Automatic welding method and system

Abstract

The invention belongs to the technical field of metal processing, and discloses an automatic welding method and system.The method comprises the steps that a functional gradient transition layer is arranged between a steel workpiece and a copper workpiece, components of the transition layer change in a gradient mode from a Ni-rich layer to a Ni-Cu solid solution layer in the thickness direction, and the transition layer is doped with nano reinforced particles or rare earth elements; the metallurgical compatibility is fundamentally improved; in the welding process, a composite energy source and an auxiliary physical field are adopted, and a sensor system is used for monitoring multi-modal data in real time; monitoring data are input into a predictive control model based on artificial intelligence, the model can predict the defect trend of a future moment, and welding parameters are actively adjusted in a feedforward mode. According to the method, the problems that in a traditional method, intermetallic compounds are difficult to inhibit, and process control lags are solved, and a welding joint which is free of cracks and pores, extremely thin in IMC layer and excellent in mechanical property can be obtained.

Description

Technical Field

[0001] This invention belongs to the field of metal processing technology, and specifically relates to an automated welding method and system. Background Technology

[0002] Welding copper and steel has broad application prospects in fields such as power, electronics, nuclear energy, and chemical industry. However, due to the significant differences in physical properties such as melting point, thermal conductivity, and coefficient of thermal expansion between copper and steel, and their near-immiscibility in metallurgy, brittle Fe-Cu intermetallic compounds (IMCs) are easily formed in the weld, leading to fatal defects such as joint cracks, low strength, and unstable performance.

[0003] Existing technologies mainly attempt to solve this problem through the following methods: 1. Using a transition layer: One or more layers of metals (such as Ni) with good compatibility with both copper and steel are pre-placed between them as a transition. However, a simple single-component transition layer can only play a limited buffering role. Under intense welding thermal cycling, Fe and Cu elements will still penetrate the Ni layer and diffuse into each other to form brittle phases, so the effect is not ideal.

[0004] 2. Optimize welding processes: Advanced processes such as laser oscillation welding and cold metal transfer (CMT) arc welding are adopted to improve weld formation and microstructure by stirring the molten pool and reducing heat input. However, these methods mainly improve the weld at the physical level and have limited effectiveness in suppressing the fundamental metallurgical problem of IMC formation.

[0005] 3. Real-time feedback control: This method monitors a single parameter (such as temperature) during the welding process using sensors, and passively adjusts the control when the parameter exceeds a threshold. This control method suffers from significant lag, often only reacting after a defect has already formed or begun to appear, and thus cannot fundamentally prevent defects.

[0006] In summary, existing technologies lack a systematic solution to the copper-steel welding problem that can work synergistically from both the metallurgical essence and the intelligent pre-control of the process.

[0007] The above description of the background technology is only for the purpose of facilitating a deeper understanding of the technical solution of the present invention (the technical means used, the technical problems solved, and the technical effects produced, etc.), and should not be regarded as an admission or in any form an implication that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0008] The present invention aims to at least partially solve the aforementioned technical problems. Therefore, the present invention aims to provide an automated welding method and system that improves the quality of copper-steel welding by organically combining a specially designed functionally graded doped transition layer with an artificial intelligence-based predictive control strategy.

[0009] The technical solution adopted in this invention is as follows: An automated welding method, comprising: a) Provide a first metal workpiece and a second metal workpiece to be welded, wherein the first metal workpiece is steel or a steel alloy and the second metal workpiece is copper or a copper alloy.

[0010] b) Design and fabricate a functionally graded doped transition layer. This step is the first key innovation of this invention. Unlike a simple single transition layer, this invention designs a functionally graded layer. For example, a transition layer with a thickness of 50-300 μm is fabricated on the steel surface to be welded using techniques such as multi-target magnetron sputtering, laser cladding, or cold spraying. The composition of this layer is not uniform but is carefully designed along the thickness direction.

