A method and system for adjusting welding process parameters of a lap joint

Abstract

The present invention discloses a method and system for adjusting welding process parameters of a lap joint, which relates to the field of welding technology and includes the following steps: determining characteristic positions during the movement of a robot in the welding process and recording the characteristic positions; vertically correcting the angle between the laser line and the weld line; building and training a control model, which is a neural network model; and realizing real-time adjustment of welding process parameters during the welding process according to a predetermined control process. The present invention establishes a real-time adjustment system for welding process parameters with a robot controller as the core, and performs secondary development on the robot controller based on the robot system architecture language to realize parallel operation of multiple programs such as robot movement, welding program, gap sampling, and welding process parameter calculation program. By vertically correcting the laser line and the weld line, it is ensured that the laser line is perpendicular to the weld, reducing detection errors, thereby obtaining an ideal welding joint, ensuring the consistency of weld formation, and reducing defects such as holes and burn-throughs.

Description

Technical Field

[0001] The present invention relates to the field of welding technology, and particularly to a method and system for adjusting welding process parameters of a lap joint. Background Art

[0002] In welding, due to the differences in the thickness, structure and service conditions of the welded parts, the joint types and groove forms are also different. The welding joint types include: butt joint, T-joint, corner joint and lap joint, etc. A lap joint is a joint formed by partial overlap of two parts.

[0003] Regarding the assembly situation of the gradually changing weld gap of a carbon steel butt joint, Gu Fan proposed in "Adaptive Planning Method for Variable Gap Welding of Robotic MAG Welding Based on Laser Sensing" to use a laser vision sensor to obtain the image of the groove, extract the gap information, and achieve multi-layer and multi-pass welding of the variable gap weld bead by adjusting the welding torch posture; however, for each weld bead, the welding torch posture and welding process parameters remain fixed.

[0004] Regarding an aluminum alloy butt joint, Fan Chongjian proposed in "Research on Visual Feature Acquisition and Intelligent Control of Variable Gap Aluminum Alloy Pulsed GTAW Molten Pool" to use the method of molten pool image sensing, designed a PID controller with variable parameters of gap feedforward, and controlled the back bead width and front bead height of the molten pool; however, this control system is complex and uses an industrial control computer as the controller.

[0005] Yan Minzhi's "Adaptive Control of Process Parameters Based on 3D Weld Modeling and Autonomous Programming Welding" used the BP neural network algorithm to model the corresponding relationship between the lap gap and welding current and voltage, and predicted the MAG welding process parameters for a lap gap of 0 - 2.1 mm, but did not establish a complete control system.

[0006] Therefore, those skilled in the art are committed to developing a method and system for adjusting welding process parameters of a lap joint. Summary of the Invention

[0007] In view of the above-mentioned defects of the prior art, the technical problem to be solved by the present invention is that the welding process parameters cannot be adjusted in real time during welding, resulting in defects such as holes and burn-through in the weld when the gap fluctuates.

[0008] To achieve the above object, the present invention provides a method for adjusting welding process parameters of a lap joint, characterized in that the method comprises the following steps:

[0009] S101: Determine the characteristic positions during the movement of the robot in the welding process, and record the characteristic positions;

[0010] S103: Vertical correction of the angle between the laser line and the weld line;

[0011] S105: Building and training a control model, wherein the control model is a neural network model;

[0012] S107: According to a predetermined control process, the welding process parameters are adjusted in real time during the welding process.

[0013] Furthermore, in step S101, the characteristic positions include a sampling starting point, a weld starting point, a welding process parameter change point, a sampling ending point, and a weld end point, wherein:

[0014] The sampling starting point, the laser line coincides with the welding starting point;

[0015] Starting the welding seam, starting the welding machine to start arcing and welding;

[0016] The welding process parameter change points are set on the weld straight line, starting from the weld starting point, at intervals of fixed weld length, P i (x) = P1(x)*(1+i*ΔL);

[0017] The sampling end point is where the laser line coincides with the welding end point;

[0018] The weld end point ends the welding and closes the welding arc;

[0019] Among them, P i (x) is the position of the welding process parameter change point on the X-axis, P1(x) is the position of the weld starting point on the X-axis, ΔL is the change in weld length, and i is a variable.

[0020] Furthermore, step S103 includes the following sub-steps:

[0021] S1031: Calibrate the initial laser line slope and the distance between the laser sensor and the welding gun: move the welding gun to the starting point of the weld, turn on the laser sensor to output laser, and record the initial laser line slope k1 and the distance d between the laser line and the welding gun;

[0022] S1032: Calculate the positional relationship between the laser line and the welding gun wire tip, as follows:

[0023] P laser (x) = P tcp (x)+d

[0024] Among them, P laser (x) is the position of the laser line on the X axis, P tcp (x) is the position of the welding gun wire tip on the X axis, and d is the distance between the laser line and the welding gun;

[0025] S1033: Adjust the angle between the laser line and the weld line to 90°, and update the rotation angle of the welding gun along the X-axis.

[0026] Furthermore, in step S1033, the rotation angle of the welding gun along the X-axis is calculated using the following method:

[0027]

[0028] Where w′ is the rotation angle of the welding gun along the X-axis, k1 is the initial laser line slope, k2 is the weld line slope, k2 = (y2-y1) / (x2-x1), (x1, y1) are the coordinates of the weld starting point, and (x2, y2) are the weld end point.

