Nine-axis linkage intelligent welding process optimization method based on analogue simulation and digital twinning

Abstract

The invention discloses a nine-axis linkage intelligent welding process optimization method based on analogue simulation and digital twinning. An existing welding process mostly depends on artificial experience and single equipment control, and an integrated processing system of simulation prediction, equipment cooperation, virtual-real linkage and iterative optimization is lacked. According to the nine-axis linkage intelligent welding process optimization method, a platform is built based on simulation software, after a three-dimensional model of a to-be-welded component is built, the three-dimensional model of the to-be-welded component is subjected to multi-physics field coupling simulation processing to obtain welding data, the welding data is input into an AI multi-target optimization algorithm to form a simulation data training model, and the simulation data training model is used for training the to-be-welded component. Screening according to the simulation data training model to obtain a processing process of an optimal welding sequence and process parameters; through deep integration of intelligent equipment integration, technological process simulation optimization and digital twin control, the problems of low efficiency, poor quality and large deformation in a traditional manufacturing mode are systematically solved.

Description

Technical Field

[0001] Specifically, this invention is a nine-axis linkage intelligent welding process optimization method based on simulation and digital twin. Background Technology

[0002] Traditional robotic welding, centered on teach-and-playback, has achieved large-scale application in standardized mass production scenarios. However, due to limitations in technical characteristics and application scenarios, it suffers from multiple bottlenecks, making it difficult to adapt to high-end manufacturing and non-standardized needs. These bottlenecks are mainly manifested in the following aspects: First: Programming teaching is highly dependent on manual operation, and line change and debugging are time-consuming, which not only leads to low equipment utilization, but also the teaching accuracy is directly affected by the operator's experience, and welding deviations are prone to occur when facing complex welds. Second: It lacks intelligent sensing and adaptive adjustment functions, and its flexible adaptation capability is weak when faced with sudden situations such as workpiece assembly errors and material fluctuations. Third: The instructors need to work in the robot's work area, which poses safety hazards such as collisions; Fourth: Traditional models are not equipped with core modules such as vision tracking and sensor perception. They can only mechanically repeat preset trajectories and cannot dynamically correct welding paths and parameters. In addition, each equipment system operates independently and is difficult to integrate seamlessly with the production management platform, resulting in untraceable process data and difficulty in iterative optimization of technology.

[0003] Robotic welding processes also face several technical bottlenecks, primarily: First, the precision of welding equipment coordination is insufficient. In existing welding systems, six-axis welding robots and positioners are mostly controlled independently, lacking a closed-loop coordination mechanism. This makes them prone to posture deviations during welding, requiring frequent manual intervention, which not only causes welding interruptions but also results in poor weld quality consistency.

[0004] Secondly, welding deformation is difficult to control precisely. When the welded components have complex structures and concentrated weld seams, uneven heat input can easily lead to severe thermal deformation. Traditional processes rely on manual experience to select the welding sequence and parameters, making it difficult to accurately control the amount of deformation, often exceeding the industry's allowable error range. Thirdly, process debugging is costly and time-consuming. When changing product models, repeated trial welding and debugging of process parameters are required, which not only leads to a high scrap rate in trial welding and a low pass rate in mass production, but also seriously restricts production efficiency.

[0005] Furthermore, existing technologies lack the integration of AI-powered intelligent decision-making capabilities and digital twin-based visual control functions, making it impossible to monitor, provide data feedback, and dynamically optimize the welding process in real time. This makes it difficult to meet the high-precision welding requirements of complex steel structures. Current welding processes largely rely on manual experience and single-equipment control, and have not yet formed an integrated technical system encompassing simulation prediction, equipment collaboration, virtual-real linkage, and iterative optimization. This fails to effectively address the aforementioned problems and severely restricts the improvement of the quality and efficiency of welded product manufacturing. Summary of the Invention

[0006] This invention provides a nine-axis linkage intelligent welding process optimization method based on simulation and digital twin to solve the above problems.

[0007] A nine-axis linkage intelligent welding process optimization method based on simulation and digital twin is proposed. The nine-axis linkage intelligent welding process optimization method relies on simulation software to build a platform, construct a three-dimensional model of the component to be welded, obtain welding data through multi-physics field coupling simulation processing of the three-dimensional model of the component to be welded, input the welding data into an AI multi-objective optimization algorithm to form a simulation data training model, and select the optimal welding sequence and process parameters based on the simulation data training model.