[0011] Gradient Design: The transition layer structure consists of a pure nickel (Ni) or nickel-rich (Ni) layer on the side immediately adjacent to the steel, utilizing the infinite solid solution relationship between nickel (Ni) and iron (Fe) to ensure a strong metallurgical bond. Then, the composition gradually transitions to a nickel-vanadium (Ni-V) or nickel-chromium (Ni-Cr) alloy. The introduction of V or Cr elements effectively hinders the diffusion of Fe to the copper side and refines the grain size. Finally, near the copper side, the composition gradually transitions to a nickel-copper (Ni-Cu) solid solution, ensuring good wetting and metallurgical bonding with the subsequently molten copper. This smooth gradient effectively mitigates abrupt changes in thermal expansion coefficients and mechanical properties, fundamentally reducing interfacial stress.

[0012] Doping reinforcement: In specific regions of the gradient layer (such as the intermediate Ni-V layer), trace amounts (0.1-2% wt.) of nano-reinforcing particles (such as titanium carbide (TiC) and tungsten carbide (WC)) or rare earth elements (such as cerium (Ce) and yttrium (Y)) are doped. The nanoparticles further enhance the strength and high-temperature stability of the transition layer through dispersion reinforcement and grain boundary pinning. Rare earth elements can purify the molten pool, change the morphology of the IMC (from coarse continuous sheets to fine dispersed dots), and significantly reduce its brittleness.

[0013] c) Assemble and fix the workpiece with the special transition layer.

[0014] d) Use a composite energy source for welding. For example, use a laser as the main heat source to provide deep penetration, while simultaneously coupling a CMT arc to stabilize the molten pool and increase filler. Alternatively, use an infrared-blue dual-wavelength laser, where the blue laser preferentially overcomes the high reflectivity of copper, effectively preheating and melting it.

[0015] e) Applying an auxiliary physical field. During welding, an alternating electromagnetic field is applied through an external electromagnetic coil. The Lorentz force generates a strong stirring effect inside the molten pool, forcing the homogenization of element distribution, breaking up dendrites at the solidification front, and further suppressing macroscopic segregation.

[0016] f) Implement AI predictive closed-loop control. This step is the second key innovation of this invention.

[0017] Multimodal data acquisition: A powerful sensor system is integrated at the coaxial or off-axis position of the welding torch to simultaneously acquire multiple information from the welding area: a high-speed camera captures the plasma / molten pool morphology, an infrared thermal imager acquires the temperature field distribution, a spectrometer analyzes the plasma composition (which can indirectly reflect element burn-off and changes in molten pool composition), an acoustic sensor monitors the acoustic emission signal of the keyhole, and an optical coherence tomography (OCT) sensor can even measure the weld depth and keyhole profile in real time.

[0018] AI model prediction: These high-dimensional, temporal multimodal data streams are fed in real time into an AI model (such as a CNN-LSTM hybrid model) pre-trained with a large amount of experimental and simulation data. CNN is responsible for extracting spatial features from data such as images and spectra, while LSTM is responsible for processing the temporal relationships of these features. The core function of this model is not judgment, but prediction. It can predict the probability of defects such as porosity, lack of fusion, and keyhole instability occurring within the next 10-50 milliseconds based on the data pattern in the current millisecond.

[0019] Feedforward proactive intervention: Once the model predicts that the probability of a defect occurring exceeds a preset safety threshold (e.g., predicting that the keyhole will become unstable in 20 milliseconds), the controller immediately generates a feedforward control signal and proactively adjusts the welding parameters before the defect occurs. For example, it instantaneously reduces the laser power by 5% while increasing the laser oscillation frequency by 10Hz to stabilize the keyhole. This proactive intervention completely overcomes the hysteresis of traditional feedback control, elevating process control to a new level of intelligence.

[0020] A system for performing the aforementioned automated welding method includes a workpiece clamping platform, a composite energy source, an auxiliary physical field generator, a sensor system, and a controller. The controller is electrically connected to the composite energy source, the auxiliary physical field generator, and the sensor system. The controller internally stores and runs an artificial intelligence-based predictive control model. The controller is configured to: receive real-time multimodal data from the sensor system; input the data into the predictive control model to predict the state evolution trend of the weld pool and the probability of potential defects; and, based on the prediction results, generate control commands to proactively adjust the operating parameters of the composite energy source and / or the auxiliary physical field generator in a feedforward manner.

