Multi-layer multi-pass weld deviation identification and regulation method based on laser vision sensing

Abstract

The application discloses a multi-layer multi-pass weld deviation identification and regulation method based on laser vision sensing, which comprises the following steps: step 1: collecting molten pool cross-section image signals and feature point position signals through laser scanning, and solving a mathematical equation of the upper edge of the molten pool by using a matlab graphic method; step 2: calculating the absolute value of a straight line slope and the size of a fan-shaped area by using a two-point method and a definite integral respectively, and judging the welding gun swing direction by comparing the size relationship between the absolute value of the slope and the area in sequence, and feeding the size signal deviation to a control center to complete the centering of the welding gun; and step 3: fusing and extracting an optimal weld deviation value by using a weighted average method, with the molten pool cross-section data signals tracked by a previous layer of welds and the predicted fan-shaped molten pool cross-section mathematical model data signals, and continuously circulating the above fusion and extraction process until the weld tracking is completed, so that the application provides a new method for weld tracking.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of weld tracking, in particular to a multi-layer multi-pass weld deviation identification and regulation method based on laser vision sensing BACKGROUND

[0002] With the development of digital information technology, automatic intelligent weld tracking technology is a key technology for the development of modern welding robots, and welding robots will gradually replace manual work. Welding robots highlight their advantages in improving weld quality, reducing labor intensity and improving weld forming. Due to the poor welding site environment, thermal deformation, vibration and processing error factors during processing cannot accurately guarantee the position of the welding gun, so the weld tracking technology is an indispensable part of the automation and intelligentization of the welding process. The laser vision sensor installed in front of the welding gun scans the V-shaped groove to feed back the position signal of the welding gun in real time, and the single-chip microcomputer closed-loop control transmits the signal to the control center to drive the actuator to correct the position deviation of the welding gun, thereby improving the weld quality.

[0003] Patent CN113042863B uses a laser vision sensor to shoot a laser line weld image, obtains feature points through the weld image, determines the groove area through the determination of the feature point position, and then obtains the groove area to determine the weld trajectory point. According to the deviation between the identified trajectory point and the position of the welding gun, the weld tracking is performed. The method of determining the groove position according to the connection line between the feature points and the fitting laser line center line of the feature points is adopted, the point on the connection line between the two feature points on the two sides of the groove area is determined as the trajectory point, and then the weld tracking is performed according to the deviation between the identified trajectory point and the position of the welding gun. However, the extraction of the trajectory point adopts the area equal method, which may cause the weld surface to be not well formed due to the influence of the welding process, and there may be a problem that the trajectory point is not the center point of the weld although the areas are equal. It cannot be real-time feedback to the position of the welding gun swing, which affects the weld forming effect due to the lag factor. If the weld formed by the base layer is not smooth enough, it is difficult to extract the feature points of the filling layer as the basis for weld tracking.

[0004] Patent CN103955927B starts from the gray scale statistics of the local image, calculates the gray scale contrast of each point in the initial image to distinguish the interference stripes formed by the reflection light from the actual stripes, and further removes the residual interference stripes by repairing the area containing the actual light strip in the binary image through the image morphological method. In the region of interest of the initial image, the gray scale of each point is Taylor expanded in the cross section direction of the light strip, and the light strip center point is determined according to the Taylor expansion formula. The two straight line equations of the light strip on the surface of the fillet are accurately fitted by using the improved iterative least square method, so as to obtain the feature points of the fillet. Although this method can effectively improve the robustness and accuracy of the feature point extraction of the fillet, the whole process is very cumbersome and the calculation amount is huge, which is completely not suitable for real-time closed-loop feedback welding seam tracking process. SUMMARY

[0005] The purpose of the present application is to provide a multi-layer multi-pass welding seam deviation identification and regulation method based on laser vision sensing, which obtains a cross-sectional image through a laser vision sensor to calculate the slope of the feature points to determine the welding gun swing direction. When the slope does not meet the welding gun judgment condition, the position of the welding gun is further judged by using arc area integration, and the welded cross section is continuously fed back to the control center to drive the actuator to track in real time. Then the data signal of the upper edge of the first layer of backing welding is fused with the predicted fan-shaped molten pool cross-sectional function through the weighted average method to obtain the optimal data signal of the double signal as the data guidance of the next layer of filling layer, and the above process is repeatedly performed to improve the accuracy of real-time tracking of the welding seam, which plays an important role in multi-layer multi-pass welding seam tracking and surfacing.

