Multi-welding robot collaborative operation system based on digital twinning

Abstract

The invention discloses a multi-welding-robot collaborative operation system based on digital twinning, and belongs to the technical field of intelligent manufacturing and robot control, the system comprises a physical space module, a virtual space module, a twinning data layer and a control module, and communication connection is established among the physical space module, the virtual space module, the twinning data layer and the control module; the physical space module comprises a plurality of welding robots, a sensor group and a data transmission network; the virtual space module comprises a multi-welding robot twinning body and a workpiece twinning model; the twin data layer is used for connecting the physical space module and the virtual space module; and the control module generates a welding path optimization scheme, a multi-welding-robot collaborative collision avoidance strategy and a welding quality feedback control instruction based on the twin data layer. According to the method, a closed-loop control framework of'physical-virtual 'bidirectional mapping is constructed, so that dynamic collaborative optimization of multi-robot motion trails, welding seam forming and thermal deformation is realized.

Description

Technical Field

[0001] The present invention relates to the field of intelligent manufacturing and robot control technology, and in particular to a multi-welding robot collaborative operation system based on digital twins. Background Art

[0002] With the rapid growth in demand for large, complex components in aerospace, shipbuilding, rail transit, and other fields, multi-robot collaborative welding technology has become a key means of improving production efficiency and quality. Multi-robot machining systems are characterized by complex structural components and operating mechanisms, multi-source heterogeneous process data, and difficulties in collaborative optimization. Implementing interference avoidance, trajectory planning, heterogeneous data fusion, process optimization, and intelligent collaboration in multi-robot machining systems is crucial to ensuring efficient, high-quality, safe, and reliable operation.

[0003] After searching, it was found that the Chinese patent publication number is: CN108544495A discloses a welding path planning method, system and equipment for multiple welding robots, including obtaining all welding trajectory points of the workpiece to be welded and the number of welding tasks of each welding robot, labeling all welding trajectory points, and the label of each welding trajectory point is different; generating multiple welding paths according to all labels and the number of welding tasks of each welding robot; judging whether the welding path meets the constraint conditions; if so, using the welding path as the execution welding path of multiple welding robots; for example, the Chinese patent publication number is: CN112453648B discloses an offline programming laser weld tracking system based on 3D vision, and its specific operation process is that a 3D camera is installed above the welding workpiece, and after the workpiece is installed, it is remotely triggered to take pictures to obtain point cloud data of the welding site such as the workpiece and the fixture; the above scheme integrates the three technologies of 3D vision model reconstruction, offline programming and laser weld tracking to realize unmanned welding site.

[0004] While the aforementioned patent improves the efficiency and flexibility of multiple welding robots during welding operations, reduces the complexity of on-site welding procedure instruction, and improves welding efficiency, it lacks the ability to integrate and analyze dynamic data from the welding process, making it difficult to address path deviations caused by thermal deformation. While it can achieve local path correction, it fails to establish a closed-loop control framework for virtual-reality collaboration. Therefore, a multi-welding robot collaborative operation system based on digital twins is urgently needed to address these issues. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to overcome the shortcomings of the above-mentioned prior art and provide a multi-welding robot collaborative operation system based on digital twins.

[0006] In order to solve the above technical problems, a technical solution adopted by the present invention is: providing a multi-welding robot collaborative operation system based on digital twin, including: a physical space module, a virtual space module, a twin data layer, and a control module;

[0007] A physical space module, comprising multiple welding robots, a sensor group, and a data transmission network. The physical space module performs welding tasks through the multiple welding robots, collects dynamic data during the welding process in real time through the sensor group, and transmits the dynamic data through the data transmission network.

[0008] A virtual space module, comprising multiple welding robot twins and a workpiece twin model, wherein the virtual space module is completely mapped from the physical space module; the virtual space module simulates the welding process in real time through the multiple welding robot twins and the workpiece twin model;

[0009] a twin data layer, which is used to connect the physical space module and the virtual space module, fuse and synchronize the dynamic data with the motion trajectory, welding formation quality, and thermal deformation state data of the multiple welding robot twins and the workpiece twin models, and predict the motion trajectory, weld formation quality, and thermal deformation state of the multiple welding robots through a preset prediction unit;

[0010] A control module generates control instructions based on the twin data layer and sends the control instructions to the multi-welding robot for execution.

[0011] The present invention is further configured such that: communication connections are established between the physical space module, the virtual space module, the twin data layer, and the control module;

[0012] The sensor group in the physical space module includes a visual sensor, a force sensor, a temperature sensor, and a current and voltage sensor;

[0013] The dynamic data in the physical space module includes welding parameter data, robot posture data, weld morphology data and ambient temperature data.

