A real-time monitoring method and system for laser arc hybrid welding wire spacing

Abstract

The present invention discloses a real-time monitoring method and system for the spacing between filaments in laser-arc hybrid welding. This method, which belongs to the field of laser-arc hybrid welding technology, addresses the technical problem of significant interference of the burning arc on the collected image during the welding process. The system comprises the following steps: Step 1: During the laser-arc hybrid welding process, a camera is used to collect a frontal image of the molten pool, and the frontal image of the molten pool is labeled to obtain a dataset; Step 2: A deep learning model is constructed and the deep learning model is trained and optimized using the dataset; Step 3: The base operating point of the composite heat source welding torch is determined, the laser-arc hybrid welding system is activated at the base operating point to begin welding, the camera is activated, and the collected frontal image of the molten pool is input into the deep learning model. The present invention introduces a visual sensing system and a deep learning model into the laser-arc hybrid welding process, achieving online prediction of the online monitoring of the spacing between filaments in laser-arc hybrid welding.

Description

Technical Field

[0001] The present invention belongs to the technical field of laser-arc hybrid welding, and in particular relates to a method and system for real-time monitoring of the spacing between light wires in laser-arc hybrid welding. Background Art

[0002] Laser-arc hybrid welding combines two heat sources, laser and MIG arc, into a single molten pool, leveraging their respective strengths while compensating for their respective shortcomings. The synergistic effect of laser and arc significantly improves arc stability and laser absorption efficiency, while also significantly reducing the heat source's requirements for weld assembly precision and weld surface quality. The resulting welds feature high aspect ratios and low stress and deformation, making them widely used in welding medium and thick plates and nonferrous metals, encompassing high-end manufacturing applications such as high-speed trains, ships, and aerospace.

[0003] Interfilament spacing, defined as the distance from the center of contact between the welding wire and the workpiece surface to the center of the laser spot, is a critical process parameter in laser-arc hybrid welding. It determines the quality and performance of the weld joint to a certain extent. During laser-arc hybrid welding, the height between the welded surface and the torch is unstable due to the influence of mechanical system accuracy and variations in substrate thickness, which directly causes variations in interfilament spacing. When the interfilament spacing is too large, the coupling effect of the composite heat source decreases or even disappears, resulting in reduced weld penetration and a weld morphology that is a repetitive superposition of the two heat source morphologies. When the interfilament spacing is too small, the guiding force between the heat sources decreases, causing arc divergence. Especially when the laser irradiates the molten droplet, the direct thermal efficiency of the laser beam on the substrate is greatly reduced, resulting in severe spatter, poor weld capping, uneven penetration, and porosity. Online monitoring and real-time adjustment of interfilament spacing during the welding process are crucial for achieving high-performance weld joints. However, direct acquisition of the relative position of the laser beam to the droplet is difficult due to interference from the arc light. Therefore, it is important to propose a method for real-time monitoring of interfilament spacing based on frontal image data of the molten pool.

[0004] Currently, there are two main methods to solve this technical problem: (1) using an infrared auxiliary measurement system to determine the positions of the laser head and arc welding torch respectively, and then calculating the filament spacing of the composite heat source; (2) using image processing methods to directly extract the filament spacing value from the molten pool image. Chinese invention patent application number 201810716264.5, "A real-time adaptive adjustment system and adjustment method for filament spacing in laser-arc hybrid welding", proposes to use two infrared auxiliary measurement systems to determine the filament spacing. One infrared measurement system monitors the distance between the laser welding torch and the substrate surface, and the other infrared measurement system monitors the distance between the arc welding torch and the laser head. The spatial position data collected by the two systems are then imported into the computer, and the filament spacing is calculated and identified using the position relationship between the laser and arc welding torches. However, this method relies on a complex auxiliary sensing system, the equipment cost is high, and the degree of freedom and convenience of the welding process are subject to certain restrictions. Chinese invention patent application number 201610464040.0, "Real-time Automatic Control Device for Laser-Arc Hybrid Welding Based on Visual Sensing and Welding Method," establishes a visual inspection system that captures images of the wire tip and laser keyhole in real time. The system then uses an image processing algorithm to determine the inter-wire spacing. While this method of obtaining inter-wire spacing based on keyhole and wire image information is simple, the burning arc significantly interferes with the captured image during welding, and the wire length and laser keyhole fluctuate dynamically, leading to significant deviations in the image processing results. Therefore, other low-cost, high-precision methods are needed to achieve online monitoring of inter-wire spacing. Summary of the Invention

[0005] The present invention discloses a real-time monitoring method and system for the spacing between light wires in laser arc hybrid welding, which aims to solve the technical problem that although the method of obtaining the spacing between light wires based on keyhole and welding wire image information is simple in equipment, the burning arc has a significant interference on the collected image during the welding process, and the welding wire length and the laser keyhole fluctuate dynamically, which will lead to a large deviation in the image processing results.

