V-shaped grooves with roughly symmetrical two sides, wherein the V-shaped grooves extend in the vertical direction;

[0009] Step S2, according to a predetermined teaching trajectory and a predetermined shooting frame rate, vertically projecting a straight laser line of a CDD industrial camera onto the V-shaped groove, continuously shooting images of the V-shaped groove, and acquiring a picture data stream;

[0010] Step S3, processing each frame of the image to obtain the size and geometric shape of the V-shaped groove;

[0011] Step S4: planning the number of weld bead layers and the metal volume filled in each weld bead layer:

[0012] The metal volume V that needs to be filled in the V-shaped groove per unit length is calculated according to the following formula:

[0013]

[0014] Plan the number of weld layers n according to the following rules:

[0015] 0.5H<n≤2H

[0016] The metal volume V1 required to be filled in the first layer of weld per unit length is calculated according to the following formula:

[0017] V1=V / 2n

[0018] Calculate the metal volume V required to fill the Kth weld bead (K ≥ 2) per unit length using the following formula: k :

[0019] V K =V k-1 +V / n(n-1)

[0020] Wherein, H is the average value of the V-shaped groove depth in each frame image, in cm, 2α is the average value of the V-shaped groove angle in each frame image, and b is the average value of the cover layer distance from the edge in each frame image;

[0021] Step S5: Assume that the transverse direction of the V-shaped groove is the u direction and the depth direction is the v direction, and obtain the coordinates of the characteristic points a, c, m1, d, and e arranged in sequence on each frame of the image:

[0022] Search from both sides to the middle in the u direction. When the gray value of the 200th pixel is 255, record the coordinates of points a and e. Search in the v direction. When the difference with points a and e is greater than 5, record the coordinates of points c1 and d1. When the difference is greater than 10, record the coordinates of points c2 and d2. Points c1 and c2 are fitted with a straight line Lc1c2. Points d1 and d2 are fitted with a straight line Ld1d2. The straight line Lc1c2 is combined with ua to obtain the coordinates of point c. The straight line Ld1d2 is combined with ue to obtain the coordinates of point d. The straight line Lc1c2 is combined with Ld1d2 to obtain the coordinates of point m1.

[0023] Step S6: The welding robot welds the first layer of welds according to the set welding trajectory;

[0024] Step S7, use the CDD industrial camera to perform a second scan on the first layer of weld bead after welding to obtain a new image data stream, and obtain the coordinates of the characteristic points a, c, f, m2, g, d, and e arranged in sequence on each frame of the image, wherein the method for obtaining a, c, d, and e is the same as step S5, and the points with the largest difference with a and e in the v direction are points f and g, and the extreme point of the image fg calculated by the second-order difference method is point m2;

[0025] Step S8, the welding robot welds the second layer of welds according to the set welding trajectory;

[0026] Step S9: Repeat steps S7 and S8 to obtain the characteristic points a, c, f, m arranged in sequence. k , g, d, e coordinates, where k is the number of weld layers, and the welding robot welds each layer of weld according to the set welding trajectory;

[0027] In steps S6, S8 and S9, the welding track is cyclically advanced in a triangular swinging manner. In one cycle, the welding gun moves from the lower end of the groove along the Trajectory motion, where B represents the endpoint of the triangular swing trajectory, the superscript represents the number of cycles, the subscript k represents the number of weld layers, l represents the left endpoint of the triangular swing trajectory, r represents the right endpoint of the triangular swing trajectory, and w represents the endpoint of the triangular swing trajectory in the depth direction. kw From m k The points selected in

[0028] Assume that the total length of the V-shaped groove is L, there are T cycles in total, B kw The coordinates of (x kw ,y kw ,z kw ),but:

[0029] The distance that needs to be moved forward along the weld in each cycle is L / T+1;

[0030] Point B kl (x kl ,y kl ,z kl ), B kr (x kr ,y kr ,z kr ) is calculated as:

[0031]

[0032] Where: h k-1 For B kw (x kw ,y kw,z kw ) point y kw Coordinate value.

[0033] Based on the above, when welding each weld in steps S6, S8 and S9, the wire feeding speed v 送 The calculation formula is:

[0034]

[0035] Where: I is the welding current per pass; α H is the cladding efficiency; d is the wire diameter; ρ is the specific gravity of the cladding metal;

[0036] Welding speed v per weld 焊 The calculation formula is:

[0037]

[0038] Where: S is the distance the welding gun travels for each weld,

[0039] Based on the above, the processing of each frame of image includes the steps of mean filtering, morphological processing, and grayscale centroid method to extract the center line of structured light.

[0040] Based on the above, when obtaining the coordinates of the feature points in step S5 and step S7, the pixel coordinates of the feature points are converted into coordinates in the motion coordinate system of the welding six-axis robot through the camera calibration parameters, hand-eye calibration parameters, and structured light plane parameters predetermined in the system.