[0014] The present invention is further configured as follows: the dynamic data collection method includes:

[0015] S1. Scanning the area to be welded of the workpiece by the visual sensor to obtain a first visual image of the weld position, shape, and size, and scanning the welding posture of the multi-welding robot by the visual sensor to generate a second visual image, and performing denoising and feature enhancement on the first and second visual images by using an image processing algorithm to generate weld morphology data and robot posture data;

[0016] S2. Collecting temperature distribution data of the welding area of ​​the multi-welding robot through the temperature sensor, recording the trend of heat input change, and generating ambient temperature data;

[0017] S3. Continuously collect the current value, voltage value and wire feeding speed data of the preset welding power supply through the current and voltage sensors. At the same time, monitor the force changes when the welding gun of the multi-welding robot contacts the workpiece in real time through the force sensor, judge the collision risk during the welding process, and record it in real time to generate welding parameter data. The welding parameter data, the robot posture data, the weld morphology data and the ambient temperature data are packaged to generate dynamic data, and the dynamic data is uploaded to the twin data layer and the virtual space module through the data transmission network.

[0018] The present invention is further configured as follows: the specific steps of completely mapping the virtual space module from the physical space module are:

[0019] M1, the virtual space module receives the dynamic data, and extracts the mechanical structure parameters, motion joint parameters and welding process parameters of the multi-welding robot based on the dynamic data, and generates model data of the multi-welding robot twin corresponding to the multi-welding robot;

[0020] M2. Based on the model data of the multi-welding robot twins, a three-dimensional geometric model and a motion model of the multi-welding robot twins are established, and combined with the weld morphology data, a workpiece twin model consistent with the workpiece is constructed, and according to the real-time dynamic data in the physical space module, the motion trajectory and welding process parameters of the multi-welding robot twins are continuously corrected, and a virtual simulation unit is preset to verify the corrected motion trajectory and welding process parameters of the multi-welding robot twins and their consistency with the real-time dynamic data in the physical space module. After successful verification, a complete virtual space module is generated.

[0021] The present invention is further configured as follows: in the step M2, a virtual simulation unit is preset to verify the consistency of the corrected motion trajectory and welding process parameters of the multi-welding robot twins with the real-time dynamic data in the physical space module (1), and the specific steps are as follows:

[0022] M21. Based on the corrected motion trajectory and welding process parameters of the multi-welding robot twins, the virtual simulation unit drives the multi-welding robot twins and the workpiece twin model to simulate the welding process and generates virtual welding parameters, virtual weld formation quality data, and virtual thermal deformation state data;

[0023] M22. Comparing the virtual welding parameters with the actual motion trajectory data of the multiple welding robots in the physical space module in real time, and performing a difference analysis between the virtual weld formation quality and the virtual thermal deformation state data and the weld morphology data and the ambient temperature data collected by the physical space module;

[0024] M23. If the difference analysis result exceeds the preset threshold, it is determined that the status of the virtual space module is inconsistent with that of the physical space module, and the virtual space module is triggered to re-correct the motion trajectory and welding parameters of the multi-welding robot twins; if the difference is within the threshold, it is confirmed that the consistency verification has passed, and the currently corrected virtual space module is locked as a valid mapping.

[0025] The present invention is further configured to fuse and synchronize the dynamic data with the motion trajectory, welding forming quality, and thermal deformation state data of the multi-welding robot twins and the workpiece twin models in the twin data layer, and the specific method includes:

[0026] Q1. The twin data layer receives the dynamic data transmitted by the physical space module, and parses the robot posture data, welding parameter data, weld morphology data and ambient temperature data in the dynamic data through a preset parsing unit;

[0027] Q2. Match the analyzed robot pose data with the motion joint parameters of the multi-welding robot twins one by one, associate the welding parameter data with the virtual welding process parameters of the welding guns of the multi-welding robot twins, and map the weld topography data and ambient temperature data to corresponding areas of the workpiece twin model;

[0028] Q3. Based on the real-time parsed dynamic data, the twin data layer drives the multi-welding robot twin to execute the same motion trajectory as the multi-welding robot in the virtual space, and updates the thermal deformation state and weld formation quality of the workpiece twin model to complete data fusion and synchronization.