[0021] Preferably, the system is also used to weld a workpiece assembly comprising a first metal workpiece and a second metal workpiece, and a functional gradient transition layer disposed at the interface between the two, the transition layer being doped with nano-reinforcing particles and / or rare earth elements.

[0022] The beneficial effects of this invention are as follows: 1. Synergistic effect to eliminate defects: This invention combines static metallurgical design (functionally graded doped transition layer) with dynamic intelligent process control (AI prediction). The transition layer inherently suppresses the formation conditions of brittle IMC, while AI predictive control eliminates the formation window of dynamic defects in the physical process. The two work synergistically to achieve a 1+1>>2 effect.

[0023] 2. Extremely high welding quality: The copper-steel joint obtained using the method of this invention has a beautiful weld formation, free from macroscopic cracks and porosity. Metallographic analysis reveals that the IMC layer at the interface is strictly controlled to below 2μm, exhibiting a fine and dispersed morphology.

[0024] 3. Excellent mechanical properties: The tensile strength of the joint can stably reach 240-250MPa, which is more than 90% of the strength of the copper base material (T2 copper, about 270MPa), far exceeding the joint strength obtained by traditional methods (usually less than 160MPa).

[0025] 4. Highly automated and stable process: AI predictive control can autonomously respond to various disturbances in the welding process (such as changes in workpiece gap and fluctuations in surface condition), ensuring high consistency and repeatability of welding quality, and greatly reducing reliance on operator skills. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments. Obviously, the described embodiments are only some embodiments of this invention, not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this invention.

[0027] The present invention provides an automated welding system comprising: a workpiece clamping platform, a composite energy source (e.g., composed of a laser and an arc welding machine), an auxiliary physical field generator (e.g., an electromagnetic coil), a sensor system (e.g., integrating a high-speed camera, an infrared thermal imager, and a spectrometer), and a core controller. The controller incorporates an AI-based predictive control model.

[0028] Example 1 This embodiment aims to weld a 10mm thick Q235 steel plate to a 10mm thick T2 copper plate.

[0029] 1. Fabrication of functionally graded transition layers: A functionally graded transition layer with a total thickness of 200 μm was prepared on the weld bevel surface of Q235 steel plate using a multi-target magnetron sputtering system. The structure of this transition layer, from the steel substrate outwards, is as follows: First layer (50μm thick): Using a pure Ni target, a Ni-rich layer is deposited on the steel surface to ensure excellent bonding with the steel substrate.

[0030] The second layer (100 μm thick): Both the Ni and V targets are simultaneously activated, and the power of both targets is controlled by a computer program to smoothly increase the V content in the deposited layer from 0% to 15%, forming a Ni-V gradient layer. During this stage, a third target containing TiC powder is simultaneously activated for low-power co-sputtering, uniformly doping the Ni-V layer with approximately 0.8% wt. of nano-TiC particles.

[0031] The third layer (50μm thick): The V target is gradually turned off while the Cu target is turned on. The power of the target material is controlled to smoothly increase the Cu content in the deposited layer from 0% to 80%, forming a Ni-Cu solid solution gradient layer to ensure excellent bonding with the subsequently molten copper substrate.

[0032] 2. Welding and AI Control: The steel plate with the prepared transition layer is butted and fixed to the copper plate with a gap of less than 0.1 mm.

[0033] Laser-CMT arc hybrid welding was employed. The laser power was set to the baseline value of 2.8kW, with the laser beam oscillating in an "infinite" shape (frequency 80Hz, amplitude 1.2mm). The CMT arc welding machine current was set to the baseline value of 130A. An alternating electromagnetic field of 50Hz and 30mT was applied to the electromagnetic coil.