[0008] As a preferred option, a nine-axis linkage welding system is constructed before building the three-dimensional model of the component to be welded. A six-axis welding robot, a three-axis positioner, and a high-performance pulse welding power supply are integrated to form a nine-axis linkage welding unit. Then, with the cooperation of the developed AI collaborative control module and kinematic algorithm, the nine-axis linkage welding unit is ensured to be in a state of motion synchronization during the welding operation.

[0009] As a preferred option: after selecting the optimal welding sequence and process parameters, a ROS welding trajectory intelligent planning process is performed. The ROS welding trajectory intelligent planning process is based on the ROS inverse kinematics solver, combined with the AI ​​obstacle recognition model and path optimization algorithm to generate a collision-free welding trajectory. The collision-free welding trajectory is then dynamically corrected for workpiece clamping errors through the AI ​​trajectory smoothing optimization algorithm until the weld quality meets the standards and the processing stops.

[0010] As a preferred solution: A digital twin platform is built based on a visualization development platform. A two-way data interaction channel between the virtual and real worlds is established by combining the industrial Ethernet protocol, welding robots and positioners. The optimal welding sequence, process parameters and ROS planning trajectory are selected and imported into the digital twin platform for virtual pre-simulation. During the actual welding process, the virtual and real data are synchronized in real time. The AI ​​algorithm compares the data deviations and iteratively optimizes the process parameters, thereby forming a closed-loop control mechanism.

[0011] As a preferred option, the nine-axis linkage welding unit is configured with a dual-station processing process. The dual-station processing process is a process in which welding operations and workpiece loading and unloading are carried out in parallel. The dual-station processing process, together with the AI ​​collaborative control module, maps the equipment operating status in real time through the digital twin platform and monitors the positioner's posture.

[0012] As a preferred option: the simulation software is Simufact Welding; the AI ​​multi-objective optimization algorithm is an optimization model based on genetic algorithm or particle swarm optimization algorithm. Welding data is input into the AI ​​multi-objective optimization algorithm to form simulation data training model, which determines the optimal heat input by automatically iterating the combination of process parameters.

[0013] As a preferred option, the optimal welding sequence is determined by comparing the simulation deformation differences of the welding sequence of the array, ensuring that the maximum overall deformation of the component in the nine-axis linkage welding unit is reduced from ≥8mm to ≤1.5mm, and the maximum equivalent stress in the weld zone is reduced by ≥11%.

[0014] As a preferred option: the inverse kinematics solver is the TRAC-IK solver, and the path optimization algorithm is... Algorithm; The AI ​​multi-objective optimization algorithm can correct for clamping errors of ±0.05mm, and the weld quality compliance rate of the AI ​​multi-objective optimization algorithm is 95%.

[0015] Compared with existing technologies, this invention provides a nine-axis linkage intelligent welding process optimization method based on simulation and digital twins, which has the following beneficial effects: This invention is a welding process optimization method that combines precision, efficiency, and intelligence. It integrates a technology system of simulation prediction, equipment collaboration, virtual-real linkage, and iterative optimization. Through the deep integration of intelligent equipment integration, process simulation optimization, and digital twin control, it systematically solves the problems of low efficiency, poor quality, and large deformation in traditional manufacturing modes. It can achieve rapid and accurate matching of welding sequence and process parameters, realize automated and precise execution of the welding process, AI-driven accurate prediction of welding deformation, and rapid iteration of process parameters. Attached Figure Description

[0016] Figure 1 This is a flowchart of the present invention; Figure 2 This is a physical image of a nine-axis linkage welding unit; Figure 3 A schematic diagram illustrating the simulation sequence of bridge expansion joint structure and welding. Figure 4 A comparison diagram of maximum stress and deformation for different welding sequence schemes; Figure 5 A comparison chart of the overall deformation before and after debugging; Figure 6 Comparison of maximum stress cloud diagrams before optimization for welding targets; Figure 7 Comparison cloud map of maximum stress after optimization for welding target; Figure 8 This is a schematic diagram of the first typical location for the welding torch accessibility simulation test. Figure 9 This is a schematic diagram of the second typical location for the welding torch accessibility simulation test. Figure 10 This is a schematic diagram of the first three-dimensional structure of the nine-axis linkage welding unit; Figure 11 This is a schematic diagram of the second three-dimensional structure of the nine-axis linkage welding unit; Figure 12 This is a block diagram illustrating the technical route of the present invention.

