Geometric feature-based real-time weld bead feature point identification method

Abstract

The invention discloses a real-time weld bead feature point identification method based on geometric features, and belongs to the technical field of welding. The problem that an existing weld joint feature point recognition method is not accurate enough in feature point recognition under the condition that a weld bead exists is solved. A welding gun and a laser are installed and subjected to hand-eye calibration, and a welding seam area laser line coordinate set corresponding to the laser is obtained; extracting a welding seam starting point and a welding seam ending point from the welding seam area laser line coordinate set, and identifying a welding seam anchor point by using the welding seam starting point and the welding seam ending point; calculating upper-end left-right feature points and lower-end left-right feature points of the weld joints, and correcting the weld joints with weld beads and the lower-end left-right feature points with the distance larger than 3 mm; and intercepting a welding seam between the left and right feature points at the upper end, performing exponential weighted average movement smoothing on the welding seam, and identifying the welding bead feature points. The method is suitable for identifying the weld bead feature points.

Description

Technical Field

[0001] The present invention belongs to the technical field of welding. Background Art

[0002] With the rapid development of industrial automation, welding robots have gradually replaced manual welding and become an important part of modern manufacturing. Due to uncertain factors such as the processing accuracy of the previous welding process and the fixture accuracy, the welding quality is not stable.

[0003] To solve the above problems, laser weld seam tracking technology has emerged. This technology uses a laser emitter and a receiving sensor to identify the positions of weld seam feature points and feedback them to the welding robot, enabling it to automatically adjust the welding trajectory and achieve high-precision and high-quality welding.

[0004] Currently, common methods for identifying weld seam feature points include feature point identification based on deep learning and feature point identification based on derivative analysis. The method based on deep learning relies on a large amount of weld seam data accumulation and has problems such as insufficient recognition stability and real-time performance; the feature point identification based on derivative analysis requires the workpiece surface to be smooth and is easily affected by laser point perturbation during the movement of the robotic arm. In addition, the existing methods for identifying weld seam feature points are not accurate enough in identifying feature points in the presence of weld beads. Summary of the Invention

[0005] The present invention is to solve the problem that the existing methods for identifying weld seam feature points are not accurate enough in identifying feature points in the presence of weld beads, and provides a real-time weld seam weld bead feature point identification method based on geometric features.

[0006] The real-time weld seam weld bead feature point identification method based on geometric features of the present invention includes:

[0007] Step 1: Install and calibrate the welding torch and the laser, and obtain the coordinate set of the weld seam area corresponding to the laser;

[0008] Step 2: Extract the starting point and the ending point of the weld seam from the laser line coordinate set of the weld seam area, and identify the weld seam anchor points by using the starting point and the ending point of the weld seam;

[0009] Step 3: Use the weld seam anchor points to calculate the left and right feature points at the upper end of the weld seam, and then use the left and right feature points at the upper end of the weld seam to calculate the left and right feature points at the lower end of the weld seam. For the weld seam with a weld bead, when the distance between the left and right feature points at the lower end is greater than 3 mm, correct the left and right feature points at the lower end; obtain the corrected left and right feature points at the lower end;

[0010] Step 4: Cut the weld seam between the left and right feature points at the upper end, perform index weighted average moving smoothing on the weld seam, and identify the weld bead feature points. When the number of weld bead feature points is less than 3, it is determined that there is no weld bead. If the number of weld bead feature points meets 3, then according to the coordinates of the 3 feature points, they are determined as the weld bead starting feature point, the midpoint feature point, and the weld bead ending feature point.

[0011] Further, in the present invention, in Step 2, the method for identifying the weld seam anchor points by using the weld seam starting point and the weld seam ending point is as follows:

[0012] Select the laser line starting point and the ending point from the laser line coordinate set of the weld seam area, mark the points in this section of the weld seam, number them from 0 starting from the weld seam starting point until the weld seam ending point is numbered n, and calculate the anchor points of this section of the weld seam. The specific formula is:

[0013]

[0014] Among them, P 0 represents the weld seam starting point coordinate, P n represents the weld seam ending point coordinate, P i represents the coordinate of the i-th point in the laser line coordinate set of the weld seam area, represents the coordinate vector from the weld seam starting point to the i-th point. Among them, represents the coordinate vector from the weld seam starting point to the weld seam ending point P 0 P n of the vector, i ⊥,max represents the anchor point serial number corresponding to the maximum vertical distance for this section of the weld seam, argmax 0<i<n represents taking the serial number of the point corresponding to the maximum distance between all points 0 < i < n of this section of the weld seam and the line segment between the weld seam starting point and the weld seam ending point.