[0006] The present application is specifically implemented by the following technical solutions: first, a laser vision sensor placed in front of the welding gun is used to obtain a V-shaped groove molten pool cross-sectional image signal, and then a matlab graphical method is used to solve the mathematical equation f AB and f BC The first step is shown in Figure 1 The present application first obtains a V-shaped groove molten pool cross-sectional image signal through a laser vision sensor placed in front of the welding gun, and then solves the mathematical equations f AB and f BC of the upper edge of the molten pool cross section based on the matlab graphical method, wherein:

[0007]

[0008]

[0009] Meanwhile, the characteristic point signal is analyzed from the molten pool cross-sectional image, the Cartesian coordinates of A(0, A1, A2), B(0, B1, B2) and C(0, C1, C2) are obtained through the determination of the characteristic point position signal, and the two-point mathematical method is used to calculate the straight lines l AB and lBC The absolute value of the slope |k AB | and |k BC |, such as Figure 3 and Figure 4 As shown:

[0010] straight line l AB The expression is:

[0011] y AB =k AB x AB +b AB

[0012] straight line l BC The expression is:

[0013] y BC =k BC x BC +b BC

[0014] And the absolute value of the slope |k AB The expression for | is:

[0015]

[0016] absolute value of slope |k AB The expression for | is:

[0017]

[0018] Finally, by comparing |k AB | and |k BC The size of the | indicates the direction of the welding torch's swing.

[0019] The specific steps include the following:

[0020] Step 1: By comparing |k AB | and |k BC The method for determining the direction of welding torch oscillation by measuring the size of the flame is as follows:

[0021] (1) When the absolute value of the slope of the straight line is |k AB |>|k BC |When, because of the straight line l AB absolute value of slope |k AB The larger the angle, the stronger the force, indicating that the left side is more tilted than the right side. Therefore, the center point of the welding torch swing (above point B) is closer to the right side of the V-shaped groove. Thus, the welding torch needs to be swung to the left so that the center point of the welding torch swing is at the weld position.

[0022] (2) When the absolute value of the slope of the straight line is |k AB |=|k BC|, the size ratio of the arc areas S1 and S2 calculated by the definite integral can be used to further determine the position of the welding gun swing direction.

[0023] (3) When the absolute value of the slope of the straight line |k AB |<|k BC |, the straight line l BC The greater the absolute value of the slope, the more inclined the right side than the left side, so the welding gun swing center point (above point B) is closer to the right side of the V-shaped groove, and thus the welding gun needs to be swung to the right to make the welding gun swing center point on the weld position.

[0024] Step 2: As known from the above (2), the two arc areas S1 and S2 can be obtained by using the mathematical calculation method of the definite integral, as shown in Figure 5 , which can be obtained as follows:

[0025]

[0026]

[0027] Further comparison of the sizes of the two arc areas can more accurately determine the direction in which the welding gun needs to be swung, and the specific determination method is as follows:

[0028] (1) When the arc area S1 > S2, it is known that the arc length AB is longer than the arc length BC, so it can be determined that the welding gun center point position should be better biased to the right side of the V-shaped groove, and at this time the welding gun needs to be adjusted to swing to the left to make the welding gun center point on the weld position.

[0029] (2) When the arc area S1 = S2, it is known that the arc length AB is the same as the arc length BC, so it is obvious that the welding gun center point position should be kept on the weld position, and at this time the welding gun does not need to swing left or right and only needs to be kept still to ensure the realization of the weld tracking.

[0030] (3) When the arc area S1 < S2, it is known that the arc length AB is shorter than the arc length BC, so it can be determined that the welding gun center point position should be better biased to the left side of the V-shaped groove, and at this time the welding gun needs to be adjusted to swing to the right to make the welding gun center point on the weld position.

[0031] Step 3: When the backing welding is completed, the molten pool cross section generated by the swing of the welding gun in the previous layer is fused with the existing mathematical equation model data signal of the fan-shaped molten pool cross section by using the weighted average method to extract the optimal data signal of the double signal, and the predicted fan-shaped area is:

[0032]

[0033] In the formula, n represents the molten pool cross section of the layer.