[0017] In the diagram: 1-Strip connecting table; 2-Welding robot; 3-Divestment seat; 4-Support frame; 5-Rotating platform; 6-Welding target. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Specific implementation method one: Combining Figures 1 to 12 This embodiment describes the nine-axis linkage intelligent welding process optimization method, which relies on simulation software to build a platform, construct a three-dimensional model of the component to be welded, obtain welding data through multi-physics field coupling simulation processing of the three-dimensional model of the component to be welded, input the welding data into an AI multi-objective optimization algorithm to form a simulation data training model, and select the optimal welding sequence and process parameters based on the simulation data training model.

[0020] Specifically, the nine-axis linkage intelligent welding process optimization method in this invention relies on simulation software to build a platform, construct a three-dimensional model of the welding equipment and the component to be welded, obtain welding data through multi-physics field coupling simulation processing of the three-dimensional model of the component to be welded, input the welding data into an AI multi-objective optimization algorithm to form a simulation data training model, and select the optimal welding sequence and process parameters based on the simulation data training model. Specifically, the optimal welding trajectory is planned in ROS based on the obtained data to control the operation of the welding robot arm, and the heat source input is adjusted in real time according to the optimized process parameters. Through data interconnection and communication, the digital twin platform monitors and intervenes in the actual production process.

[0021] This invention utilizes AI-driven decision-making and digital twin-based virtual-real linkage as its main logic. It acquires basic data through multi-physics coupling simulation, and AI algorithms generate optimal process solutions. The ROS system is used to automate and collision-free generate welding trajectories. A high-fidelity digital twin is constructed using Unity3D, synchronizing data between the virtual and real ends in real time. AI algorithms continuously iterate and optimize process parameters, achieving a synergistic improvement in welding quality and efficiency. The simulation software platform enables intelligent synchronization of robot and positioner movements. The digital twin platform maps equipment operating status in real time, remotely monitoring key parameters such as joint angles and positioner posture during welding to ensure the consistency and dimensional accuracy of the weld trajectory. The dual-station parallel design allows for the parallel execution of welding operations and workpiece loading and unloading, significantly reducing equipment downtime.

[0022] Specific Implementation Method Two: This implementation method further defines Specific Implementation Method One. Before constructing the 3D model of the component to be welded, a nine-axis linkage welding system is built. A six-axis welding robot, a three-axis positioner, and a high-performance pulse welding power supply are integrated to form a nine-axis linkage welding unit. Then, through the cooperation of a developed AI collaborative control module and kinematic algorithms, the nine-axis linkage welding unit is ensured to be in a synchronized state during welding operations. Specifically, the six-axis welding robot and the three-axis positioner are integrated into a closed-loop collaborative nine-axis welding system. Based on the AI ​​kinematic collaborative control algorithm, the action timing is dynamically adjusted. The equipment operating status is mapped in real time through a digital twin platform, achieving millisecond-level synchronization between the positioner's rotation and the robot's welding trajectory. This ensures the optimal posture of the robot when welding complex welds and improves the accuracy of welding action coordination.

[0023] The nine-axis linkage welding unit in this embodiment, as shown in the figure, is essentially a nine-axis fully automatic welding system. It includes a strip connecting table 1, a welding robot 2, a steering seat 3, a support frame 4, and two rotating platforms 5. The strip connecting table 1 is horizontally arranged. The welding robot 2 and the steering seat 3 are respectively arranged at both ends of the strip connecting table 1. The support frame 4 is arranged on the steering seat 3. The steering seat 3 is an existing rotating platform that can achieve 360° rotation, driving the support frame 4 to move synchronously. The support frame 4 is a support frame with notches at both ends, which is formed by two U-shaped frames integrally connected. The two notches of the support frame 4 are arranged separately, and a rotating platform 5 is hinged in each notch. The rotating platform 5 can achieve 360° rotation. The welding target 6 is arranged on the rotating platform 5. The welding target 6 is a bridge expansion joint component, ship section, pressure vessel, or other related complex steel structure.

[0024] The welding robot 2 in this embodiment is an existing welding robot, specifically the Canop CRP-RH14-06-W six-axis vertical multi-joint industrial welding robot. Its components are connected in series to form a hierarchical transmission and attitude adjustment system from the base to the end effector. The motion of each axis component is coupled to achieve spatial positioning and trajectory control of high-precision welding operations.