[0015] Further, in the present invention, in Step 3, the method for calculating the left and right feature points at the upper end of the weld seam by using the weld seam anchor points is as follows:

[0016] In Step 3, the method for calculating the left and right feature points at the upper end of the weld seam by using the weld seam anchor points is as follows:

[0017] Calculate the vertical distance from each point P i =(x i1 , y i1 ) i1 ∈ (0, a) between the weld seam starting point and the anchor point of this section of the weld seam to the line segment P 0 P a . The point with the maximum vertical distance is identified as the left feature point at the upper end of the weld seam. The line segment P 0 P a represents the connecting line segment from the weld seam starting point P 0 to the anchor point P a of this section of the weld seam;

[0018] Calculate each point P between the anchor point and the end point of this section of the weld seam i2 =(x i2 ,y i2 ), where i∈(a,n), to the vertical distance of the line segment P a P n . The point with the maximum vertical distance is identified as the right feature point at the upper end of the weld seam. The line segment P a P n represents the line segment connecting the starting point and the end point of this section of the weld seam.

[0019] Furthermore, in the present invention, in step three, the method for calculating the left and right feature points at the lower end of the weld seam is as follows:

[0020] Set the left feature point at the upper end as P tl =(x tl ,y tl ), and the right feature point at the upper end as P tr =(x tr ,y tr );

[0021] Select all points on the weld seam between the left feature point at the upper end of the weld seam and the anchor point of this section of the weld seam, and connect them to the starting point of the weld seam. Calculate the slope of each connecting line P 0 P i3 . Then, connect the starting point of the weld seam and the left feature point at the upper end of the weld seam P 0 P tl , and calculate the slope of the connecting line P 0 P tl .

[0022] Subtract the slope of each connecting line P 0 P i from the slope of the connecting line P 0 P tl . The weld point corresponding to the maximum difference is the left feature point P bl at the lower end of the weld seam, and the serial number is:

[0023]

[0024] Among them, P i =(x i3 ,y i3 ), where i3∈(tl,a) represents the i3th point on the weld seam between the left feature point at the upper end and the anchor point of this section of the weld seam, and argmax i＝tl+1,tl+2,…,a represents taking the serial number of the point corresponding to the maximum distance between all points on the weld seam between the left feature point at the upper end and the anchor point of this section of the weld seam and the connecting line P 0 P tl between the starting point of the weld seam and the left feature point at the upper end;

[0025] Connect all the points between the weld anchor point of this section of the weld and the upper right feature point to the weld end point, and calculate each connection line P i4 P n the slope of, and then connect the upper right feature point to the weld end point to get line P tr P n , calculate the slope of line P tr P n ; find the difference between the slope of each connection line P i4 P n and the slope of line P tr P n . The point with the largest slope difference is the lower right feature point P br , and the serial number is:

[0026]

[0027] Among them, P i4 =(x i4 , y i4 ) i4 ∈ (a, tr) represents the i4th point in the corresponding weld from the weld anchor point of this section of the weld to the weld end point; argmax i＝a+1,a+2,…,tr represents taking the serial number of the point corresponding to the maximum distance from all points between the weld anchor point and the upper right feature point in this section of the weld to the line P tr P n between the upper right feature point and the weld end point.

[0028] Furthermore, in the present invention, in step three, the method for correcting the left and right feature points at the lower end of the weld is as follows:

[0029] For the weld with weld beads, if the distance between the left feature point and the right feature point at the lower end exceeds 3 mm, a correction operation is performed. Specifically:

[0030] Correct the left feature point at the lower end:

[0031] Calculate the distance from all points on the weld between the upper left feature point and the lower right feature point to the line segment P tl P br . The point with the maximum vertical distance is used as the corrected left feature point at the lower end of the weld. The line segment P tl P br is the connection line between the upper left feature point and the lower right feature point;

[0032] Correct the right feature point at the lower end:

[0033] Calculate the vertical distance from all laser points on the weld between the left feature point and the right feature point at the lower end to the line segment P bl P tr . The point with the maximum vertical distance is used as the corrected right feature point at the lower end of the weld. The line segment P bl Pbr It is the line from the lower left feature point to the lower right feature point.