[0006] In order to solve the above-mentioned technical problems, the technical solutions adopted by the present invention are as follows:

[0007] A method for real-time monitoring of the spacing between laser arc hybrid welding wires comprises the following steps:

[0008] Step 1: During the laser-arc hybrid welding process, a camera is used to capture the front image of the molten pool. The front image of the molten pool is labeled using the actual filament spacing value corresponding to the front image of the molten pool to obtain a dataset.

[0009] Step 2: Build a deep learning model, use the dataset to train and optimize the deep learning model, and call the optimized deep learning model online;

[0010] Step 3: Determine the basic working point of the composite heat source welding torch, start the laser-arc composite welding system at the basic working point and start welding. Turn on the camera and capture a frame rate of f. Input the captured front image of the molten pool into the deep learning model to monitor the distance between the laser-arc composite heat source filaments in real time.

[0011] Preferably, the step 1 specifically includes the following steps:

[0012] Step 1.1: Adjust the laser-arc hybrid heat source torch posture so that the angle between the arc torch and the substrate is θ1, the wire stickout is set to d1, the angle between the laser beam and the substrate is θ2, the camera shooting angle is θ3, and the distance between the camera lens and the molten pool is d2. When the wire tip and the beam focus intersect at a point on the substrate, define the wire spacing value D LA is 0, the distance H between the laser-arc composite heat source torch and the substrate is 0, D LA It satisfies the following geometric relationship with H:

[0013] D LA =H×(tan(π / 2-θ1)+tan(π / 2-θ2))

[0014] Step 1.2: Get different D LA The front image of the melt pool under the condition of 0.01% is shown, using the filament spacing value D. LA Label the front images of the melt pool to obtain the dataset.

[0015] Preferably, the angle θ1 in step 1 has a value range of 55°-75°, the value range of d1 is 10-16 mm, the value range of θ2 is 70°-85°, the value range of θ2 is 30°-65°, and the value range of d2 is 130-180 mm.

[0016] Preferably, the data set in step 1 is randomly divided into a training set and a test set according to a ratio of β:(1-β), wherein the value range of β is 0.65-0.85.

[0017] Preferably, the step 2 specifically includes:

[0018] Step 2.1: Use the front image of the melt pool in the dataset as input and the label as output, and iterate to train the deep learning model;

[0019] Step 2.2: Use the root mean square loss function to evaluate the deviation between the true value and the predicted value, use the gradient momentum descent optimizer to update the model weight coefficient, and continuously adjust the hyperparameters of the deep learning model to minimize the loss function value to tune the deep learning model. After the deep learning model is tuned, its accuracy is tested using the test set;

[0020] Step 2.3: Deploy the tuned deep learning model online through system deployment;

[0021] The system deployment is to realize online calling of deep learning model files through the Tensorflow.NET library.

[0022] Preferably, the deep learning model architecture in step 2 comprises a convolutional neural network consisting of 1 input layer, n convolutional layers, n pooling layers, m fully connected layers and 1 linear regression layer, wherein the value range of n is an integer from 3 to 7, and the value range of the parameter m is an integer from 1 to 4.

[0023] Preferably, the determination of the basic operating point of the laser-arc composite heat source in step 3 is specifically as follows:

[0024] Measure the height between the arc welding torch contact tip and the base plate. This is the maximum height H that the composite heat source welding torch can be lowered to. max1 , start welding under the conditions of wire feeding speed V0, welding speed V1, and laser power P, and gradually increase the height H of the composite heat source torch from 0 until the arc can no longer burn. Record the composite heat source torch rising height H at this time max2 , in [-ηH max1 ,ηH max2 ] arbitrarily select the composite heat source torch height value H0 as the basic working point within the interval, where

[0025] Preferably, the camera in step 1 is a CCD camera, and the resolution of the front image of the laser-arc hybrid welding molten pool captured by the camera is 750 pixels × 300 pixels.

[0026] Preferably, in step 3, the value range of parameter f is 10 Hz-50 Hz.