[0029] Furthermore, in step S105, the control model is an automatic adjustment control model of the lap joint welding process parameters, including input parameters, control coefficients and output parameters, wherein the input parameters are control targets, the control targets include weld width, weld height, weld penetration and lap gap, the control coefficients include neuron weights and neuron biases, and the output parameters are control quantities of welding process parameters, and the control quantities include welding current, welding speed and welding gun swing.

[0030] Furthermore, the relevant parameters of the trained control model are input in sequence, and output parameters are obtained through calculation. The output parameters include the values of the welding process parameters, specifically:

[0031] y_data[l,1]=W2[l,k]×hid_data[k,1]+theta2[l,1]

[0032] Among them, y_data is the output layer parameter matrix, W2[l,k] is the weight coefficient matrix of the intermediate layer neuron node, the order is l*k, hid_data is the hidden layer parameter matrix,

[0033]

[0034] Among them, theta2[l,1] is the bias coefficient matrix of the intermediate layer neuron nodes, with an order of l*1, W1[k,j] is the weight coefficient matrix of the input layer neuron nodes, with an order of k*j, in_date is the input layer parameter matrix, with an order of j*1, theta1[k,1] is the bias coefficient matrix of the input layer neuron nodes, with an order of k*1, j is the number of input layer neurons, k is the number of hidden layer neurons, and l is the number of output layer neurons.

[0035] Furthermore, when building the neural network model, the following steps are included:

[0036] S1051: Design a welding process test according to welding indexes, and establish a database of welding process parameters and weld formation. The welding indexes include lap joint gap, weld width, and weld penetration depth.

[0037] S1052: Determine the input layer parameters and output layer parameters of the neural network model according to the database.

[0038] S1053: Establish a neural network model in the industrial control computer and train it. After the model converges, extract the neuron weights and neuron biases.

[0039] Furthermore, in the step S107, the control process includes a main process and a calculation process. The main process includes the robot movement and welding process. The calculation process includes gap sampling and welding process parameter calculation. By accessing and operating on each register in the robot controller, the real-time adjustment of the welding process parameters during the welding process is achieved. The real-time adjustment is to adjust the welding process parameters once every fixed weld length.

[0040] Furthermore, the main process and the calculation process cooperate to achieve the adjustment of the lap joint welding process parameters, which specifically includes the following steps:

[0041] S1071: The system calls the main process, and the robot is located at the operation origin.

[0042] S1072: The robot starts to move. When the robot reaches the sampling starting point, turn on the laser sensor and start the calculation process. The main process and the calculation process run in parallel.

[0043] S1073: The robot continues to move. When the robot reaches the weld starting point, start the welding power supply and start welding. The welding process parameters are obtained from the calculation process.

[0044] S1074: When the robot reaches the welding process parameter change point, update the welding process parameters.

[0045] S1075: When the robot reaches the sampling end point, turn off the laser sensor.

[0046] S1076: When the robot reaches the weld end point, extinguish the arc and turn off the welding power supply.

[0047] On the other hand, the present invention also provides a system for adjusting lap joint welding process parameters, which is characterized in that the system uses the method for adjusting lap joint welding process parameters provided by the present invention. The system includes a robot, a laser sensor, a welding power supply, and an industrial control computer, where

[0048] The robot, including a robot controller and a robot body, is a motion execution mechanism of the welding system, responsible for adjusting the welding speed. A laser sensor and a welding torch are installed on the robot body;

[0049] The laser sensor samples the lap joint gap. It is installed on the robot body. The laser sensor sends the sampled signal to the robot controller, and the laser sensor outputs a laser line;

[0050] The welding power source provides the welding current and welding voltage during welding. The output value of the welding power source is controlled by the robot controller;

[0051] The industrial control computer completes the construction and training of the neural network model.

[0052] In a preferred embodiment of the present invention, compared with the prior art, the present invention has the following beneficial effects:

[0053] 1. The present invention performs secondary development on the robot controller based on the robot system architecture language, enabling the parallel operation of multiple programs such as robot movement, welding program, gap sampling, and welding process parameter calculation program, ensuring an ideal welding joint, ensuring the consistency of weld formation, and reducing defects such as holes and burn-through;

[0054] 2. The present invention corrects the perpendicularity between the laser line and the weld line. After calculating the angle between the laser line and the weld line, the rotation angle of the welding torch is adjusted, so as to ensure that the angle between the laser line and the weld bead direction always remains 90°, ensuring that the laser line is perpendicular to the weld, and reducing the detection error.

[0055] The following will further illustrate the concept, specific structure, and technical effects of the present invention in conjunction with the drawings to fully understand the purpose, features, and effects of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 is a schematic flow chart of the method for adjusting the welding process parameters of the lap joint in a preferred embodiment of the present invention;

[0057] Figure 2 is a schematic diagram of the lap joint in a preferred embodiment of the present invention;

[0058] Figure 3 is a schematic diagram of the perpendicular correction of the laser line and the weld bead direction vector in a preferred embodiment of the present invention;

[0059] Figure 4 is a schematic control flow chart in a preferred embodiment of the present invention;

[0060] Figure 5It is a schematic diagram of the movement during the robotic welding process of a preferred embodiment of the present invention;

[0061] Figure 6 It is a schematic diagram of the composition of a lap joint welding process parameter adjustment system of a preferred embodiment of the present invention.