[0029] The present invention is further configured as follows: the specific steps of predicting the motion trajectory, weld formation quality and thermal deformation state of the multiple welding robots by a preset prediction unit in the twin data layer include:

[0030] Q4. Based on the motion trajectory data of the multiple welding robot twins and the current thermal deformation state of the workpiece twin model, the prediction unit simulates the welding process in the future time period to generate predicted motion trajectory data, predicted weld formation size data, and predicted thermal deformation distribution data;

[0031] Q5. Performing real-time comparative analysis on the predicted motion trajectory data and the real-time motion trajectory of the multiple welding robots in the physical space module, and performing difference calculation on the predicted weld formation size data and the predicted thermal deformation distribution data, and the weld morphology data and the ambient temperature data collected by the physical space module;

[0032] Q6. If the difference calculation result exceeds the preset threshold, the weight coefficient of the welding process parameters in the prediction unit and the preset thermal deformation simulation algorithm are adjusted according to the difference direction, and steps Q4 to Q5 are re-executed until the difference between the predicted motion trajectory data, the predicted weld formation size data and the predicted thermal deformation distribution data and the actual data collected by the physical space module is within the threshold range; if the difference calculation result is less than the preset threshold, the predicted motion trajectory data, the predicted weld formation size data and the predicted thermal deformation distribution data are output to the control module.

[0033] The present invention is further configured as follows: the specific steps of calculating the difference in step Q5 include:

[0034] Q51. Align the predicted motion trajectory data, the predicted weld size data, and the predicted thermal deformation distribution data generated by the prediction unit with the actual motion trajectory data, the actual weld morphology data, and the actual ambient temperature data collected by the physical space module during the same time period;

[0035] Q52. Compare the coordinate deviations of the predicted motion trajectory data and the actual motion trajectory data at each time point, calculate the average offset and maximum offset of each robot joint position, obtain the trajectory offset, compare the weld width and height in the predicted weld formation size data with the measured values ​​of the weld morphology data, calculate the width error percentage and the height absolute error value, obtain the weld error, compare the predicted deformation state data with the actual ambient temperature data, calculate the difference between the predicted temperature and the actual temperature of the key area of ​​the workpiece, and the deviation amplitude of the deformation variable, obtain the thermal deformation deviation, the trajectory offset, the weld error and the thermal deformation deviation are the difference calculation results.

[0036] The present invention is further configured as follows: the specific steps of the control module are as follows:

[0037] H1. Based on the trajectory offset, the weld error, and the thermal deformation deviation, a multi-robot collaborative collision avoidance path planning scheme is generated. Meanwhile, compensation values ​​of welding current and voltage parameters of the multiple welding robots are adjusted according to the weld error, and a welding sequence is optimized based on the thermal deformation deviation.

[0038] H2, sending the generated multi-robot collaborative collision avoidance path planning scheme, compensation value and welding sequence to the multi-welding robots in the physical space module (1) in real time through the logic controller, and driving the multi-welding robots to perform the corrected welding action;

[0039] H3. During the welding process, the multi-welding robot continuously receives the updated prediction data of the twin data layer (3) and the dynamic data of the physical space module (1). If it is detected that the error of the robot posture data or the weld morphology data exceeds a preset threshold, steps H2 to H3 are re-executed until the welding task is completed.

[0040] The beneficial effects of the present invention are as follows:

[0041] 1. This invention uses a physical space module to collect multi-source heterogeneous data such as welding parameters, posture, and temperature in real time. Combining the fusion and synchronization of the twin data layer with a high-fidelity twin model in the virtual space, it builds a closed-loop control framework for "physical-virtual" bidirectional mapping, enabling dynamic collaborative optimization of multi-robot motion trajectories, weld formation, and thermal deformation, thereby improving welding accuracy and efficiency.

[0042] 2. The present invention uses a virtual simulation unit to predict the thermal deformation state of the workpiece in real time, combined with dynamic adjustment of welding parameters, to actively correct the robot's motion path and welding process parameters, effectively suppressing weld offset and workpiece deformation caused by heat accumulation, and improving the welding quality of complex components. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is a system flow chart of the present invention;

[0044] Figure 2 This is a flow chart of the dynamic data collection method of the present invention;

[0045] Figure 3 A flowchart showing the complete mapping of the virtual space module of the present invention from the physical space module;

[0046] Figure 4 A flowchart for verifying the virtual simulation unit of the present invention;

[0047] In the figure: 1. Physical space module; 2. Virtual space module; 3. Twin data layer; 4. Control module. DETAILED DESCRIPTION

[0048] The preferred embodiments of the present invention are described in detail below with reference to the accompanying drawings so that the advantages and features of the present invention can be more easily understood by those skilled in the art, thereby making a clearer and more precise definition of the protection scope of the present invention.