[0027] The range of the angle θ1 between the arc welding torch and the substrate is 55°-75°, and the range of the angle θ2 between the laser beam and the substrate is 70°-85° because it is necessary to ensure that the arc heat source and the laser beam act better in the same molten pool to ensure the synergistic effect of the laser arc heat source; the range of the value of d1 is 10-16mm because it is necessary to maintain the stability of arc combustion and droplet transfer. When the dry extension length is too large, the resistance of the welding wire increases, the actual current decreases, causing the welding wire to break and severe spatter. When the dry extension length is too small, the actual current increases due to the decrease in resistance, causing the conductive nozzle to overheat and be damaged; the range of the camera shooting angle θ3 The angle of 40°-70° is set because it is necessary to ensure that the front image of the molten pool is complete, and the camera installation position needs to avoid the spatial area illuminated by the light beam. The value range of d2 is set to 130-180m because the shooting distance is too close, which will result in an incomplete molten pool, and the shooting distance is too far, which will make it impossible to capture the details of the molten pool. η is set to two-thirds because the arc combustion is stable within this range, and it is easy to select the basic working point. The value range of the CCD camera acquisition frame rate f is set to 10Hz-50Hz because f is too small to meet the real-time prediction requirements, and f is too large, which will lead to a significant increase in the calculation cost of the deep learning model.

[0028] Preferably, the laser-arc hybrid welding system includes: a substrate, an arc power supply, an arc welding torch, a laser welding torch, a computer, a CCD camera, and an ABB robot; wherein the negative pole of the arc power supply is connected to the substrate, the positive pole is connected to the arc welding torch, and the CCD camera is connected to the computer via a USB data interface; the motion actuator is a mobile workbench, the arc welding torch and the laser welding torch are fixed to the ABB robot through a clamp, and the ABB robot controls the movement of the heat source.

[0029] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are:

[0030] In the present invention, during the laser-arc hybrid welding process, a visual sensing system is used to capture the front image of the laser-arc hybrid welding molten pool on the welding path in real time, and a deep learning model directly predicts the wire spacing in the laser-arc hybrid welding process based on the front image of the molten pool. Compared with traditional methods, the present invention introduces the visual sensing system and the deep learning model into the laser-arc hybrid welding process, providing a new idea for solving the problem of online monitoring of wire spacing due to arc interference in the welding process, and realizing online prediction of online monitoring of wire spacing in laser-arc hybrid welding. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 The figure shows the structure diagram of the front molten pool visual sensing system of laser-arc hybrid welding;

[0032] Figure 2The figure shows the workflow of the wire spacing monitoring method in the laser-arc hybrid welding process based on the front image of the molten pool and the deep learning method.

[0033] Figure 3 Shown is the deep learning model architecture and training flow chart;

[0034] Figure 4 Shown are visual images of the laser-arc hybrid welding molten pool under different wire spacing conditions;

[0035] Among them, 1-CCD camera, 2-arc welding torch, 3-laser head, 4-stainless steel substrate, 5-fixture, 6-optical wire spacing real-time prediction host computer. DETAILED DESCRIPTION

[0036] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in combination with the embodiments of the present application and the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all of the embodiments. The components of the embodiments of the present application generally described and indicated in the accompanying drawings here can be arranged and designed in various configurations. Therefore, the following detailed description of the embodiments of the present application provided in the accompanying drawings is not intended to limit the scope of the application for which protection is claimed, but merely represents the selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of this application.

[0037] In the description of the embodiments of the present application, it should be noted that the terms "upper", "lower", "left", "right", "vertical", "horizontal", "inside", "outside", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the accompanying drawings, or are the orientation or positional relationship in which the product of the invention is usually placed when in use. They are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, they cannot be understood as limiting the present application. In addition, the terms "first", "second", "third", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0038] like Figure 1-Figure 4As shown, the laser-arc hybrid welding system of the present invention comprises: a substrate (1), an arc power supply (2), an arc welding torch (3), a laser welding torch (4), a computer (5), a CCD camera (6), and an ABB robot (7); wherein the negative electrode of the arc power supply (2) is connected to the substrate (1), and the positive electrode is connected to the arc welding torch (3); the CCD camera (6) is connected to the computer (5) via a USB data interface; the motion actuator is a mobile workbench, the arc welding torch (3) and the laser welding torch (4) are fixed to the ABB robot (7) via a clamp, and the ABB robot controls the movement of the heat source.