[0034] During the welding process, the sensor system collects plasma morphology, temperature field distribution, and Fe / Cu / Ni elemental spectral intensity data and sends them to the controller at a rate of 1000 frames per second.

[0035] The controller runs a CNN-LSTM model pre-trained on a high-performance computing server. By learning from more than 20,000 sets of "(sensor data) - (defect label)" samples, the model is able to predict the probability of keyhole collapse within the next 30ms based on the input real-time data.

[0036] In a welding experiment, at t=5.31s, the model predicted that the keyhole collapse probability would reach 95% after 25ms. The controller immediately made a feedforward adjustment: at t=5.315s, the laser power was instantly reduced to 2.6kW, and the electromagnetic field frequency was increased to 60Hz. Observation via a high-speed camera showed that around t=5.34s, the molten pool experienced a violent fluctuation but was quickly stabilized, preventing the formation of a collapse defect.

[0037] Example 2 This embodiment aims to weld stainless steel (316L) and oxygen-free copper (TU1).

[0038] The transition layer was prepared using laser cladding technology, with the following gradient: 316L substrate -> Ni60 alloy layer -> Ni60 + 0.2% wt. rare earth Ce mixed layer -> Cu-Ni alloy layer.

[0039] The energy source uses a composite heat source of infrared laser (1070nm) + blue laser (450nm). The blue laser has a power of 2kW and its spot size leads the infrared laser spot size by 2mm, used for preheating and melting highly reflective copper. The infrared laser has a power of 3.5kW and is responsible for generating deep penetration welds.

[0040] The AI ​​model employs a GRU (Gated Recurrent Unit) network, specifically designed for time-series prediction of spectral and temperature field data, primarily used to predict excessive element burn-off and porosity defects.

[0041] To highlight the inventiveness and beneficial effects of this invention, the following comparative examples are provided: Comparative Example 1: A conventional method was used. A 200μm thick pure Ni transition layer (non-gradient, undoped) was pre-placed between Q235 steel and T2 copper. The welding process parameters were the same as in Example 1, but no electromagnetic field was applied, and the control method was simple PID feedback control (based solely on infrared thermometry, the laser power was linearly reduced when the maximum temperature exceeded 1450°C).

[0042] Comparative Example 2: Partially employs the technology of this invention. It uses the exact same functionally graded doped transition layer as Example 1, but the control method remains the simple PID feedback control from Comparative Example 1.

[0043] Experimental Data and Results Analysis The welded joints obtained in Example 1, Comparative Example 1, and Comparative Example 2 were sampled and tested. The results are shown in the table below:

[0044] Data Analysis: 1. Strength comparison: The tensile strength of Example 1 (248MPa) is much higher than that of Comparative Example 1 (145MPa) and Comparative Example 2 (215MPa), reaching 91.8% of the strength of T2 copper base material, showing a huge performance advantage.

[0045] 2. Synergistic Effect of Transition Layer and AI: Comparing Comparative Example 2 and Example 1, both used the same advanced transition layer. However, Example 1, due to the adoption of AI predictive control, still achieved a 15.3% increase in strength, and the IMC layer thickness was further reduced from 4.5 μm to 1.8 μm, resulting in a significant improvement in yield. This strongly demonstrates that AI predictive control is not simply an added bonus, but rather forms an indispensable synergistic effect with the advanced transition layer, which is key to obtaining the highest quality joint.

[0046] 3. Microstructure Comparison: Microstructure analysis of the weld interface using scanning electron microscopy (SEM) revealed that the interface of Comparative Example 1 exhibited a thick, continuous network or lamellar IMC layer, which was the initiation point of cracks. In contrast, the interface of Example 1 was clean and clear, with an extremely thin IMC layer that was dispersed as discontinuous fine particles at the interface. This was attributed to the blocking effect of the functionally graded layer and the refining modification effect of rare earth / nanoparticles.

[0047] In summary, this invention, through its unique technical solution comprised of two core innovations—a functionally graded doped transition layer and AI predictive control—successfully addresses a long-standing pain point in the field of copper-steel dissimilar metal welding. The resulting technical effects are unpredictable by simple combinations of existing technologies, demonstrating outstanding substantive features and significant progress.