[0062] Among them, 1 - robot, 11 - robot body, 12 - robot controller, 2 - laser sensor, 3 - welding power source, 31 - welding torch. Specific embodiments

[0063] The following introduces multiple preferred embodiments of the present invention with reference to the accompanying drawings of the specification, making its technical content clearer and easier to understand. The present invention can be embodied in many different forms of embodiments, and the protection scope of the present invention is not limited to the embodiments mentioned in the text.

[0064] In the drawings, components with the same structure are denoted by the same reference numerals, and components with similar structures or functions everywhere are denoted by similar reference numerals. The size and thickness of each component shown in the drawings are arbitrarily shown, and the present invention does not limit the size and thickness of each component. In order to make the illustration clearer, the thickness of some parts in the drawings is appropriately exaggerated.

[0065] As Figure 1 shown, a lap joint welding process parameter adjustment method provided by an embodiment of the present invention includes the following steps:

[0066] S101: Determine the characteristic positions during the movement of the robot in the welding process and record these characteristic positions.

[0067] The above-mentioned characteristic positions include the sampling starting point, the weld starting point, the welding process parameter change point, the sampling ending point, and the weld ending point. Among them, at the sampling starting point, the laser line coincides with the welding starting point; at the weld starting point, the welding machine is controlled to start arcing and start welding; the welding process parameter change point is on the weld straight line, starting from the weld starting point, set at intervals of a fixed weld length. At the position of the welding process parameter change point, the welding process parameters are adjusted to obtain better welding quality. The sampling ending point is when the laser line coincides with the welding ending point, and at the weld ending point, the welding is ended and the welding arc is turned off. Among them, the welding process parameter change point is calculated using the following formula:

[0068] P i (x) = P1(x) * (1 + i * ΔL)

[0069] P i (x) is the position of the welding process parameter change point on the X-axis, P1(x) is the position of the weld starting point on the X-axis, ΔL is the change amount of the weld length, and i is a variable.

[0070] As Figure 5As shown, the characteristic positions of the movement in the robot welding process, where

[0071] Po - Robot operation origin;

[0072] P0 - Sampling starting point, where the laser line coincides with the welding starting point;

[0073] P1 - Weld starting point, control the welding machine to start welding;

[0074] Pi - Welding process parameter change point, on the straight weld line, starting from the weld starting point, at a fixed interval of the weld length;

[0075] P3 - Sampling end point, where the laser line coincides with the welding end point;

[0076] P2 - Weld end point, turn off the welding arc.

[0077] S103: Vertical correction of the angle between the laser line and the straight weld line.

[0078] As Figure 2 、 3 shown, the vertical correction between the laser line and the straight weld line includes the following sub - steps:

[0079] S1031: Calibrate the initial slope of the laser line and the distance between the laser sensor and the welding torch: Move the welding torch to the weld starting point, turn on the laser sensor to output laser, and record the initial slope k1 of the laser line and the distance d between the laser line and the welding torch;

[0080] S1032: Calculate the positional relationship between the laser line and the tip of the welding wire of the welding torch, specifically as follows:

[0081] P laser (x) = P tcp (x) + d

[0082] where, P laser (x) is the position of the laser line on the X - axis, P tcp (x) is the position of the tip of the welding wire of the welding torch on the X - axis, and d is the distance between the laser line and the welding torch;

[0083] S1033: Adjust the angle between the laser line and the straight weld line to 90°, and update the rotational angle of the welding torch along the X - axis.

[0084] The rotational angle of the welding torch along the X - axis is calculated by the following method:

[0085]

[0086] Among them, w′ is the rotational angle of the welding torch along the X-axis, k1 is the initial laser line slope, k2 is the weld line slope, k2 = (y2 - y1) / (x2 - x1), (x1, y1) is the coordinate of the weld starting point position, and (x2, y2) is the weld end position.

[0087] S105: Build and train a control model, and the control model is a neural network model.

[0088] The above control model is an automatic adjustment control model for lap joint welding process parameters, including input parameters, control coefficients, and output parameters. Among them, the input parameters are the regulation targets, and the regulation targets include parameters such as weld width, weld height, weld penetration, and lap gap. The control coefficients include parameters such as neuron weights and neuron biases. The output parameters are the control quantities of welding process parameters, including parameters such as welding current, welding speed, and welding torch swing amplitude.

[0089] After building and training the neural network model, input the relevant parameters of the trained control model in sequence. Through calculation, the output parameters are obtained. The output parameters include the values of welding process parameters, specifically:

[0090] y_data[l,1] = W2[l,k] × hid_data[k,1] + theta2[l,1]

[0091] Among them, y_data is the output layer parameter matrix, W2[l,k] is the weight coefficient matrix of the intermediate layer neuron nodes, with the order of l*k, and hid_data is the hidden layer parameter matrix.

[0092]

[0093] Among them, theta2[l,1] is the bias coefficient matrix of the intermediate layer neuron nodes, with the order of l*1, W1[k,j] is the weight coefficient matrix of the input layer neuron nodes, with the order of k*j, in_date is the input layer parameter matrix, with the order of j*1, theta1[k,1] is the bias coefficient matrix of the input layer neuron nodes, with the order of k*1, j is the number of input layer neurons, k is the number of hidden layer neurons, and l is the number of output layer neurons.

[0094] When building the above neural network model, it includes the following steps:

[0095] S1051: Design a welding process experiment according to welding indicators, and establish a database of welding process parameters and weld formation. The welding indicators include parameters such as lap gap, weld width, and weld penetration.