[0034] Furthermore, in the present invention, in step 4, the specific method for identifying the characteristic points of the weld bead is:

[0035] Step 51: intercept the weld between the upper left and right feature points, normalize the intercepted weld, and obtain a normalized weld curve;

[0036] Step 52: Calculate the difference between the horizontal coordinate x and the vertical coordinate y of all points of the normalized weld curve to obtain a difference curve;

[0037] Step 53: extracting points in the difference curve where the growth rate of the y-value of the ordinate decreases, and taking the points where the growth rate of the y-value of the ordinate decreases as candidate points;

[0038] Step 54: Calculate the threshold of each candidate point:

[0039]

[0040] in, and Indicates the sn on the curve after normalization i+1 Dot and sn i The horizontal coordinate of the point, represents the ordinate of the candidate point, S=1, m is the number of candidate points;

[0041] Step 55: Screen the candidate points along the positive x-axis of the difference curve. When there is a difference curve value between the previous candidate point and the next candidate point that is less than the threshold of the previous candidate point, the previous candidate point is used as a feature point. Otherwise, the previous candidate point is deleted and the next candidate point is screened until all candidate points are screened and all feature points are obtained.

[0042] Further, in the present invention, in step 51, the normalized weld curve is:

[0043]

[0044] Among them, x s and s Represents the set of horizontal and vertical coordinates of all points on the curve before normalization. and Represents the s on the curve before normalization i The horizontal and vertical coordinates of the point, and Indicates the sn on the curve after normalization i The horizontal and vertical coordinates of the point.

[0045] Furthermore, in the present invention, in step 52, the difference curve:

[0046]

[0047] Among them, and represent the abscissa and ordinate of point d on the difference curve. i of the point.

[0048] Furthermore, in the present invention, in step five-three, the formula for extracting the points where the growth rate of the y value in the difference curve decreases is:

[0049]

[0050] Among them, and represent the abscissa and ordinate of the candidate point.

[0051] The existing conventional method for detecting weld feature points is to calculate the second derivative of each point on the weld curve and determine the feature points by finding the maximum and minimum values of the second derivative. However, since the surface of the welded base material is not absolutely smooth, the protrusions or depressions on the surface of the base material will affect the value of the second derivative. The present invention finds the weld feature points through the geometric features of the weld, is not affected by the smoothness of the base material, has accurate recognition and higher robustness, and the error of the weld feature points is within 0.3 mm. The traditional weld recognition method is difficult to accurately quantify the features of the weld bead in the weld. The present invention can simultaneously identify the feature points of the weld and the weld bead, and can more comprehensively describe the morphological features of the weld. The weld bead feature points of the present invention are accurately recognized, and the error is within 1 mm; and the recognition speed of the weld and weld bead feature points is fast, and the single recognition time is within 15 ms; it has strong anti-mechanical vibration ability, and when the robotic arm moves rapidly, the deviation of the feature points does not exceed 1 mm. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 is the flow chart of the method of the present invention;

[0053] Figure 2 is the schematic diagram of the weld and weld bead feature points. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0054] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention. It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments may be combined with each other.

[0055] DETAILED DESCRIPTION OF THE EMBODIMENT ONE: Refer to Figure 1 and Figure 2Specifically describing this embodiment, the method for real-time recognition of weld bead feature points based on geometric features in this embodiment includes:

[0056] Step 1: Install and perform hand-eye calibration on the welding torch and the laser, and obtain the coordinate set of the weld area corresponding to the laser;

[0057] Step 2: Extract the weld start point and the weld end point from the laser line coordinate set of the weld area, and use the weld start point and the weld end point to identify the weld anchor point;