[0039] The specific experimental platform for this example is: the arc welding power source is a Fronius TPS4000 welder. The arc welding torch is positioned at a 65° angle to the substrate, the laser beam at a 90° angle to the substrate, and the CCD camera at a 40° angle to the substrate. The arc welding torch, laser welding torch, and CCD camera are all mounted on an ABB robot fixture, with the ABB robot controlling the composite heat source's position and movement speed. The substrate is a 304 stainless steel plate measuring 300 mm × 60 mm × 5 mm, and the welding wire is 304 stainless steel wire. The process parameters are: wire feed speed of 5 m / min, composite heat source torch travel speed of 20 mm / s, laser power of 3000 W, shielding gas of 99.9% Ar, and gas flow rate of 20 L / min.

[0040] A method for real-time monitoring of the spacing between laser arc hybrid welding wires comprises the following steps:

[0041] Step 1: During the laser-arc hybrid welding process, a camera is used to capture the front image of the molten pool. The front image of the molten pool is labeled using the actual filament spacing value corresponding to the front image of the molten pool to obtain a dataset.

[0042] The camera is a CCD camera, and the resolution of the front image of the laser-arc hybrid welding pool collected by the camera is 750 pixels × 300 pixels;

[0043] The step 1 specifically includes the following steps:

[0044] Step 1.1: Adjust the laser-arc hybrid heat source torch posture so that the angle between the arc torch and the substrate is θ1, the wire stickout is set to d1, the angle between the laser beam and the substrate is θ2, the camera shooting angle is θ3, and the distance between the camera lens and the molten pool is d2. When the wire tip and the beam focus intersect at a point on the substrate, define the wire spacing value D LA is 0, the distance H between the laser-arc composite heat source torch and the substrate is 0, D LA It satisfies the following geometric relationship with H:

[0045] D LA=H×(tan(π / 2-θ1)+tan(π / 2-θ2))

[0046] The angle θ1 is in the range of 55°-75°, the value range of d1 is 10-16 mm, the value range of θ2 is 70°-85°, the value range of θ2 is 30°-65°, and the value range of d2 is 130-180 mm;

[0047] Step 1.2: Get different D LA The front image of the melt pool under the condition of 0.01% is shown, using the filament spacing value D. LA Label the front images of the melt pool to obtain a dataset;

[0048] The dataset is randomly divided into training and test sets according to the ratio of β:(1-β), where the value of β ranges from 0.65 to 0.85.

[0049] Step 2: Build a deep learning model, use the dataset to train and optimize the deep learning model, and call the optimized deep learning model online;

[0050] The step 2 specifically includes:

[0051] Step 2.1: Use the front image of the melt pool in the dataset as input and the label as output, and iterate to train the deep learning model;

[0052] The deep learning model architecture consists of a convolutional neural network with 1 input layer, n convolutional layers, n pooling layers, m fully connected layers, and 1 linear regression layer, where n is an integer between 3 and 7, and m is an integer between 1 and 4.

[0053] Step 2.2: Use the root mean square loss function to evaluate the deviation between the true value and the predicted value, use the gradient momentum descent optimizer to update the model weight coefficient, and continuously adjust the hyperparameters of the deep learning model to minimize the loss function value to tune the deep learning model. After the deep learning model is tuned, its accuracy is tested using the test set;

[0054] Step 2.3: Deploy the tuned deep learning model online through system deployment;

[0055] The system deployment is to realize online calling of deep learning model files through the Tensorflow.NET library.

[0056] Step 3: Determine the base operating point of the hybrid heat source torch, start the laser-arc hybrid welding system at this base operating point, turn on the camera, capture a frame rate of f, and input the captured frontal image of the molten pool into the deep learning model to monitor the distance between the laser-arc hybrid heat source filaments in real time.

[0057] The value range of parameter f is 10Hz-50Hz;

[0058] The basic working point of the laser-arc composite heat source is determined as follows in step 3:

[0059] Measure the height between the arc welding torch contact tip and the base plate. This is the maximum height H that the composite heat source welding torch can be lowered to. max1 , start welding under the conditions of wire feeding speed V0, welding speed V1, and laser power P, and gradually increase the height H of the composite heat source torch from 0 until the arc can no longer burn. Record the composite heat source torch rising height H at this time max2 , in [-ηH max1 ,ηH max2 ] arbitrarily select the composite heat source torch height value H0 as the basic working point within the interval, where