[0025] Welding robot 2 is connected to strip connecting platform 1 via a vertical connecting platform. The vertical connecting platform is a vertical column structure, serving as the load-bearing and fixing foundation for the entire robot. The vertical connecting platform is connected to steering seat 3 via strip connecting platform 1. The upper end of the vertical connecting platform is rigidly connected to the base of welding robot 2. The drive and control module integrated inside the base provides the power source for the movement of each axis. The relevant positions and rotation parameters of the six-axis vertical multi-joint welding robot 2 are as follows: The single-axis component is the waist rotation joint. One end of the waist rotation joint is rigidly connected to the base, and the other end is connected to the two-axis component. It can drive all the components above to rotate horizontally, thereby adjusting the robot's working position in the horizontal plane. The angle of the single-axis component is limited to -170°~170°. The two-axis component, namely the main arm, is hinged to the end of the first axis, enabling pitch and swing in the vertical plane. Its coordinated movement with the first axis can significantly expand the robot's working radius. The angle of the two-axis component is limited to -90° to 150°. The three-axis component, also known as the forearm, has one end connected to the end of the two-axis component and the other end connected to the four-axis component. The rotation angle range of the forearm is -89° to 150°. The four-axis component is the wrist rotation joint, which is a hollow structure. One end of the wrist rotation joint is connected to the end of the forearm, and the other end is connected to the five-axis component. The hollow channel can be used to store the welding torch's user line, air pipe, and other pipelines to avoid the pipelines from getting tangled and affecting the movement. At the same time, the four axes can drive the end component to rotate around the forearm axis, adjusting the circumferential angle of the welding torch. The angle range of the wrist rotation joint is -190° to 190°. The five-axis component, namely the wrist pitch joint, is hinged to the end of the four-axis actuator, which can realize the vertical pitch fine adjustment of the end effector to adapt to welds of different angles. The angle range of the five-axis component is -110° to 130°. The six-axis component, namely the wrist torsion joint, serves as the end joint. One end is connected to the five-axis component, while the other end directly mounts end effectors such as welding torches. It also adopts a hollow design to further optimize pipeline layout. The six-axis component can achieve precise torsion of the end effector. Combined with the movement of the first five axes, it can complete the precise posture matching of complex spatial welds. The angle range of the six-axis component is -190° to 190°. Among them, the first, second, and third axes are the main motion axes of the robot, which mainly realize the positional movement of the end effector in a large space and determine the coverage of the operation; the fourth, fifth, and sixth axes are wrist posture adjustment axes, which are mainly used to adjust the posture and angle of the end welding torch to ensure the precise alignment of the welding torch with the weld. The transmission system of each axis component is uniformly coordinated by the control module in the base to achieve precise synchronization of the movement of each component and meet the precise requirements of welding operation for welding trajectory and posture.

[0026] The workpiece clamping mechanism formed by the cooperation of the support frame 4 and the two rotating platforms 5 is a three-axis positioner, which can replace existing three-axis positioners, or can take the following structural forms: The two rotating platforms 5 are located at two independent workstations, which are also twin workstations. The two workstations share a three-axis motion drive system, which can realize 360-degree rotational attitude adjustment of the workpiece. Each workstation is equipped with an independent workpiece clamping and positioning mechanism, which can respectively carry the workpiece to be welded and complete the welding attitude adjustment.

[0027] The base of the positioner, namely the steering seat 3, is rigidly connected to the strip connecting platform 1, and the integrated drive and control module inside provides a power source for the movement of each axis. The positioner support frame 4 and the steering seat 3 are connected by a rotary joint, which can drive the two rotating platforms 5 on it to rotate horizontally, so as to realize the parallel operation of welding and loading and unloading of workpieces, with an angle limit of ±180°.

[0028] The rotating platform 5 consists of a rotary table and a clamping plate. Each station has a rotary table and a clamping plate. The clamping plate is rigidly connected to the rotary table, and the rotary table is connected to the worktable through a rotary joint. Its angle range is -180° to 180°. The rotating platform 5 can adjust the welding posture of the workpiece to cooperate with the welding robot to complete the operation.