[0058] Step 3: Use the weld anchor point to calculate the left and right feature points at the upper end of the weld, and then use the left and right feature points at the upper end of the weld to calculate the left and right feature points at the lower end of the weld. For the weld with a weld bead, when the distance between the left and right feature points at the lower end is greater than 3 mm, correct the left and right feature points at the lower end; obtain the corrected left and right feature points at the lower end;

[0059] Step 4: Intercept the weld between the left and right feature points at the upper end, perform exponential weighted average moving smoothing on the weld, and perform weld bead feature point recognition. When the number of weld bead feature points is less than 3, it is determined that there is no weld bead. If the number of weld bead feature points meets 3, then according to the coordinates of the 3 feature points, determine them as the weld bead start point feature point, the midpoint feature point, and the weld bead end point feature point.

[0060] Further, in the present invention, in Step 2, the method for using the weld start point and the weld end point to identify the weld anchor point is:

[0061] Select the laser line start point and the end point from the laser line coordinate set of the weld area, mark the points in this section of the weld, label them from 0 starting from the weld start point until the weld end point is labeled as n, and calculate the anchor point of this section of the weld. The specific formula is:

[0062]

[0063] where P 0 represents the weld start point coordinate, P n represents the weld end point coordinate, P i represents the coordinate of the i-th point in the laser line coordinate set of the weld area, represents the coordinate vector from the weld start point to the i-th point, where, represents the coordinate vector from the weld start point to the weld end point P 0 P n of, i ⊥,max represents the anchor point serial number corresponding to the maximum vertical distance for this section of the weld, argmax 0<i<n represents taking the serial number of the point corresponding to the maximum distance between all points 0 < i < n in this section of the weld and the line segment between the weld start point and the weld end point.

[0064] Further, in the present invention, in step three, the method for calculating the left and right feature points at the upper end of the weld seam by using the weld seam anchor points is as follows:

[0065] In step three, the method for calculating the left and right feature points at the upper end of the weld seam by using the weld seam anchor points is as follows:

[0066] Calculate each point P between the starting point of the weld seam and the weld seam anchor point of this section i =(x i1 , y i1 ) i1 ∈ (0, a) to the vertical distance of the line segment P 0 P a . The point with the maximum vertical distance is identified as the left feature point at the upper end of the weld seam. The line segment P 0 P a represents the connecting line segment from the starting point P 0 of the weld seam to the weld seam anchor point P a of this section;

[0067] Calculate each point P between the weld seam anchor point of this section of the weld seam and the end point of this section of the weld seam i2 =(x i2 , y i2 ) i ∈ (a, n) to the vertical distance of the line segment P a P n . The point with the maximum vertical distance is identified as the right feature point at the upper end of the weld seam. The line segment P a P n represents the line segment connecting the starting point and the end point of this section of the weld seam.

[0068] Further, in the present invention, in step three, the method for calculating the left and right feature points at the lower end of the weld seam is as follows:

[0069] Set the left feature point at the upper end as P tl =(x tl , y tl ), and the right feature point at the upper end as P tr =(x tr , y tr );

[0070] Select all the points on the weld seam between the left feature point at the upper end of the weld seam and the weld seam anchor point of this section and connect them with the starting point of the weld seam. Calculate the slope of each connecting line P 0 P i3 . Then, for the connecting line P 0 P tl connecting the starting point of the weld seam and the left feature point at the upper end of the weld seam, calculate the slope of the connecting line P 0 P tl ;

[0071] Compare the slope of each connecting line P 0 P i with the slope of the connecting line P 0 Ptl Take the difference of the slopes, and the weld point corresponding to the largest difference value is the left lower feature point P of the weld bl , and the serial number is:

[0072]

[0073] Among them, P i =(x i3 , y i3 ) i3 ∈ (tl, a) represents the i3-th point in the weld between the upper left feature point and the anchor point of this section of the weld. argmax i＝tl+1,tl+2,…,a represents taking the serial number of the point corresponding to the maximum distance between all points in the weld between the upper left feature point and the anchor point of this section of the weld to the line connecting the weld starting point and the upper left feature point of the weld P 0 P tl ;

[0074] Select all points in the weld between the anchor point of this section of the weld and the upper right feature point and connect them to the weld end point, calculate the slope of each connecting line P i4 P n , and then connect the upper right feature point and the weld end point P tr P n , calculate the slope of the connecting line P tr P n ; find the difference between the slope of each connecting line P i4 P n and the slope of the connecting line P tr P n . The point with the largest slope difference is the right lower feature point P of the weld br , and the serial number is:

[0075]

[0076] Among them, P i4 =(x i4 , y i4 ) i4 ∈ (a, tr) represents the i4-th point in the weld corresponding to the anchor point of this section of the weld to the weld end point; argmax i＝a+1,a+2,…,tr represents taking the serial number of the point corresponding to the maximum distance between all points in the weld between the anchor point of this section of the weld and the upper right feature point to the line connecting the upper right feature point and the weld end point P tr P n .

[0077] Furthermore, in the present invention, in step three, the method for correcting the left and right lower feature points of the weld is:

[0078] For the weld with weld beads, when the distance between the left lower feature point and the right lower feature point exceeds 3 mm, a correction operation is performed, specifically:

[0079] Correct the lower left feature point:

[0080] Calculate all points on the weld from the upper left feature point to the lower right feature point to the line segment P tl P br The point with the largest vertical distance is the left characteristic point at the lower end of the weld after correction. The line segment P tl P br It is the line connecting the upper left feature point and the lower right feature point;

[0081] Correct the lower right feature point:

[0082] Calculate all laser points on the weld from the lower left feature point to the lower right feature point to the line segment P bl P tr The vertical distance is the point with the largest vertical distance as the right characteristic point of the lower end of the weld after correction. The line segment P bl P tr It is the line from the lower left feature point to the lower right feature point.

[0083] Furthermore, in the present invention, in step 4, the specific method for identifying the characteristic points of the weld bead is:

[0084] Step 51: intercept the weld between the upper left and right feature points, normalize the intercepted weld, and obtain a normalized weld curve;

[0085] Step 52: Calculate the difference between the horizontal coordinate x and the vertical coordinate y of all points of the normalized weld curve to obtain a difference curve;

[0086] Step 53: extracting points in the difference curve where the growth rate of the y-value of the ordinate decreases, and taking the points where the growth rate of the y-value of the ordinate decreases as candidate points;

[0087] Step 54: Calculate the threshold of each candidate point:

[0088]

[0089] in, and Indicates the sn on the curve after normalization i+1 Dot and sn i The horizontal coordinate of the point, represents the ordinate of the candidate point, S=1, m is the number of candidate points;

[0090] Step 55: Screen the candidate points along the positive x-axis of the difference curve. When there is a difference curve value between the previous candidate point and the next candidate point that is less than the threshold of the previous candidate point, the previous candidate point is used as a feature point. Otherwise, the previous candidate point is deleted and the next candidate point is screened until all candidate points are screened and all feature points are obtained.

[0091] Further, in the present invention, in Step 5-1, the normalized weld curve is:

[0092]

[0093] where x s and y s represent the sets of the abscissa and ordinate of all points on the curve before normalization, and represent the abscissa and ordinate of the s i point on the curve before normalization, and represent the abscissa and ordinate of the sn i point on the curve after normalization.

[0094] Further, in the present invention, in Step 5-2, the difference curve:

[0095]

[0096] where and represent the abscissa and ordinate of the d i point on the difference curve.

[0097] Further, in the present invention, in Step 5-3, the formula for extracting the points where the growth rate of the y value in the difference curve decreases is:

[0098]

[0099] where and represent the abscissa and ordinate of the candidate point.

[0100] Specific application example:

[0101] The system hardware includes: a welding robot, a PC host computer, a laser tracker, etc.

[0102] Step 1: Before welding, grind, clean the workpiece to be welded, and fix it on the fixture;

[0103] Step 2: Fix the laser tracker on the robotic arm of the welding robot using a bracket;

[0104] Step 3: Before welding, complete the motion trajectory teaching of the robotic arm of the welding robot;

[0105] Step 4: Turn on the laser tracking sensor switch through the PC host computer to start scanning, and the laser transmits the laser line coordinate data to the PC host computer through tcp / ip communication;

[0106] Step 5: Calculate the upper left and right feature points, lower left and right feature points of the weld seam, and the coordinates of the starting, midpoint, and ending feature points of the weld bead by the method of the present invention;

[0107] Step 6: The PC host computer sends the weld seam feature point coordinate data to the welding robot through tcp / ip communication;

[0108] Step 7: The welding robot corrects the motion trajectory according to the weld seam feature point coordinates.