[0034] The above-mentioned weld tracking is repeated to realize that the real weld deviation value is constantly close to the weld deviation value in the ideal state, thereby improving the weld tracking precision.

[0035] The execution mechanism of the welding gun swing is controlled by the output signal from the single-chip microcomputer to drive the stepping motor driving circuit, so as to drive the execution mechanism of the welding gun swing to realize real-time weld tracking. Meanwhile, the actual weld tracking value is compared with the predicted value of the sector mathematical model in real time to realize deviation threshold calibration, real-time feedback to the weld, and real-time correction, so as to improve the weld quality and welding efficiency.

[0036] The sector molten pool cross-section mathematical equation model is that the welding current and voltage signals are collected by the Hall sensor, and then the welding current and voltage signals are established in a functional relationship with the sector molten pool cross-section area, wherein the sector model area is:

[0037]

[0038] In the formula, k 扇 represents the correlation coefficient related to the welding material and the welding wire and other factors; I represents the welding current; U represents the arc voltage; and V represents the welding speed.

[0039] The sector area formula is substituted into the established functional relationship, and the sector molten pool cross-section mathematical equation model signal is obtained.

[0040] The present application has the beneficial effects that: the present application provides a method combining the laser vision sensor and the mathematical calculation method of the slope and the area, the position of the welding gun is determined by comparing the slope and the arc area of the feature points collected by the laser vision sensor, then the output signal fed back to the center of the control system is applied to the cross slide execution mechanism, so as to control the welding gun to realize real-time accurate swing. Finally, when the filler layer welding is performed, the data signal of the upper layer groove upper edge equation and the predicted sector cross-section mathematical equation model signal are fused by the weighted average method to realize the optimal data signal extraction of the double-layer signal. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 is the flow chart of the deviation identification and regulation method of the multi-layer multi-pass weld based on the laser vision sensing of the present application.

[0042] Figure 2 is the control principle diagram of the deviation identification and regulation method of the multi-layer multi-pass weld based on the laser vision sensing of the present application.

[0043] Figure 3 and Figure 4The diagram shows the welding torch swinging directions, which are biased to the left and right, respectively, in the method for identifying and controlling deviations in multi-layer, multi-pass welds based on laser vision sensing, according to the present invention.

[0044] Figure 5 This is a diagram showing the special position of the welding torch oscillation in the method for deviation identification and control of multi-layer and multi-pass welds based on laser vision sensing, which is based on the present invention.

[0045] Figure 6 This is the predictive sector equation diagram of the deviation identification and control method for multi-layer and multi-pass welds based on laser vision sensing in this invention. Detailed Implementation

[0046] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0047] See also Figure 1 The method in this embodiment firstly involves a method for identifying and controlling deviations in multi-layer, multi-pass welds. This includes acquiring the cross-sectional image signal of the V-groove molten pool using a laser vision sensor placed in front of the welding torch, and then solving the mathematical equation f along the cross-section of the molten pool using a MATLAB graphical method. AB and f BC Simultaneously, feature point signals are extracted from the cross-sectional image of the molten pool. By determining the position signals of the feature points, the Cartesian coordinates of A(0, a1, a2), B(0, b1, b2), and C(0, c1, c2) can be obtained. The straight line l is then calculated using a two-point mathematical method. AB and l BC The absolute value of the slope |k AB | and |k BC Finally, by comparing |k AB | and |k BC The direction of the welding torch's oscillation is determined by the magnitude of the [value], and the specific steps include the following:

[0048] Step 1: By comparing |k AB | and |k BC The method for determining the direction of welding torch oscillation by the magnitude of the | is detailed in the following article. Figure 3 , Figure 4 as follows:

[0049] (1) When the absolute value of the slope of the straight line is |k AB |>|k BC |When, because of the straight line l AB absolute value of slope |k AB The larger the angle, the stronger the force, indicating that the left side is more tilted than the right side. Therefore, the center point of the welding torch swing (above point B) is closer to the right side of the V-shaped groove. Thus, the welding torch needs to be swung to the left so that the center point of the welding torch swing is at the weld position.