[0096] S1052: Determine the input layer parameters and output layer parameters of the neural network model according to the database.

[0097] S1053: Establish a neural network model in the industrial control computer and train it. After the model converges, extract the neuron weights and neuron bias parameters.

[0098] S107: Implement real-time adjustment of the welding process parameters during the welding process according to a predetermined control flow.

[0099] As Figure 4 shown, the above control flow includes a main flow and a calculation flow. The main flow includes the robot movement and the welding process. The calculation flow includes gap sampling and welding process parameter calculation. Through accessing and operating on each register in the robot controller, real-time adjustment of the welding process parameters during the welding process is achieved, and the welding process parameters are adjusted once every fixed weld length.

[0100] Through the cooperation of the main flow and the calculation flow, adjustment of the welding process parameters for the lap joint is achieved, specifically including the following steps:

[0101] S1071: The system calls the main flow, and the robot is at the operation origin.

[0102] S1072: The robot starts to move. When the robot reaches the sampling starting point, turn on the laser sensor and start the calculation flow. The main flow and the calculation flow run in parallel.

[0103] S1073: The robot continues to move. When the robot reaches the weld starting point, start the welding power supply and start welding. The welding process parameters are obtained from the calculation flow.

[0104] S1074: When the robot reaches the welding process parameter change point, update the welding process parameters.

[0105] S1075: When the robot reaches the sampling end point, turn off the laser sensor.

[0106] S1076: When the robot reaches the weld end point, extinguish the arc and turn off the welding power supply.

[0107] As Figure 6 、 2As shown in the figure, the embodiment of the present invention also provides a lap joint welding process parameter adjustment system. This system uses a lap joint welding process parameter adjustment method, and includes a robot, a laser sensor, a welding power source, and an industrial control computer. Among them, the robot includes a robot controller and a robot body, which is the motion execution mechanism of the welding system and is responsible for adjusting the welding speed. A laser sensor and a welding torch are installed on the robot body; the laser sensor realizes the sampling of the lap joint gap and is installed on the robot body. The laser sensor sends the sampled signal to the robot controller, and the laser sensor outputs a laser line; the welding power source provides the welding current and welding voltage during welding, and the output value of the welding power source is controlled by the robot controller; the industrial control computer completes the construction and training of the neural network model.

[0108] A lap joint welding process parameter adjustment method and system provided by a preferred embodiment of the present invention aim at the problem that the prior art cannot adjust the welding process parameters in real time during welding, resulting in defects such as holes and burn-throughs in the weld seam when the gap fluctuates. A real-time adjustment system for welding process parameters with the robot controller as the core is established. Based on the robot system architecture language, secondary development is carried out on the robot controller to realize the parallel operation of multiple programs such as robot movement, welding program, gap sampling, and welding process parameter calculation program. Based on the robot system architecture language, secondary programming development is carried out on the robot controller, and multiple programs run in a parallel multi-threaded manner to ensure an ideal welding joint, ensure the consistency of the weld seam formation, and reduce the defects of holes and burn-throughs.

[0109] Preferably, when the laser line is not perpendicular to the weld seam straight line, the measured lap joint gap error is large and does not match the actual gap, and the detection accuracy needs to be improved. In the present invention, after calculating the angle between the laser line and the weld seam straight line through the perpendicular correction of the laser line and the weld seam straight line, the rotation angle of the welding torch is adjusted, so as to ensure that the angle between the laser line and the weld bead direction always remains 90°, ensure that the laser line is perpendicular to the weld seam, and reduce the detection error.

[0110] Aiming at the problems that most of the current systems using neural network control need to rely on industrial control computers, etc. to implement, the system is complex and has poor portability. In the present invention, the complex neural network calculation algorithm is embedded and transplanted into the robot controller for execution. The weights and biases of each neuron node in the neural network model obtained after training are stored in the robot controller, and the result of the output layer is solved according to matrix calculation. The control system structure is simplified and can be realized based on the secondary development in the robot controller;

[0111] In view of the inherent deviation between the position of the laser line and the TCP position of the welding torch, there is no good solution in the prior art on how to ensure the correspondence between the sampling position of the gap and the welding process parameters. In view of the fact that the laser line is in front of the welding torch, the present invention first obtains the gap, and then refreshes the welding process parameters after the TCP of the welding torch moves to the corresponding position, so as to ensure that the output welding process parameters match the gap value at the corresponding position, calibrate the position deviation between the laser line and the TCP of the welding torch, record it in the corresponding register, calculate the position of the laser line and the real-time position of the TCP of the welding torch by reading the position system variables of the robot in real time, and the control is more accurate, ensuring the matching between the gap and the welding process parameters, and the control effect is better.

[0112] The following will combine the preferred embodiments of the present invention to describe the present invention in detail.

[0113] Embodiment 1

[0114] A method for adjusting welding process parameters of a lap joint provided by a preferred embodiment of the present invention includes working steps:

[0115] Step 1: Connect all components and determine the characteristic positions during the movement of the robot 1.

[0116] The specific method is as follows:

[0117] Determine the weld start point P1, sampling start point P0, sampling end point P3 and weld end point P2 according to the position of the weld respectively, and calculate the welding process parameter change point Pi.

[0118] The specific definitions and characteristics of each point are as follows:

[0119] P0 - Sampling start point, where the laser line coincides with the welding start point.