[0029] In this embodiment, the workflow of the nine-axis fully automatic welding system is as follows: The worker manually clamps the workpiece to be welded, starts the welding program, and the positioner's steering seat 3 rotates 180° to deliver the workpiece to the welding station, thus beginning the welding process. Taking a bridge expansion joint as an example: First working stage: The welding station rotary table rotates 15 degrees, and the robot welding gun transitions from the initial position to the weld seam and begins welding according to the optimized welding parameters. The welding voltage is 25V, the welding current is 220A, the arc starting distance is 1.5mm, and the welding speed is controlled at 30cm / min. This stage can be completed in 6 minutes and 40 seconds. After the first working stage is completed, the second working stage is carried out. The second working stage is to restore the welding station rotary table to a horizontal angle, and the robot starts welding according to the optimized welding parameters. The welding voltage is 25V, the welding current is 220A, the arc distance is 1.5mm, and the welding speed is controlled at 30cm / min. This stage can be completed in 11 minutes and 16 seconds. After the second working stage is completed, the third working stage is carried out. In the third working stage, the welding station rotary table is adjusted to -90 degrees, and the robot starts welding according to the optimized welding parameters. The welding voltage is 25V, the welding current is 220A, the arc distance is 1.5mm, and the welding speed is controlled at 30cm / min. This stage can be completed in 3 minutes and 0 seconds. After the third working stage is completed, the fourth working stage is carried out. In the fourth working stage, the welding station rotary table is adjusted to +90 degrees, and the robot starts welding according to the optimized welding parameters. The welding voltage is 25V, the welding current is 220A, the arc distance is 1.5mm, and the welding speed is controlled at 30cm / min. This stage can be completed in 2 minutes and 0 seconds. Finally, after welding is completed, the positioner steering seat 3 rotates 180° to send the workpiece away from the welding station. The worker can then disassemble it and replace it with the next workpiece to be welded, repeating the above work process.

[0030] Specific Implementation Method 3: This implementation method is a further limitation of Specific Implementation Method 1 or 2. In this implementation method, after selecting the optimal welding sequence and process parameters, a ROS welding trajectory intelligent planning process is performed. The ROS welding trajectory intelligent planning process is based on the ROS inverse kinematics solver, combined with the AI ​​obstacle recognition model and path optimization algorithm to generate a collision-free welding trajectory, which is also the welding trajectory with the lowest cost. The collision-free welding trajectory is then dynamically corrected for workpiece clamping errors through the AI ​​trajectory smoothing optimization algorithm until the weld quality meets the standards and the processing stops.

[0031] Specific Implementation Method Four: This implementation method is a further limitation of Specific Implementation Method One, Two, or Three. In this implementation method, a digital twin platform is built based on a visualization development platform. A virtual-real two-way data interaction channel is established by combining industrial Ethernet protocol, welding robot, and positioner. The optimal welding sequence, process parameters, and ROS planning trajectory are selected and imported into the digital twin platform for virtual pre-simulation. During the actual welding process, the data at both ends of the virtual and real sides are processed synchronously in real time. The AI ​​algorithm compares the data deviation and iteratively optimizes the process parameters, thereby forming a closed-loop control mechanism.

[0032] Specifically, in combination Figure 3 and Figure 4As shown, the dynamic correlation between temperature field, stress field, and phase transformation effect is integrated, and the coupled simulation of welding thermal cycle and deformation is realized based on a double ellipsoidal heat source model. Through AI multi-objective optimization algorithm, the welding sequence is optimized from ①→②→③→④→⑤→⑥ to ④→①→③→⑥→②→⑤.

[0033] Combination Figure 5 As shown, by optimizing the welding sequence, the maximum overall deformation of the component was reduced from the initial 8mm to 1.48mm, with the toothed plate having a maximum deformation of 0.22mm, meeting the industry standard of ≤0.3mm. The digital twin platform compares the simulation and measured deformation data in real time, and AI automatically analyzes the reasons for the deviation, providing a basis for subsequent optimization.

[0034] Combination Figure 6 and Figure 7 As shown, in terms of stress uniformity distribution, the maximum equivalent stress in the weld zone decreased from 462.5 MPa to 410.9 MPa. The AI ​​algorithm automatically identified stress concentration areas and visualized the stress distribution through the digital twin platform. Through the AI ​​parameter optimization algorithm, it automatically iterated 100+ sets of parameter combinations to determine the optimal welding parameters as 220A current, 25V voltage, 30cm / min speed, and 4950W heat input, balancing penetration effect and thermal deformation control.

[0035] Specific Implementation Method 5: This implementation method is a further limitation of Specific Implementation Method 1, 2, 3 or 4. In this implementation method, the nine-axis linkage welding unit is configured with a dual-station processing process. The dual-station processing process is a process in which welding operations and workpiece loading and unloading are carried out in parallel. The dual-station processing process, together with the AI ​​collaborative control module, maps the equipment operating status in real time through the digital twin platform and monitors the positioner posture.