[0050] (2) When the absolute value of the slope of the straight line is |kAB |=|k BC When |, the ratio of the arc areas S1 and S2 obtained by definite integral can be used to further determine the position of the welding torch swing direction.

[0051] (3) When the absolute value of the slope of the straight line is |k AB |<|k BC |When, because of the straight line l BC A larger absolute value of the slope indicates that the right side is more inclined than the left side. Therefore, the center point of the welding torch swing (above point B) is closer to the right side of the V-shaped groove. Thus, the welding torch needs to be swung to the right so that the center point of the welding torch swing is at the weld position.

[0052] Step 2: As shown in (2) above, by using the mathematical calculation method of definite integrals to obtain the two arc areas S1 and S2, the size of the two arc areas can be compared further to more accurately determine the direction in which the welding torch needs to be swung. For the specific judgment method, please refer to [link to relevant documentation]. Figure 5 as follows:

[0053] (1) When the arc area S1 > S2, it can be known that the arc length AB is longer than the arc length BC. Therefore, it can be determined that the center point of the welding torch should be more biased towards the V-shaped groove on the right side. At this time, it is necessary to adjust the welding torch to swing to the left so that the center point of the welding torch is placed on the weld.

[0054] (2) When the area of ​​the arc S1 = S2, it means that the arc length AB and the arc length BC are the same. Therefore, it is obvious that the center point of the welding gun should be kept at the weld position. At this time, the welding gun does not need to swing left or right, but only needs to remain still to ensure weld tracking.

[0055] (3) When the arc area S1 < S2, it can be seen that the arc length AB is shorter than the arc length BC. Therefore, it can be determined that the center point of the welding torch should be more biased towards the V-shaped groove on the left side. At this time, it is necessary to adjust the welding torch to swing to the right so that the center point of the welding torch is placed on the weld.

[0056] Step 3: After the root pass welding is completed, when welding the filler layer, the data signal along the equation of the cross-section of the molten pool generated by the previous welding torch swing and the data signal of the mathematical equation model of the predicted fan-shaped molten pool cross-section are fused using a weighted average method to extract the optimal data signal of the dual signals. This provides an accurate reference model for the next layer of weld tracking. By repeating the above weld tracking work, the actual weld deviation value can be made to continuously approach the weld deviation value under the ideal state, thereby improving the weld tracking accuracy.

[0057] See Figure 2, the execution mechanism of the welding gun swing is a step motor driven circuit receiving output signals from the single-chip microcomputer to drive the execution mechanism of the welding gun swing to realize real-time weld tracking. Meanwhile, the actual weld tracking value is compared with the predicted value of the sector mathematical model in real time to calibrate the deviation threshold, and real-time feedback is provided for real-time weld deviation correction to improve weld quality and welding efficiency.

[0058] In case 1, multi-layer multi-pass V-groove weld tracking is performed. A key technology for multi-layer multi-pass weld tracking is to constantly plan the welding route, and the weld deviation information during each welding process cannot be guaranteed to be a constant value, that is, weld tracking needs to be performed at all times. Therefore, the execution mechanism of the welding gun swing is a step motor driven circuit receiving output signals from the single-chip microcomputer to drive the execution mechanism of the welding gun swing to realize real-time weld tracking. Meanwhile, the actual weld tracking value is compared with the predicted value of the sector mathematical model in real time to calibrate the deviation threshold, and real-time feedback is provided for real-time weld deviation correction to improve weld quality and welding efficiency. Figure 1 、 Figure 2 It can be seen that the double-layer deviation judgment method is adopted to ensure the accuracy of the weld deviation, and the optimal weld deviation information can be extracted by using the introduced predicted sector cross-section model, providing a better choice for multi-layer multi-pass weld tracking. Since this method uses size judgment to execute the welding gun swing position, the calculation accuracy is higher, the response is more sensitive, and the cost is lower.