[0120] Since the installation position of the laser sensor 2 is in front of the welding torch 31, turn on the laser sensor 2 and operate the robot 1 to move. When the position of the laser line coincides with the weld start point, the position of this point is the sampling start point position, recorded as P0. Set the sampling start point position register PR0 in the robot controller 12, and PR0 is used to save the coordinates of the sampling start point P0.

[0121] P1 - Weld start point, control the welding machine to start welding with arc starting.

[0122] Pi - Welding process parameter change point, on the weld straight line, starting from the weld start point, at intervals of a fixed weld length ΔL. The position of Pi is calculated according to the position of the weld start point P1, and the formula is as follows:

[0123] P i (x) = P1(x) * (1 + i * ΔL)

[0124] Among them, Pi (x) is the position of the welding process parameter change point on the X-axis, P1(x) is the position of the weld start point on the X-axis, ΔL is the fixed weld length, and i is a variable.

[0125] The fixed weld length ΔL can be set according to the actual welding control effect. In the preferred embodiment of the present application, ΔL is 10 mm.

[0126] P3 - Sampling end point. The laser line position coincides with the welding end point. At this time, the position of this point is the sampling end point position, recorded as P3. Set the sampling end point position register PR3 in the robot controller 12. PR3 is used to save the coordinates of the sampling end point position P3.

[0127] P2 - Weld end point, turn off the welding arc.

[0128] Step 2: Laser line perpendicular correction to the weld line

[0129] Step 2.1: Calibrate the slope of the laser line, the distance between the laser sensor 2 and the welding torch 31, and the positional relationship between the laser line and the tip of the welding torch wire (TCP).

[0130] Set the initial laser line slope register R01 and the light-gun distance register R02 in the robot controller 12 to save the above parameter values respectively.

[0131] The calibration method of the slope of the laser line is as follows: When the posture of the welding torch 31 is set to (180, 0, 0), move the welding torch 31 to the start point P1 of the weld, turn on the laser sensor 2 to output the laser line, record the slope k1 of the laser line and the distance d between the laser line and the welding torch 31, and record the results in the initial laser line slope k1 register R01 and the light-gun distance d register R02, as Figure 3 shown. Among them, in the tool coordinate system of the robot 1, the posture of the welding torch 31 is the rotation angle of the tip of the welding torch around the X, Y, and Z axes. The posture of the welding torch 31 is (180, 0, 0), which means: the rotation angle of the welding torch 31 along the X-axis is 180°, the rotation angle along the Y-axis is 0°, and the rotation angle along the Z-axis is 0°. The calibration method of the positional relationship between the laser line and the tip of the welding torch wire (TCP) can refer to relevant literature.

[0132] According to the installation method of the laser line and the welding torch 31, it can be determined that the position of the laser line and the tip of the welding torch only has a deviation on the X-axis, and other coordinates are the same as the tip of the welding torch TCP. The position of the tip of the welding torch TCP can be obtained by reading the system parameters of the robot 1. According to the distance d between the laser line and the welding torch 31, the positional coordinate relationship between the laser line and the welding torch 31 on the X-axis can be calculated as follows:

[0133] P laser (x) = Ptcp P(x)+d

[0134] Where, P laser (x) is the position of the laser line on the X-axis, P tcp (x) is the position of the tip of the welding wire of the welding torch 31 on the X-axis, and d is the distance between the laser line and the welding torch 31.

[0135] Step 2.2: Perpendicular correction of the laser line and the weld bead direction vector.

[0136] Set a weld position register in the robot controller 12 to record the weld position values, including a weld start position register PR1 and a weld end position register PR2, which respectively record the weld start position data (x1, y1, z1, w, 0, 0) and the weld end position data (x2, y2, z2, w, 0, 0), and calculate the slope k2 of the weld line. Where, x, y, and z are coordinate values, and w is the rotational angle of the welding torch 31 along the X-axis, and w is generally set to 180°.

[0137] According to the initial laser line slope k1 and the weld line slope k2, adjust the angle between the laser line and the weld line to 90°, calculate the rotational angle w′ of the welding torch 31 along the X-axis, and update the posture of the welding torch 31:

[0138]

[0139] Where, w′ is the rotational angle of the welding torch 31 along the X-axis. w′ greater than 0 indicates clockwise rotation; less than 0 indicates counterclockwise rotation; equal to 0 indicates no rotation. k1 is the initial laser line slope, and k2 is the weld line slope.

[0140] k2 = (y2 - y1) / (x2 - x1), (x1, y1) is the weld start position coordinate, (x2, y2) is the weld end position coordinate. Update the results of the PR1 and PR2 registers according to the above results, which are (x1, y1, z1, w′, 0, 0) and (x2, y2, z2, w′, 0, 0) respectively.

[0141] Step 3: Embedded calculation of the output variables of the neural network model.

[0142] Set a control model input variable register, a control model control coefficient register, and a control model output variable register in the robot controller 12. Input the relevant parameters of the trained control model into the corresponding registers in sequence. Calculate the value of the input variable according to the formula, so as to obtain the value of the welding process parameter.

[0143] The calculation process is: first calculate the values of the hidden layers in sequence, and then calculate the values of the output variables.