[0036] Specific Implementation Method Six: This implementation method is a further limitation of Specific Implementation Method One, Two, Three, Four or Five. In this implementation method, the simulation software is Simufact Welding; the AI ​​multi-objective optimization algorithm is an optimization model based on genetic algorithm or particle swarm optimization algorithm. The welding data is input into the AI ​​multi-objective optimization algorithm to form a simulation data training model, which determines the optimal heat input by automatically iterating the combination of process parameters.

[0037] In this embodiment, a thermo-mechanical-phase change multiphysics coupling simulation technique is used to simulate the temperature field, stress field distribution, and deformation patterns under different welding parameters and welding sequences. An AI multi-objective optimization algorithm is introduced, and the model is trained using a large amount of simulation data to establish a mapping relationship between autonomously learned process parameters such as current, voltage, and welding speed and deformation and stress. For example, the optimal combination of skip welding and sequential welding is used as a process to automatically select the optimal welding sequence and determine the optimal heat input.

[0038] Specific Implementation Method Seven: This implementation method is a further limitation of Specific Implementation Method One, Two, Three, Four, Five or Six. In this implementation method, the optimal welding sequence is determined by comparing the simulation deformation differences of the welding sequence of the array, ensuring that the maximum overall deformation of the component in the nine-axis linkage welding unit is reduced from ≥8mm to ≤1.5mm, and the maximum equivalent stress in the weld zone is reduced by ≥11%.

[0039] Specific Implementation Method Eight: This implementation method is a further limitation of Specific Implementation Methods One, Two, Three, Four, Five, Six, or Seven. In this implementation method, the inverse kinematics solver is the TRAC-IK solver, and the path optimization algorithm is... Algorithm; The AI ​​multi-objective optimization algorithm can correct for clamping errors of ±0.05mm, ensuring that each weld is in the optimal welding posture and avoiding defects such as incomplete fusion and weld offset. The weld quality compliance rate of the AI ​​multi-objective optimization algorithm is 95%.

[0040] This invention enables a high degree of integration between virtual simulation and actual production, as well as data interconnection. It develops an AI-driven visual interface to synchronize welding data and welding process parameters in real time, thereby achieving comprehensive monitoring of the production process. Simultaneously, it supports the distribution of ROS-optimized welding trajectory data to the virtual robot, driving the actual robot to precisely execute the trajectory.

[0041] This invention can solve the problem of limited space and difficult welding in some areas of expansion joints. It customizes an 800mm extended welding gun and optimizes the gun head angle. The specific angle parameters are defined as an elevation angle of 30° and a side deviation angle of 15°, thereby ensuring the precise accessibility of the weld.

[0042] In this embodiment, the welding posture is optimized by automatically calculating joint angles using an AI-enhanced TRAC-IK solver to meet process posture requirements. The posture compliance rate is increased from the traditional 78% to 95%, avoiding defects such as incomplete fusion and weld misalignment caused by manual teaching or angle deviation.

[0043] Through AI obstacle recognition models and The algorithm's cost calculation and collision detection mechanism identify interference risks in the path, generate collision-free paths, and avoid collisions between the robotic arm and the workpiece.

[0044] By replacing B-spline interpolation with an AI trajectory smoothing optimization algorithm, the trajectory is smoothed, reducing path inflection points, minimizing the jerking of the robotic arm's movements, and avoiding defects such as porosity and undercut caused by welding torch vibration. This reduces the incomplete fusion defect rate of welds from 40% to 12%, and increases the first-pass yield from 83% to over 99%, reducing subsequent rework costs.

[0045] In this embodiment, the visualization development platform is Unity3D; the industrial Ethernet protocol is ModbusTCP protocol, and the sampling and transmission frequency can be optimized to ensure the accuracy of virtual and real data mapping, with simulation and measurement errors ≤6%.

[0046] The digital twin platform described in this embodiment has an AI-driven visual interface that can synchronize welding current, voltage, and speed parameters, as well as stress field and temperature field cloud maps in real time, and supports remote monitoring and parameter adjustment.

[0047] Combination Figures 8 to 9 As shown, this embodiment also includes a step of customizing an extended welding torch for welding scenarios in confined spaces: the accessibility of the welding torch is simulated through a digital twin platform, and the welding torch parameters are optimized to a length of 800mm, an elevation angle of 30°, and a side deviation angle of 15° to ensure the feasibility of welding seams in confined spaces.