[0059] In real-time case 2, surfacing is performed. Surfacing in additive manufacturing has developed rapidly in recent years and is a key project supported by the state. One of the key technologies for surfacing is how to perform multi-layer multi-pass welding on the weld, and how to quickly and effectively extract the weld surface deviation information of different layers. Referring to Figure 3 , Figure 4 , Figure 5 The arc of the welding gun is swung by the swing device, and the characteristic points on the weld surface are collected to prepare for the next calculation of the weld equation. Two mathematical size judgment methods are adopted. The computer is used to calculate the mathematical problems in the whole process, and then the swing direction of the welding gun is determined by comparing the values, so as to realize the multi-layer surfacing process. This method can simply and quickly calculate the direction in which the welding gun should be swung, and the operation process is simple, improving the responsiveness and real-time performance of the surfacing welding process.

Claims

1. A method for identifying and controlling deviations in multi-layer, multi-pass welds based on a laser vision sensor, characterized in that: A laser vision sensor positioned in front of the welding torch acquires an image signal of the cross-section of the V-groove molten pool. Then, the mathematical equations along the cross-section of the molten pool are solved using a graphical method based on MATLAB. and Simultaneously, feature point signals are extracted from the cross-sectional image of the molten pool. A(0, ...) can be obtained by determining the position signals of the feature points. , B(0) , ) and C(0, , Using Cartesian coordinates, calculate the line using the two-point mathematical method. , absolute value of slope , Meanwhile, the area is solved using definite integrals. , Finally, what will be obtained , and , The deviation value is input into the actuator, which outputs a signal to drive the welding torch to swing in the direction of the swing, thereby realizing weld deviation identification and weld tracking control. The specific steps are as follows: Step 1: By comparison The method for determining the direction of welding torch oscillation based on the magnitude of the force is as follows: (1) When the absolute value of the slope of the straight line > At that time, due to the straight line absolute value of slope The larger the force, the more the left side is tilted than the right side. Therefore, the center point of the welding torch swing is closer to the right side of the V-groove. Thus, the welding torch needs to be swung to the left so that the center point of the welding torch swing is at the weld position. (2) When the absolute value of the slope of the straight line = When, the area of ​​the arc can be calculated by definite integral. and The size ratio is used to further determine the direction and position of the welding torch swing; (3) When the absolute value of the slope of the straight line < At that time, due to the straight line A larger absolute value of the slope indicates that the right side is more inclined than the left. Therefore, the center point of the welding torch swing is closer to the left side of the V-groove. Thus, the welding torch needs to be swung to the right so that the center point of the welding torch swing is at the weld position. Step 2: As can be seen from (2) above, the areas of the two arcs can be obtained by using the mathematical calculation method of definite integrals. and To further compare the sizes of the two arc areas, the direction in which the welding torch needs to be swung can be determined more accurately. The specific method for this determination is as follows: (1) When the area of ​​the arc > When it is known that the arc length AB is longer than the arc length BC, it can be determined that the center point of the welding torch should be more biased towards the V-shaped groove on the right side. At this time, it is necessary to adjust the welding torch to swing to the left so that the center point of the welding torch is placed on the weld. (2) When the area of ​​the arc = When the arc length AB and arc length BC are the same, it is obvious that the center point of the welding torch should be kept at the weld position. At this time, the welding torch does not need to swing left or right, but only needs to remain still to ensure weld tracking. (3) When the area of ​​the arc < When the arc length AB is shorter than the arc length BC, it can be determined that the center point of the welding torch should be more biased towards the left side of the V-groove. At this point, the welding torch needs to be adjusted to swing to the right so that the center point of the welding torch is placed on the weld seam. Step 3: After the root pass welding is completed, when welding the filler layer, the data signal along the equation of the cross-section of the molten pool generated by the previous layer welding torch swing and the data signal of the mathematical equation model of the predicted fan-shaped molten pool cross-section are fused using a weighted average method to achieve optimal data signal extraction of the dual signals. This provides a reference model for the next layer of weld tracking. By repeating the previous work, the weld deviation value can be made closer to the ideal weld deviation value, thereby continuously improving the deviation identification accuracy of multi-layer and multi-pass welds.

2. The weld deviation identification and control method according to claim 1, characterized in that: The mathematical equation model for the predicted cross-section of the sector-shaped molten pool refers to: acquiring welding current and voltage signals through a Hall element, establishing a functional relationship between the welding current and voltage signal values ​​and the cross-sectional area of ​​the sector-shaped molten pool, and then substituting the area formula of the sector into the established functional relationship to obtain the mathematical equation model for the cross-section of the sector-shaped molten pool.

Images