[0144] The calculation formula is:

[0145] The calculation formula for the values of the hidden layer is as follows:

[0146]

[0147] The calculation formula for the output layer variables is as follows:

[0148] y_data[l,1] = W2[l,k] × hid_data[k,1] + theta2[l,1],

[0149] where in_data[j,1] is the input layer parameter matrix with an order of j*1; W1[k,j] is the weight coefficient matrix of the input layer neuron nodes with an order of k*j, and theta1[k,1] is the bias coefficient matrix of the input layer neuron nodes with an order of k*1; W2[l,k] is the weight coefficient matrix of the intermediate layer neuron nodes with an order of l*k, and theta2[l,1] is the bias coefficient matrix of the intermediate layer neuron nodes with an order of l*1. The value of j is 1, 2,..., the number of input layer neurons, the value of k is 1, 2,..., the number of hidden layer neurons, and the value of l is 1, 2,..., the number of output layer neurons.

[0150] The input variable register of the control model is used to store the input layer parameters of the neural network model, that is, the control targets, such as: weld width register, weld height register, penetration depth register, and lap joint gap register, etc. The values of the weld width register and the weld depth register are user-set values and remain unchanged during the welding process. The value of the lap joint gap register is a real-time sampled value obtained by the laser sensor 2 and is refreshed at a fixed interval during the welding process.

[0151] The control coefficient register of the control model includes a neuron weight register and a neuron bias register. The number of registers is determined by the number of neurons in the input layer, intermediate layer, and output layer of the trained neural network model.

[0152] The output variable register of the control model is used to store the output layer parameters of the neural network model, that is, the control quantities of the welding process parameters, such as: welding current register, welding speed register, and welding torch swing register, etc.

[0153] The control model described above is an automatic adjustment control model for the welding process parameters of lap joints. The control model adopted in this method is a neural network model. The establishment method is as follows: Design welding process tests according to indicators such as lap gap, weld width, and weld penetration, and establish a database of welding process parameters and weld formation. Determine the input layer parameters and output layer parameters of the neural network model according to the database. Establish and train the neural network model in an industrial control computer with the help of relevant software (typical software, such as Matlab). When the model converges, extract the neuron weights and neuron biases. There are many literatures that have mentioned the specific methods and steps for neural network control model modeling, and this method will not be elaborated here.

[0154] In the preferred embodiment of the present application, for the lap joint of 1.2 mm stainless steel plate, through process tests, the neural network model structure is finally determined to be a three-layer structure with 1 input layer, 1 middle layer, and 1 output layer. Among them, the number of neurons in the input layer is 2 (i.e., j = 2), which are weld width and lap gap respectively; the number of neurons in the middle layer is 12 (i.e., k = 12); the number of neurons in the output layer is 2 (i.e., l = 2), which are welding current and welding speed respectively.

[0155] The register values of each matrix are as follows:

[0156] w1[1,1] = -2.7517452237403019488; w1[1,2] = -3.7755721500145358327;

[0157] w1[2,1] = 3.8974229811484186747; w1[2,2] = -3.1010137808156490458;

[0158] w1[3,1] = 2.2853700769648401803; w1[3,2] = 3.9809291784589118102;

[0159] w1[4,1] = 0.92844177573306829743; w1[4,2] = -5.6049941032870105317;

[0160] w1[5,1] = 3.908883663325845248; w1[5,2] = -2.7923152973196279092;

[0161] w1[6,1] = -4.3983945132935096112; w1[6,2] = -3.3725762660730418396;

[0162] w1[7,1] = 4.4999661455540795885; w1[7,2] = -1.3253468283426776964;

[0163] w1[8,1] = 2.4231818963369664566; w1[8,2] = -5.1941813489611758925;

[0164] w1[9,1] = -0.082908224151521242584; w1[9,2] = 4.8398725884898841088;

[0165] w1[10,1] = -2.7455100415434516847; w1[10,2] = -3.4616144145952216782;

[0166] w1[11,1] = 0.75766757078375490675; w1[11,2] = -4.9629656804772075418;

[0167] w1[12,1] = -3.5380305953421906651; w1[12,2] = -3.3446533594368443865;

[0168] w2[1,1] = -0.42569638508399909593;

[0169] w2[1,2] = -0.39462715069032489046;

[0170] w2[1,3] = -0.20370316897697676728;

[0171] w2[1,4] = -1.5928674376273865843;

[0172] w2[1,5] = -0.4638598882201663276;

[0173] w2[1,6] = -0.35734761489893940301;

[0174] w2[1,7] = -0.32976466840448315221;

[0175] w2[1,8] = 1.7868360951479664855;

[0176] w2[1,9] = 0.43804208150857920634;

[0177] w2[1,10] = 0.73701691758597898119;

[0178] w2[1,11] = 1.307435932297749126;

[0179] w2[1,12] = -0.36275760437338594455;

[0180] w2[2,1] = -0.19952343997177640622;

[0181] w2[2,2] = -0.17732805640299056926;

[0182] w2[2,3] = -0.15711683233437748197;

[0183] w2[2,4] = -0.97368933027893878052;

[0184] w2[2,5] = -0.64664350544370330365;

[0185] w2[2,6] = 0.16794241924301533953;

[0186] w2[2,7] = -0.41014225796567860272;

[0187] w2[2,8] = 1.675511,1169336776022;

[0188] w2[2,9] = -0.62484608057438983941;

[0189] w2[2,10] = 0.39151840574518698146;

[0190] w2[2,11] = 0.97634833228778039604;