[0041] The present invention has outstanding substantive features and significant progress compared to the prior art. Specifically, the present invention has the following advantages:

[0042] The effective planning of the triangular swing trajectory is realized. When the welding gun swings to both sides of the groove, it moves toward the open end of the groove. When it swings to the middle of the groove, it moves toward the closed end of the groove, and the shape of the groove is well fitted. By calculating the trajectory coordinates, the travel route of the welding gun is accurately delineated. By matching the calculated welding parameters, it can be ensured that each position of the welding gun is swung into place and the molten pool is filled in place, so as to realize the automatic welding of the welding robot. The V-shaped groove is scanned frame by frame using a CCD industrial camera, and each frame image is processed separately. By obtaining the mean of each size of each frame image, the accuracy of calculating the number of weld layers and the filling metal of the entire weld and each weld can be improved. The triangular swing trajectory calculated based on the processing results of each frame image is used as a reference. When the welding gun performs vertical welding according to the trajectory, it can better match the different positions of the weld, improve the fault tolerance of the shape defect position, and improve the welding quality of the entire weld. When extracting feature points, a certain distance is left from the endpoints and corners to avoid defects caused by endpoint light or corner shapes, so that the extraction of feature points can more accurately reflect the true shape of the groove and ensure the accuracy of the calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a structural schematic diagram of the welding robot scanning the vertical weld in the present invention.

[0044] Figure 2 It is a schematic diagram for calculating the weld metal volume in the present invention.

[0045] Figure 3 It is a schematic diagram of laser stripes of each weld in the present invention.

[0046] Figure 4 It is a schematic diagram of extracting feature points from each frame of image in the present invention.

[0047] Figure 5 It is a schematic diagram of the planning of the triangular swing trajectory in the present invention.

[0048] In the figure: 1. Six-axis welding robot; 2. Line laser generator; 3. CCD industrial camera; 4. Visual sensor; 5. Camera lens; 6. Robot control cabinet; 7. Industrial computer; 8. Workpiece to be welded; 9. Linear laser line; 10. First layer weld bead; 11. Second layer weld bead. DETAILED DESCRIPTION

[0049] The technical solution of the present invention is further described in detail below through specific implementation methods.

[0050] like Figure 1As shown, the welding robot used in the present invention includes a six-axis welding robot 1, a visual sensor 4, a robot control cabinet 6, and an industrial computer 7; the visual sensor 4 is installed on the end flange of the six-axis welding robot 1; the visual sensor 4 is composed of a CCD industrial camera 3, a camera lens 5 and a line laser sensor 3, the line laser generator 2 is connected to a 24V DC power supply, the CCD industrial camera 3 is connected to the industrial computer 7 through a Gige network port, the six-axis welding robot 1, the visual sensor 4, and the robot control cabinet 6 constitute a motion control system, and the robot control cabinet 6 is connected to the industrial computer 7 through a Gige network port; the visual sensor 4 transmits the weld image with laser stripes to the industrial computer 7 in real time. The welding robot is an existing device, and the present welding method is implemented based on the device.

[0051] like Figure 1-5 As shown, a vertical welding method based on structured light vision includes the following steps:

[0052] Step S1, assembling the workpieces 8 to be welded into V-shaped grooves with roughly symmetrical two sides, wherein the V-shaped grooves extend in the vertical direction.

[0053] Step S2: according to a predetermined teaching trajectory and a predetermined shooting frame rate, the straight laser line 9 of the CDD industrial camera 3 is vertically projected onto the V-shaped groove, and images of the V-shaped groove are continuously shot to obtain a picture data stream.

[0054] Step S3, processing each frame of the image, including mean filtering, morphological processing, and grayscale centroid method to extract the center line of the structured light, and obtain the size and geometric shape of the V-shaped groove.

[0055] Step S4: planning the number of weld bead layers and the metal volume filled in each weld bead layer:

[0056] The metal volume V that needs to be filled in the V-shaped groove per unit length is calculated according to the following formula:

[0057]

[0058] Plan the number of weld layers n according to the following rules:

[0059] 0.5H<n≤2H

[0060] The metal volume V1 required to be filled in the first layer of weld per unit length is calculated according to the following formula:

[0061] V1=V / 2n

[0062] Calculate the metal volume V required to fill the Kth weld bead (K ≥ 2) per unit length using the following formula: k :

[0063] V K =V k-1+V / n(n-1)

[0064] Among them, H is the average value of the V-groove depth in each frame image, in cm, 2α is the average value of the V-groove angle in each frame image, and b is the average value of the distance from the cover layer to the edge in each frame image. The calculation of the average value here is conducive to improving the accuracy of calculating the number of weld layers and the filler metal of the entire weld and each weld.