[0048] In this embodiment, the AI ​​obstacle recognition model is trained through deep learning and can automatically identify protruding structures on the component to be welded, combined with... The algorithm's cost calculation and collision detection mechanism avoid collisions between the robotic arm and the workpiece, reducing the probability of collisions.

[0049] This invention is particularly suitable for welding complex steel structures such as bridge expansion joint steel components, ship sections, and pressure vessels. When changing product types, the process debugging cycle is reduced from 3 days to 2 hours, and the first-pass yield rate of welds is increased from 83% to over 99%.

[0050] After establishing a high-fidelity 3D model of the expansion joint steel component, this invention conducts multiphysics thermo-mechanical coupling simulation processing based on Simufact Welding software to study the application of AI algorithms in welding parameter optimization. This invention trains the model with a large amount of simulation data to achieve intelligent matching of welding current, voltage, and speed; compares the deformation differences between skip welding and sequential welding, and automatically selects the optimal welding sequence by AI; finally, the simulation results are uploaded to a digital twin platform to build a simulation-actual data mapping library, providing data support for the subsequent precise AI control of the welding process.

[0051] In this invention, the closed-loop optimization process for digital twin virtual-real interaction involves building a high-fidelity digital twin platform based on Unity3D, developing a visual interface, establishing a two-way data interaction channel between the virtual and real worlds, and using the Modbus TCP protocol to achieve real-time data synchronization between the actual equipment and the virtual model, mapping key information such as joint angles, positioner posture, and welding parameters. Simultaneously, it supports sending ROS-optimized welding trajectory data to the virtual robot, driving the actual robot to precisely execute the trajectory.

[0052] When changing product models, such as when changing welding products, this invention first requires simulation in a pre-built Unity3D virtual environment to customize welding torches with appropriate lengths and elevation angles for areas difficult to weld. Simultaneously, the changed workpiece model is imported into CAE, and optimal welding parameters and the best welding sequence are determined through multiphysics dynamic simulation. Subsequently, the changed model, after the above adaptation and parameter optimization, is imported into ROS for intelligent welding trajectory planning, and then imported into Unity and the actual robot respectively, thus quickly achieving the goals of product changeover and process adjustment.

[0053] This invention adopts an AI-driven approach across the entire technical route, starting with the construction of a nine-axis system to build the hardware foundation, then proceeding to multiphysics simulation to output process solutions, and finally using ROS trajectory planning to ensure execution accuracy. Ultimately, it achieves virtual-real linkage and closed-loop iteration through a digital twin platform, forming a closed-loop intelligent welding technology system that integrates equipment, simulation, trajectory, and control. This completely breaks through the bottlenecks of traditional processes, such as reliance on experience, high trial-and-error costs, and difficulty in controlling deformation.

[0054] In practical applications, this invention employs a flexible production line layout, enabling mixed-line production of bridge expansion joints in sizes 80, 120, and 160. Bridge expansion joints are expansion devices installed at the ends or connections of bridge sections, primarily located between the ends of two beams, between a beam end and an abutment, or at the hinged joints of the bridge. Their main function is to regulate the displacement and connection between the superstructure caused by vehicle loads and bridge construction materials. Bridge expansion joints are core force-transmitting components of bridges, and their manufacturing precision directly affects the bridge's service life and traffic safety.

[0055] This invention has completed the full-dimensional output of sample results, with complete physical objects, software, and experimental data. At the physical level, a nine-axis intelligent welding system consisting of a six-axis robot and a three-axis positioner has been built. At the software level, a trajectory planning module based on ROS, a Simufact welding simulation model, and a Unity3D digital twin interaction platform have been developed, achieving seamless integration of virtual and physical data. Regarding experimental data, a multi-dimensional dataset has been formed: simulation data covers temperature field, stress field cloud maps, and deformation prediction values ​​for nine welding sequences; measured data, collected through equipment such as laser displacement sensors and stress testing instruments, includes over 500 sets of key indicators such as welding deformation, weld formation parameters, and production efficiency, with a simulation-to-measurement error of ≤6%. The technological achievements are reproducible and scalable. A comparison of core performance test data is shown in Table 1. Table 1: Comparison of Core Performance Test Data

[0056] Sample testing demonstrates that this invention significantly improves both product quality and production efficiency, achieving three core breakthroughs in the intelligent welding production line: First, by using the AI+nine-axis collaborative system, the welding quality compliance rate has increased from 78% to 95%, solving the posture deviation defect; Secondly, by leveraging AI-driven multiphysics simulation and digital twin technology, the trial-and-error time for new product processes has been reduced from 3 days to 2 hours, and the cost has been reduced by 90,000 yuan for each model change. Third, through AI trajectory planning and digital twin control, the welding cycle time for a single piece has been reduced from 59 minutes to 19 minutes, with equipment utilization exceeding 90%. In actual production, the welding deformation of the product is stably controlled within 1.5mm, and the first-pass yield rate of the weld is over 99%, fully verifying the practical value of combining AI and digital twin technologies.