[0060] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for real-time monitoring of the spacing between laser arc hybrid welding wires, characterized in that: The following steps are involved: Step 1: During the laser-arc hybrid welding process, a camera is used to capture the front image of the molten pool. The front image of the molten pool is labeled using the actual filament spacing value corresponding to the front image of the molten pool to obtain a dataset. The step 1 specifically includes the following steps: Step 1.1: Adjust the laser-arc composite heat source torch posture so that the included angle between the arc torch and the substrate is θ 1. Set the wire extension length to d 1. The angle between the laser beam and the substrate is θ 2. The camera's shooting angle is θ 3. The distance between the camera lens and the molten pool is d 2. When the tip of the welding wire and the focus of the beam intersect at one point on the substrate, define the wire spacing value D LA =0, the distance between the laser-arc composite heat source torch and the substrate H is 0, D LA and H The following geometric relationship is satisfied: D LA = H × (tan(π / 2 - θ 1) + tan(π / 2 - θ 2)) Step 1.2: Get Different D LA Front view of the melt pool under the conditions, using the filament spacing value D LA Label the front images of the melt pool to obtain a dataset; Step 2: Build a deep learning model, use the dataset to train and optimize the deep learning model, and call the optimized deep learning model online; The deep learning model architecture in step 2 contains 1 input layer. n convolutional layers, n Pooling layers, m A convolutional neural network with a fully connected layer and a linear regression layer, where n The value range is an integer from 3 to 7. m The value range is an integer from 1 to 4; Step 3: Determine the basic working point of the composite heat source welding torch, start the laser-arc composite welding system at the basic working point to start welding, turn on the camera, and collect frame rates. f ,The collected front image of the molten pool is input into the deep learning model to ,real-time monitor the filament spacing of the laser-arc composite ,heat source.

2. The method for real-time monitoring of the spacing between laser arc hybrid welding wires according to claim 1, characterized in that: The angle described in step 1 θ The value range of 1 is 55°-75°, d The value range of 1 is 10-16mm, θ The value range of 2 is 70°-85°, θ The value range of 2 is 30°-65°, d The value range of 2 is 130-180mm.

3. The method for real-time monitoring of the spacing between laser arc hybrid welding wires according to claim 1, characterized in that: The data set in step 1 is randomly divided into a training set and a test set according to the ratio of β:(1-β), where β The value range is 0.65-0.

85.

4. The method for real-time monitoring of the spacing between laser arc hybrid welding wires according to claim 3, characterized in that: The step 2 specifically includes: Step 2.1: Use the front image of the melt pool in the dataset as input and the label as output, and iterate to train the deep learning model; Step 2.2: Use the root mean square loss function to evaluate the deviation between the true value and the predicted value, use the gradient momentum descent optimizer to update the model weight coefficient, and continuously adjust the hyperparameters of the deep learning model to minimize the loss function value to tune the deep learning model. After the deep learning model is tuned, its accuracy is tested using the test set; Step 2.3: Deploy the tuned deep learning model online through system deployment; The system deployment is to realize online calling of deep learning model files through the Tensorflow.NET library.

5. The method for real-time monitoring of the spacing between laser arc hybrid welding wires according to claim 1, characterized in that: The basic working point of the laser-arc composite heat source is determined as follows in step 3: Measure the height between the arc welding torch contact tip and the base plate. This is the maximum height that the composite heat source welding torch can be lowered. H max1 , when the wire feeding speed is V 0 , welding speed is V 1 , the laser power is P Start welding under the condition of H Start from 0 and gradually rise until the arc can no longer burn, and record the height of the composite heat source torch at this time. H max2 ,exist Randomly select the composite heat source torch height value within the range H 0 is used as the basic working point, where .

6. The method for real-time monitoring of the spacing between laser arc hybrid welding wires according to claim 1, characterized in that: The camera described in step 1 is a CCD camera, and the resolution of the front image of the laser-arc hybrid welding molten pool collected by the camera is 750 pixels × 300 pixels.

7. The method for real-time monitoring of the distance between laser arc hybrid welding wires according to claim 1, characterized in that: In step 3, the parameters f The value range is 10Hz-50Hz.

8. A real-time monitoring system for the spacing between wires in laser arc hybrid welding, applying the method according to any one of claims 1 to 7, characterized in that: The invention comprises a substrate (1), an arc power supply (2), an arc welding torch (3), a laser welding torch (4), a computer (5), a CCD camera (6), and an ABB robot (7); wherein the negative electrode of the arc power supply (2) is connected to the substrate (1), and the positive electrode is connected to the arc welding torch (3); the CCD camera (6) is connected to the computer (5) via a USB data interface; the motion actuator is a mobile workbench, the arc welding torch (3) and the laser welding torch (4) are fixed on the ABB robot (7) via a clamp, and the ABB robot controls the movement of the heat source.

Images