[0065] Step S5: Assume that the transverse direction of the V-shaped groove is u direction and the depth direction is v direction, and obtain the coordinates of the characteristic points a, c, m1, d, and e arranged in sequence on each frame image:

[0066] Search from both sides to the middle in the u direction. When the gray value of the 200th pixel is 255, record the coordinates of points a and e. This can avoid inaccurate point finding caused by end light or uneven plate. Search in the v direction. When the difference with points a and e is greater than 5, record the coordinates of points c1 and d1 (for example, points with a difference of 6, 7, 8, and 9). When the difference is greater than 10, record the coordinates of points c2 and d2 (for example, points with a difference of 11, 12, 13, and 14). Points c1 and c2 are fitted with a straight line Lc1c2, and points d1 and d2 are fitted with a straight line Ld1d2. The straight line Lc1c2 is combined with ua to obtain the coordinates of point c, the straight line Ld1d2 is combined with ue to obtain the coordinates of point d, and the straight line Lc1c2 is combined with Ld1d2 to obtain the coordinates of point m1. This can avoid inaccurate straight line fitting caused by irregular corner shapes, so that the extraction of feature points can more accurately reflect the true shape of the groove.

[0067] In this step and the subsequent coordinate calculation step, the pixel coordinates of the feature points are converted into coordinates in the motion coordinate system of the six-axis welding robot 1 through the camera calibration parameters, hand-eye calibration parameters, and structured light plane parameters predetermined in the system.

[0068] Step S6: The welding robot welds the first layer weld bead 10 according to the set welding trajectory.

[0069] Step S7, use the CDD industrial camera 3 to perform a second scan on the first layer weld 10 after welding to obtain a new image data stream, and obtain the coordinates of the sequentially arranged feature points a, c, f, m2, g, d, and e in each frame of the image, where a, c, d, and e are obtained in the same way as step S5. The points with the largest difference with a and e in the v direction are points f and g. The extreme point of the image fg calculated by the second-order difference method is point m2.

[0070] Step S8: The welding robot welds the second layer weld bead 11 according to the set welding trajectory.

[0071] Step S9: Repeat steps S7 and S8 to obtain the characteristic points a, c, f, m arranged in sequence.k , g, d, and e are coordinates, where k is the number of weld layers. In this embodiment, there are 3 weld layers in total, and the welding robot welds each layer of weld according to the set welding trajectory.

[0072] In steps S6, S8 and S9, the welding track is cyclically advanced in a triangular swinging manner. In one cycle, the welding gun moves from the lower end of the groove along the Trajectory motion, where B represents the endpoint of the triangular swing trajectory, the superscript represents the number of cycles, the subscript k represents the number of weld layers, l represents the left endpoint of the triangular swing trajectory, r represents the right endpoint of the triangular swing trajectory, and w represents the endpoint of the triangular swing trajectory in the depth direction. kw From m k The points selected in

[0073] Assume that the total length of the V-shaped groove is L, there are T cycles in total, B kw The coordinates of (x kw ,y kw ,z kw ),but:

[0074] The distance that needs to be moved forward along the weld in each cycle is L / T+1;

[0075] Point B kl (x kl ,y kl ,z kl ), B kr (x kr ,y kr ,z kr ) is calculated as:

[0076]

[0077] Where: h k-1 For B kw (x kw ,y kw ,z kw ) point y kw Coordinate value.

[0078] Since the triangular swing trajectory is calculated based on the processing results of each frame of the image, when the welding gun performs vertical welding according to the trajectory, it can better match the different positions of the weld, improve the fault tolerance of the shape defect position, and improve the welding quality of the entire weld.

[0079] When welding each weld in steps S6, S8 and S9, the wire feeding speed v 送 The calculation formula is:

[0080]

[0081] Where: I is the welding current per pass; α H is the cladding efficiency; d is the wire diameter; ρ is the specific gravity of the cladding metal;

[0082] Welding speed v per weld 焊 The calculation formula is:

[0083]

[0084] Where: S is the distance the welding gun travels for each weld,

[0085] This welding method realizes the effective planning of the triangular swing trajectory. When the welding gun swings to both sides of the groove, it moves toward the open end of the groove. When it swings to the middle of the groove, it moves toward the closed end of the groove, and fits well with the shape of the groove. By calculating the trajectory coordinates, the travel route of the welding gun is accurately delineated, and the calculated welding parameters are matched to ensure that each position of the welding gun is swung into place and the molten pool is filled in place. The recognition process adopts image processing technology without manual teaching, thereby improving welding efficiency and quality.

[0086] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or some technical features can be replaced by equivalents without departing from the spirit of the technical solution of the present invention, which should be included in the scope of the technical solution for protection of the present invention.