[0191] w2[2,12] = 0.064310594843141805077;

[0192] theta1[1,1] = 4.9972035419016567204;

[0193] It should be noted that in the translation of "w2[2,8] = 1.675511,1169336776022;", there seems to be an extra comma in the original text which might be a typo. The translation is presented as accurately as possible based on the provided content.theta1[2,1] = -3.7594781160101700124;

[0194] theta1[3,1] = -2.5565532669951562816;

[0195] theta1[4,1] = 1.9379292989775016576;

[0196] theta1[5,1] = -1.3500407096406381591;

[0197] theta1[6,1] = 0.81599619979576798467;

[0198] theta1[7,1] = 0.82942854055492931664;

[0199] theta1[8,1] = 0.42618553964692096336;

[0200] theta1[9,1] = 2.6038749585140932297;

[0201] theta1[10,1] = -4.1915446553255932827;

[0202] theta1[11,1] = 4.0310664703347827142;

[0203] theta1[12,1] = -4.8059155761906637849;

[0204] theta2[1,1] = -0.35658976360306615661;

[0205] theta2[2,1] = 0.46785524641722658901;

[0206] Step 4: Write the main program for robot movement and welding, the program for gap sampling and calculation of welding process parameters, and perform actual welding.

[0207] The control flow of the welding process is as Figure 4As shown in the figure. A program is written in the robot controller 12. According to a predetermined control process, through the access and operation of each register in the robot controller 12, the real-time adjustment of the welding process parameters during the welding process is achieved. Among them, the welding process parameters are adjusted once every fixed weld length ΔL. The written program is divided into a robot movement and welding main program and a gap sampling and welding process parameter calculation program, and the two programs are parallel threads.

[0208] The system calls the robot movement and welding main program. Initially, the robot 1 is located at the operation origin Po. The robot 1 starts to move. When the robot 1 moves to the sampling starting point P0, the laser sensor 2 is turned on, and the gap sampling and welding process parameter calculation program is started, and the two programs run in parallel. The robot 1 continues to move, continuously reads the position variable of the robot 1, and judges whether the robot 1 has reached the weld starting point position P1. When the robot 1 reaches P1, welding starts. The welding process parameters are obtained from the gap sampling and welding process parameter calculation program. Pi is the welding process parameter change point, on the weld line, the corresponding point at intervals of the fixed weld length ΔL starting from the weld starting point. When the robot 1 moves to Pi, the values of the lap joint gap register and the control model output variable register are updated, that is: the welding process parameters are updated. When the robot 1 reaches the sampling end point P3, the laser sensor 2 is turned off. When the robot 1 reaches the weld end point P2, the arc is extinguished and the welding power supply 3 is turned off.

[0209] Embodiment 2

[0210] A preferred embodiment of the present invention provides a lap joint welding process parameter adjustment system, which includes: a robot 1, a laser sensor 2, a welding power supply 3 and an industrial control computer. Among them, the robot 1 includes a robot controller 12 and a robot body 11, as Figure 6 、 Figure 2 shown.

[0211] The connection relationship of the above components is as follows:

[0212] The robot 1 is the motion execution mechanism of the welding system, responsible for realizing the adjustment of the welding speed; a laser sensor 2 and a welding torch 31 are installed on the sixth axis of the robot body 11;

[0213] The laser sensor 2 is a sensor for lap joint gap sampling, installed on the 6th axis of the robot body 11, and the signal sampled by the laser sensor 2 is sent to the robot controller 12;

[0214] The welding power supply 3 is used to provide the welding current and welding voltage during welding, and the output value of the welding power supply 3 is given by the robot controller 12.

[0215] A method and system for adjusting welding process parameters of a lap joint provided by a preferred embodiment of the present invention. Aiming at the problem that the prior art cannot adjust welding process parameters in real time during welding, resulting in defects such as holes and burn-through in the weld seam when the gap fluctuates, a real-time adjustment system for welding process parameters with the robot controller 12 as the core is established. Based on the robot system architecture language, secondary development of the robot controller 12 is carried out to realize the parallel operation of multiple programs such as robot movement, welding program, gap sampling, and welding process parameter calculation program. Based on the robot system architecture language, secondary programming development of the robot controller 12 is carried out, and multiple programs run in a parallel multi-threaded manner to ensure an ideal welded joint, ensure the consistency of weld formation, and reduce the defects of holes and burn-through. When the laser line is not perpendicular to the weld line, the measured lap joint gap error is large and does not match the actual gap, and the detection accuracy needs to be improved. In the present invention, after calculating the angle between the laser line and the weld line through the perpendicular correction of the laser line and the weld line, the rotation angle of the welding torch 31 is adjusted, so as to ensure that the angle between the laser line and the weld bead direction always remains 90°, ensure that the laser line is perpendicular to the weld seam, and reduce the detection error.

[0216] The preferred specific embodiments of the present invention have been described in detail above. It should be understood that those of ordinary skill in the art can make many modifications and variations based on the concept of the present invention without creative labor. Therefore, all technical solutions that can be obtained by those skilled in the art in the technical field of the present invention through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art shall fall within the protection scope determined by the claims.