[0109] Although the present invention has been described herein with reference to particular embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the invention. It should thus be understood that numerous modifications may be made to the exemplary embodiments, and other arrangements may be devised, without departing from the spirit and scope of the invention as defined by the appended claims. It should be understood that the features described in the different dependent claims and in the present text may be combined in a manner different from that described in the original claims. It should also be understood that the features described in connection with a separate embodiment may be used in other described embodiments.

Claims

1. A real-time weld bead feature point recognition method based on geometric features, characterized in that: include: Step 1: Install the welding gun and laser and calibrate the hand-eye to obtain the weld area coordinate set corresponding to the laser; Step 2: extract the weld start point and weld end point from the weld area laser line coordinate set, and use the weld start point and weld end point to identify the weld anchor point; Step 3: Calculate the left and right feature points at the upper end of the weld using the weld anchor point, and then calculate the left and right feature points at the lower end of the weld using the left and right feature points at the upper end of the weld, and for welds with weld beads, when the distance between the left and right feature points at the lower end is greater than 3 mm, correct the left and right feature points at the lower end; and obtain the corrected left and right feature points at the lower end; Step 4: intercept the weld between the left and right feature points at the upper end, perform weighted average moving smoothing on the weld index, and identify the weld feature points. When the number of weld feature points is less than 3, it is determined that there is no weld. If the number of weld feature points meets 3, the three feature points are determined as the weld starting feature point, the midpoint feature point, and the weld end feature point according to their coordinates.

2. The real-time weld bead feature point recognition method based on geometric features according to claim 1 is characterized in that: In step 2, the method of identifying the weld anchor point using the weld start point and weld end point is: Select the starting point and end point of the laser line from the laser line coordinate set in the weld area, mark the points in the weld section, starting from the weld starting point, which is numbered 0, until the weld end point, which is numbered n, and calculate the anchor point of the weld section. The specific formula is: Among them, P0 represents the coordinate of the starting point of the weld seam, and P n represents the coordinate of the ending point of the weld seam, and P i represents the coordinate of the i-th point in the set of laser line coordinates of the weld seam area. represents the coordinate vector from the starting point of the weld seam to the i-th point. Among them, represents the vector of the coordinate vector from the starting point of the weld seam to the ending point P0P n of the weld seam, and i ⊥,max represents the anchor point number corresponding to the maximum vertical distance for this section of the weld seam, and argmax 0<i<n represents taking the serial number of the point corresponding to the maximum distance between all points 0 < i < n of this section of the weld seam and the line segment between the starting point and the ending point of the weld seam.

3. The real-time weld bead feature point recognition method based on geometric features according to claim 2 is characterized in that: In step 3, the method of calculating the left and right feature points at the upper end of the weld using the weld anchor point is: In step 3, the method of calculating the left and right feature points at the upper end of the weld using the weld anchor point is: Calculate each point P between the weld start point and the weld anchor point i =(x i1 ,y i1 )i1∈(0,a) to line segment P0P a The point with the largest vertical distance is identified as the left characteristic point of the upper end of the weld. The line segment P0P a Indicates the distance from the weld starting point P0 to the weld anchor point P a ; Calculate the value of each point P between the anchor point of the weld section and the end point of the weld section i2 =(x i2 ,y i2 )i∈(a,n) to line segment P a P n The point with the largest vertical distance is identified as the right characteristic point at the upper end of the weld. a P n Represents the line segment connecting the start and end points of the weld.