[0048] This invention is not limited to the optional embodiments described above, and anyone can derive other various forms of products based on the inspiration of this invention. The specific embodiments described above should not be construed as limiting the scope of protection of this invention; the scope of protection of this invention should be determined by the claims, and the specification can be used to interpret the claims.

Claims

1. An automated welding method, characterized in that... Includes the following steps: a) Provide a first metal workpiece and a second metal workpiece to be welded, wherein the first metal workpiece is steel or a steel alloy and the second metal workpiece is copper or a copper alloy; b) On the welding joint surface of the first metal workpiece and the second metal workpiece, a functional gradient transition layer is prepared and disposed, wherein the functional gradient transition layer is doped with at least one selected from nano-reinforcing particles or rare earth elements. c) Align and fix the first and second metal workpieces with the functional gradient transition layer; d) Using a composite energy source, the workpiece is welded from the side of the second metal workpiece or the center of the joint; e) During the welding process, an auxiliary physical field is applied to the welding area to stir the molten pool; f) During the welding process, multimodal data of the welding area are monitored in real time using a sensor system; g) The multimodal data is input in real time into a pre-trained, artificial intelligence-based predictive control model, which predicts the state evolution trend of the weld pool and the probability of potential defects. h) Based on the prediction results output by the model, the parameters of the composite energy source and / or the parameters of the auxiliary physical field are actively adjusted in a feedforward manner to suppress the formation of potential defects and achieve closed-loop predictive control of the welding process.

2. The method according to claim 1, characterized in that, The composition of the functional gradient transition layer varies in a gradient along its thickness direction to smoothly transition the differences in physical and chemical properties between the first metal workpiece and the second metal workpiece.

3. The method according to claim 1, characterized in that, The composition gradient of the functional gradient transition layer is set as follows: the side closest to the first metal workpiece is a nickel-rich layer, and towards the second metal workpiece, the composition gradually transitions to a nickel-vanadium alloy layer or a nickel-chromium alloy layer, and finally transitions to a nickel-copper solid solution layer.

4. The method according to any one of claims 1-3, characterized in that, The nano-reinforcing particles are one or more nanoscale particles selected from titanium carbide, tungsten carbide, silicon carbide, and titanium nitride, with a doping mass percentage of 0.1% to 2%.

5. The method according to any one of claims 1-3, characterized in that, The rare earth element is selected from one or more of cerium, yttrium, and lanthanum, and its doping mass percentage is 0.05% to 0.5%.

6. The method according to claim 1, characterized in that, The predictive control model based on artificial intelligence is a model composed of convolutional neural networks, recurrent neural networks, long short-term memory networks, or combinations thereof.

7. The method according to claim 1, characterized in that, The sensor system includes a combination of at least two sensors selected from the following: high-speed camera, infrared thermal imager, spectral analyzer, acoustic sensor, and optical coherence tomography sensor.

8. The method according to claim 1, characterized in that, The composite energy source is a laser-arc composite heat source or a dual-wavelength laser composite heat source.

9. A system for performing the automated welding method according to any one of claims 1-8, characterized in that, include: A workpiece clamping platform; A composite energy source; An auxiliary physics field generator; A sensor system; A controller; The controller is electrically connected to the composite energy source, the auxiliary physical field generator, and the sensor system. The controller internally stores and runs an artificial intelligence-based predictive control model. The controller is configured to: Receive real-time multimodal data from the sensor system; The data is input into the predictive control model to predict the state evolution trend of the weld pool and the probability of potential defects; Based on the prediction results, control commands are generated to proactively adjust the operating parameters of the composite energy source and / or auxiliary physical field generator in a feedforward manner.

10. The system according to claim 9, characterized in that, The system is also used to weld a workpiece assembly comprising a first metal workpiece and a second metal workpiece, and a functional gradient transition layer disposed at the interface between the two, the transition layer being doped with nano-reinforcing particles and / or rare earth elements.

Images