Claims

1. A method for adjusting welding process parameters of a lap joint, characterized in that: The method comprises the following steps: S101: determining a characteristic position of the robot during the welding process, and recording the characteristic position; S103: vertical correction of the angle between the laser line output by the laser sensor and the weld line; S105: Building and training a control model, wherein the control model is a neural network model; S107: According to a predetermined control process, the welding process parameters are adjusted in real time during the welding process; in, In step S105, the control model is an automatic adjustment control model for the lap joint welding process parameters, including input parameters, control coefficients and output parameters, wherein the input parameters are control targets, including weld width, weld height, weld penetration and lap gap, the control coefficients include neuron weights and neuron biases, and the output parameters are control quantities of welding process parameters, including welding current, welding speed and welding gun swing amplitude; the relevant parameters of the trained control model are input in sequence, and output parameters are obtained by calculation, and the output parameters include the values of the welding process parameters, specifically: y_data[l,1]=W2[l,k]×hid_data[k,1]+theta2[l,1], Among them, y_data is the output layer parameter matrix, W2[l,k] is the weight coefficient matrix of the intermediate layer neuron node, the order is l*k, hid_data is the hidden layer parameter matrix, Among them, theta2[l,1] is the bias coefficient matrix of the intermediate layer neuron nodes, with an order of l*1, W1[k,j] is the weight coefficient matrix of the input layer neuron nodes, with an order of k*j, in_date is the input layer parameter matrix, with an order of j*1, theta1[k,1] is the bias coefficient matrix of the input layer neuron nodes, with an order of k*1, j is the number of input layer neurons, k is the number of hidden layer neurons, and l is the number of output layer neurons; When building the neural network model, the following steps are included: S1051: Designing a welding process test based on welding indicators, and establishing a database of welding process parameters and weld formation, wherein the welding indicators include lap gap, weld width, and weld penetration; S1052: Determine input layer parameters and output layer parameters of the neural network model according to the database; S1053: Establish a neural network model in the industrial computer and train it. When the model converges, extract the neuron weights and neuron biases.

2. The method according to claim 1, wherein In step S101, the characteristic positions include the sampling starting point, the weld starting point, the welding process parameter change point, the sampling ending point and the weld end point, wherein: The sampling starting point is where the laser line coincides with the welding starting point; The starting point of the weld seam controls the welding machine to start arcing and welding; The welding process parameter change points are set on the weld straight line, starting from the weld starting point, at intervals of fixed weld length, P i (x) = P1(x)*(1+i*ΔL); The sampling end point is where the laser line coincides with the welding end point; The weld end point ends the welding and closes the welding arc; Among them, P i (x) is the position of the welding process parameter change point on the X-axis, P1(x) is the position of the weld starting point on the X-axis, ΔL is the fixed weld length, and i is a variable.

3. The method according to claim 2, wherein The step S103 includes the following sub-steps: S1031: Calibrate the initial laser line slope and the distance between the laser sensor and the welding gun: move the welding gun to the starting point of the weld, turn on the laser sensor to output the laser line, and record the initial laser line slope k1 and the distance d between the laser line and the welding gun; S1032: Calculate the positional relationship between the laser line and the welding gun wire tip, as follows: P laser (x)=P tcp (x)+d Among them, P laser (x) is the position of the laser line on the X axis, P tcp (x) is the position of the welding gun wire tip on the X axis, and d is the distance between the laser line and the welding gun; S1033: Adjust the angle between the laser line and the weld line to 90°, and update the rotation angle of the welding gun along the X-axis.

4. The method according to claim 3, wherein In step S1033, the rotation angle of the welding gun along the X-axis is calculated using the following method: Where w′ is the rotation angle of the welding gun along the X-axis, k1 is the initial laser line slope, k2 is the weld line slope, k2 = (y2-y1) / (x2-x1), (x1, y1) are the coordinates of the weld starting point, and (x2, y2) are the weld end point.

5. The method according to claim 3, wherein In step S107, the control process includes a main process and a calculation process. The main process includes robot movement and welding process, and the calculation process includes gap sampling and welding process parameter calculation. By accessing and calculating each register in the robot controller, real-time adjustment of the welding process parameters is achieved. The real-time adjustment is to adjust the welding process parameters once every fixed weld length.

6. The method according to claim 5, wherein The main process and the calculation process cooperate to implement the lap joint welding process parameter adjustment, specifically including the following steps: S1071: The system calls the main process, and the robot is located at the operation origin; S1072: The robot starts to move. When the robot reaches the sampling starting point, the laser sensor is turned on and the calculation process is started. The main process and the calculation process run in parallel. S1073: The robot continues to move, and when the robot reaches the starting point of the weld, the welding power supply is started to start welding, and the welding process parameters are obtained from the calculation process; S1074: When the robot reaches the welding process parameter change point, update the welding process parameters; S1075: When the robot reaches the sampling end point, turning off the laser sensor; S1076: When the robot reaches the end point of the weld, the arc is extinguished and the welding power supply is turned off.

7. A lap joint welding process parameter adjustment system, characterized in that: The system uses the method according to any one of claims 1 to 6, and the system includes a robot, a laser sensor, a welding power supply and an industrial computer, wherein: The robot, including a robot controller and a robot body, is the motion actuator of the welding system and is responsible for adjusting the welding speed. The laser sensor and welding gun are installed on the robot body. The laser sensor is used to sample the overlap gap and is installed on the robot body. The laser sensor sends a sampled signal to the robot controller and outputs a laser line. The welding power supply provides welding current and welding voltage during welding, and the output value of the welding power supply is controlled by the robot controller; The industrial computer completes the construction and training of the neural network model.

Images