Claims

1. A vertical welding method based on structured light vision, characterized in that , including the following steps: Step S1, assembling the workpieces to be welded into V-shaped grooves with roughly symmetrical two sides, wherein the V-shaped grooves extend in the vertical direction; Step S2, according to a predetermined teaching trajectory and a predetermined shooting frame rate, vertically projecting a straight-line laser line of a CCD industrial camera onto the V-shaped groove, continuously shooting images of the V-shaped groove, and acquiring a picture data stream; Step S3, processing each frame of the image to obtain the size and geometric shape of the V-shaped groove; Step S4: planning the number of weld bead layers and the metal volume filled in each weld bead layer: The metal volume V that needs to be filled in the V-shaped groove per unit length is calculated according to the following formula: Plan the number of weld layers n according to the following rules: 0.5H<n≤2H The metal volume V1 required to be filled in the first layer of weld per unit length is calculated according to the following formula: V1=V / 2n Calculate the metal volume V required to fill the Kth weld bead (K ≥ 2) per unit length using the following formula: k : V K =V k-1 +V / n(n-1) Wherein, H is the average value of the V-shaped groove depth in each frame image, in cm, 2α is the average value of the V-shaped groove angle in each frame image, and b is the average value of the cover layer distance from the edge in each frame image; Step S5: Assume that the transverse direction of the V-shaped groove is the u direction and the depth direction is the v direction, and obtain the coordinates of the characteristic points a, c, m1, d, and e arranged in sequence on each frame of the image: Search from both sides to the middle in the u direction. When the gray value of the 200th pixel is 255, record the coordinates of points a and e. Search in the v direction. When the difference with points a and e is greater than 5, record the coordinates of points c1 and d1. When the difference is greater than 10, record the coordinates of points c2 and d2. Points c1 and c2 are fitted with a straight line Lc1c2. Points d1 and d2 are fitted with a straight line Ld1d2. The straight line Lc1c2 is combined with ua to obtain the coordinates of point c. The straight line Ld1d2 is combined with ue to obtain the coordinates of point d. The straight line Lc1c2 is combined with Ld1d2 to obtain the coordinates of point m1. Step S6: The welding robot welds the first layer of welds according to the set welding trajectory; Step S7, use the CCD industrial camera to perform a second scan on the first layer of weld bead after welding to obtain a new image data stream, and obtain the coordinates of the characteristic points a, c, f, m2, g, d, and e arranged in sequence on each frame of the image, wherein the method for obtaining a, c, d, and e is the same as step S5, and the points with the largest difference with a and e in the v direction are points f and g, and the extreme point of the image fg calculated by the second-order difference method is point m2; Step S8, the welding robot welds the second layer of welds according to the set welding trajectory; Step S9: Repeat steps S7 and S8 to obtain the characteristic points a, c, f, m arranged in sequence. k , g, d, e coordinates, where k is the number of weld layers, and the welding robot welds each layer of weld according to the set welding trajectory; In steps S6, S8 and S9, the welding track is cyclically advanced in a triangular swinging manner. In one cycle, the welding gun moves from the lower end of the groove along B t kw →B t kr →B t kl →B t+1 kw Trajectory motion, where B represents the endpoint of the triangular swing trajectory, the superscript represents the number of cycles, the subscript k represents the number of weld layers, l represents the left endpoint of the triangular swing trajectory, r represents the right endpoint of the triangular swing trajectory, and w represents the endpoint of the triangular swing trajectory in the depth direction. kw From m k The points selected in Assume that the total length of the V-shaped groove is L, there are T cycles in total, B kw The coordinates of (x kw ,y kw ,z kw ),but: The distance that needs to be moved forward along the weld in each cycle is L / T+1; Point B kl (x kl ,y kl ,z kl ), B kr (x kr ,y kr ,z kr ) is calculated as: Where: h k-1 For B kw (x kw ,y kw ,z kw ) point y kw Coordinate value.

2. The vertical welding method based on structured light vision according to claim 1 is characterized in that: When welding each weld in steps S6, S8 and S9, the wire feeding speed v 送 The calculation formula is: Where: I is the welding current per pass; α H is the cladding efficiency; d is the wire diameter; ρ is the specific gravity of the cladding metal; Welding speed v per weld 焊 The calculation formula is: Where: S is the distance the welding gun travels for each weld, 3. The vertical welding method based on structured light vision according to claim 1 or 2, characterized in that: The processing of each frame of image includes the steps of mean filtering, morphological processing, and grayscale centroid method to extract the center line of structured light.

4. The vertical welding method based on structured light vision according to claim 3 is characterized in that: When obtaining the coordinates of the feature points in step S5 and step S7, the pixel coordinates of the feature points are converted into coordinates in the motion coordinate system of the six-axis welding robot through the camera calibration parameters, hand-eye calibration parameters, and structured light plane parameters predetermined in the system.

Images