Claims

1. A nine-axis linkage intelligent welding process optimization method based on simulation and digital twin, characterized in that: The nine-axis linkage intelligent welding process optimization method relies on simulation software to build a platform, construct a three-dimensional model of the component to be welded, obtain welding data through multi-physics field coupling simulation processing of the three-dimensional model of the component to be welded, input the welding data into the AI ​​multi-objective optimization algorithm to form a simulation data training model, and select the optimal welding sequence and process parameters based on the simulation data training model.

2. The nine-axis linkage intelligent welding process optimization method based on simulation and digital twin as described in claim 1, characterized in that: Before constructing the 3D model of the component to be welded, a nine-axis linkage welding system is constructed. A six-axis welding robot, a three-axis positioner, and a high-performance pulse welding power supply are integrated to form a nine-axis linkage welding unit. Then, with the cooperation of the developed AI collaborative control module and kinematic algorithm, the nine-axis linkage welding unit is ensured to be in a state of motion synchronization during the welding operation.

3. The nine-axis linkage intelligent welding process optimization method based on simulation and digital twin as described in claim 1, characterized in that: After selecting the optimal welding sequence and process parameters, a ROS welding trajectory intelligent planning process is performed. The ROS welding trajectory intelligent planning process is based on the ROS inverse kinematics solver, combined with the AI ​​obstacle recognition model and path optimization algorithm to generate a collision-free welding trajectory. The collision-free welding trajectory is then dynamically corrected for workpiece clamping errors through the AI ​​trajectory smoothing optimization algorithm until the weld quality meets the standards, at which point the processing stops.

4. A nine-axis linkage intelligent welding process optimization method based on simulation and digital twin as described in claim 1, 2, or 3, characterized in that: A digital twin platform is built based on a visualization development platform. A two-way data interaction channel between the virtual and real worlds is established by combining industrial Ethernet protocol, welding robots and positioners. The optimal welding sequence, process parameters and ROS planning trajectory are selected and imported into the digital twin platform for virtual pre-simulation. During the actual welding process, the virtual and real data are synchronized in real time. AI algorithms compare data deviations and iteratively optimize process parameters, thereby forming a closed-loop control mechanism.

5. The nine-axis linkage intelligent welding process optimization method based on simulation and digital twin as described in claim 2, characterized in that: The nine-axis linkage welding unit is configured with a dual-station processing process, which is a process in which welding operations and workpiece loading and unloading are carried out in parallel. The dual-station processing process, together with the AI ​​collaborative control module, maps the equipment operating status in real time through the digital twin platform and monitors the positioner posture.

6. The nine-axis linkage intelligent welding process optimization method based on simulation and digital twin as described in claim 1, characterized in that: The simulation software is Simufact Welding; the AI ​​multi-objective optimization algorithm is an optimization model based on genetic algorithm or particle swarm optimization algorithm. Welding data is input into the AI ​​multi-objective optimization algorithm to form simulation data training model, which determines the optimal heat input by automatically iterating the combination of process parameters.

7. The nine-axis linkage intelligent welding process optimization method based on simulation and digital twin as described in claim 1, characterized in that: The optimal welding sequence was determined by comparing the differences in simulated deformation between the welding sequences of the array, ensuring that the maximum overall deformation of the component in the nine-axis linkage welding unit was reduced from ≥8mm to ≤1.5mm, and the maximum equivalent stress in the weld zone was reduced by ≥11%.

8. The nine-axis linkage intelligent welding process optimization method based on simulation and digital twin as described in claim 3, characterized in that: The inverse kinematics solver is the TRAC-IK solver, and the path optimization algorithm is... Algorithm; The AI ​​multi-objective optimization algorithm can correct for clamping errors of ±0.05mm, and the weld quality compliance rate of the AI ​​multi-objective optimization algorithm is 95%.

Images