4. The real-time weld bead feature point recognition method based on geometric features according to claim 2 or 3, characterized in that: In step 3, the method for calculating the left and right characteristic points at the lower end of the weld is: Set the upper left feature point to P tl =(x tl ,y tl ), the upper right feature point is P tr =(x tr ,y tr ); Select all points in the weld between the left feature point at the upper end of the weld and the anchor point of the weld and connect them with the starting point of the weld, and calculate the P0P of each connection line. i3 Then connect the starting point of the weld with the left characteristic point at the upper end of the weld P0P tl , calculate the connection P0P tl The slope of Connect each P0P i The slope of the line P0P tl The maximum difference corresponds to the weld point at the left characteristic point P at the lower end of the weld. bl , the serial number is: Among them, P i =(x i3 ,y i3 )i3∈(tl,a) represents the i3th point in the weld between the upper left feature point and the anchor point of the weld, argmax i＝tl+1,tl+2,…,a Indicates that the line P0P is taken from all points in the weld between the upper left feature point of the weld and the anchor point of the weld to the weld starting point and the upper left feature point of the weld. tl The serial number of the point with the maximum distance between them; Select all points in the weld between the weld anchor point and the upper right feature point and connect them with the weld end point, and calculate the P of each connection line. i4 P n The slope of the weld is then connected by a line P tr P n , calculate the connection P tr P n The slope of each connecting line P i4 P n The slope of the line P tr P n The point with the largest slope difference is the right characteristic point P at the lower end of the weld. br , the serial number is: Among them, P i4 =(x i4 ,y i4 )i4∈(a,tr) represents the i4th point in the weld from the anchor point to the end point of the weld; argmax i＝a+1,a+2,…,tr Indicates that the line P between all points in the weld from the weld anchor point to the upper right feature point and the weld end point is taken. tr P n The serial number of the point with the maximum distance between them.

5. The real-time weld bead feature point recognition method based on geometric features according to claim 4 is characterized in that: In step 3, the method for correcting the left and right characteristic points at the lower end of the weld is: For welds with weld beads, if the distance between the lower left feature point and the lower right feature point exceeds 3mm, correction operation is performed, specifically: Correct the lower left feature point: Calculate all points on the weld from the upper left feature point to the lower right feature point to the line segment P tl P br The point with the largest vertical distance is the left characteristic point at the lower end of the weld after correction. The line segment P tl P br It is the line connecting the upper left feature point and the lower right feature point; Correct the lower right feature point: Calculate all laser points on the weld from the lower left feature point to the lower right feature point to the line segment P bl P tr The vertical distance is the point with the largest vertical distance as the right characteristic point of the lower end of the weld after correction. The line segment P bl P tr It is the line from the lower left feature point to the lower right feature point.

6. The real-time weld bead feature point recognition method based on geometric features according to claim 5 is characterized in that: In step 4, the specific method for identifying the characteristic points of the weld bead is: Step 51: intercept the weld between the upper left and right feature points, normalize the intercepted weld, and obtain a normalized weld curve; Step 52: Calculate the difference between the horizontal coordinate x and the vertical coordinate y of all points of the normalized weld curve to obtain a difference curve; Step 53: extracting points in the difference curve where the growth rate of the y-value of the ordinate decreases, and taking the points where the growth rate of the y-value of the ordinate decreases as candidate points; Step 54: Calculate the threshold of each candidate point: in, and Indicates the sn on the curve after normalization i+1 Dot and sn i The horizontal coordinate of the point, represents the ordinate of the candidate point, S=1, m is the number of candidate points; Step 55: Screen the candidate points along the positive x-axis of the difference curve. When there is a difference curve value between the previous candidate point and the next candidate point that is less than the threshold of the previous candidate point, the previous candidate point is used as a feature point. Otherwise, the previous candidate point is deleted and the next candidate point is screened until all candidate points are screened and all feature points are obtained.

7. The real-time weld bead feature point recognition method based on geometric features according to claim 6 is characterized in that: In step 51, the normalized weld curve is: Among them, x s and s Represents the set of horizontal and vertical coordinates of all points on the curve before normalization. and Represents the s on the curve before normalization i The horizontal and vertical coordinates of the point, and Indicates the sn on the curve after normalization i The horizontal and vertical coordinates of the point.

8. The real-time weld bead feature point recognition method based on geometric features according to claim 7 is characterized in that: In step 52, the difference curve: in, and Indicates the difference curve d i The horizontal and vertical coordinates of the point.

9. The real-time weld bead feature point recognition method based on geometric features according to claim 8, characterized in that: In step 53, the formula for extracting the point where the y value growth rate decreases in the difference curve is: in, and Represents the horizontal and vertical coordinates of the candidate point.

Images