[0049] See also Figures 1-4, a multi-welding robot collaborative operation system based on digital twin, including: a physical space module 1, a virtual space module 2, a twin data layer 3, and a control module 4. Communication connections are established among the physical space module 1, the virtual space module 2, the twin data layer 3, and the control module 4;

[0050] Physical space module 1, which includes multiple welding robots, a sensor group, and a data transmission network. The physical space module 1 performs welding tasks through the multiple welding robots and collects dynamic data during the welding process in real time through the sensor group. The dynamic data is then transmitted through the data transmission network. The data transmission network is a communication system connecting the multiple welding robots, the sensor group, and external modules in the physical space module, and includes industrial protocols, switches, gateways, and a real-time streaming transmission mechanism.

[0051] Among them, the sensor group in the physical space module 1 includes a visual sensor, a force sensor, a temperature sensor, and a current and voltage sensor;

[0052] The dynamic data in the physical space module 1 include welding parameter data, robot posture data, weld morphology data and ambient temperature data. Among them, the welding parameter data include: welding current, voltage, wire feeding speed, welding speed and welding gun force value; the robot posture data include: robot joint angle, welding gun position coordinates and posture angle parameters; the weld morphology data include: weld position, geometry, width, height and depth measurement values; the ambient temperature data include: temperature values ​​of each monitoring point in the welding area and heat input change trend.

[0053] The dynamic data collection method includes:

[0054] S1. Scan the area to be welded of the workpiece by a visual sensor to obtain a first visual image of the weld position, shape and size, and scan the welding posture of the multi-welding robot by a visual sensor to generate a second visual image, and use an image processing algorithm to denoise and enhance the features of the first and second visual images. The first and second visual images are processed as follows: first, random noise is eliminated by median filtering, and background interference is smoothed by Gaussian filtering; then histogram equalization is used to enhance the contrast between the weld and the background, and the weld edge and robot joint contour are sharpened by combining Sobel or Canny operators (existing technology), and image distortion is corrected by affine transformation; finally, edge detection is performed on the first image to extract the weld position, size and shape parameters, key point matching is performed on the second image, and the three-dimensional coordinates and posture angle of the welding gun are calculated to generate weld morphology data and robot posture data with an error of ≤0.1mm;

[0055] S2. Collect temperature distribution data of the welding area of ​​multiple welding robots through temperature sensors, record the trend of heat input changes, and generate ambient temperature data;

[0056] Among them, the heat input change trend includes: the temperature change curve of each monitoring point in the welding area over time (such as peak temperature, heating rate, cooling rate), and the cumulative distribution trend of heat on the workpiece during continuous welding. Specifically, it is manifested as the expansion range of the heat-affected zone along the welding path, the temperature superposition effect of adjacent welding points and the dynamic evolution process of the overall thermal deformation. By recording the real-time data collected by the temperature sensor, the temperature gradient map and heat input history curve are generated to analyze the impact of the welding thermal cycle on the deformation of the workpiece.

[0057] S3. Continuously collect the current value, voltage value, and wire feed speed data of the preset welding power source through current and voltage sensors. At the same time, use force sensors to monitor the force changes when the welding guns of multiple welding robots contact the workpiece in real time, determine the collision risk during welding, and record it in real time to generate welding parameter data. The welding parameter data, robot posture data, weld morphology data, and ambient temperature data are packaged to generate dynamic data, and the dynamic data is uploaded to the twin data layer 3 and the virtual space module 2 through the data transmission network;

[0058] The specific steps for determining the collision risk during welding are as follows:

[0059] S31. Using a force sensor, collect data on force values ​​and torque changes in multiple directions when the welding gun contacts the workpiece in real time;

[0060] S32. Setting the dynamic safety threshold range of the force in each direction according to the current welding parameters and welding gun posture;

[0061] S33. If it is detected that the force value and torque change data of the welding gun exceed the dynamic safety threshold within a preset time period, or there is a sudden abnormal force peak, it is determined that there is a collision risk;

[0062] S34. Generate a welding gun avoidance path adjustment instruction based on the collision risk determination result, and send it to multiple welding robots for execution in real time through the data transmission network.

[0063] Virtual space module 2, which includes multiple welding robot twins and workpiece twin models, is completely mapped from physical space module 1. The specific steps for completely mapping virtual space module 2 from physical space module 1 are as follows:

[0064] M1 and the virtual space module 2 receive dynamic data, and extract the mechanical structure parameters, motion joint parameters and welding process parameters of the multi-welding robot based on the dynamic data, and generate model data of the multi-welding robot twin corresponding to the multi-welding robot. The method for generating the model data of the multi-welding robot twin is as follows: First, the virtual space module 2 receives the dynamic data packet from the physical space module 1, and parses the robot mechanical structure parameters (including the length of the robot arm, the joint connection method and the welding gun installation position), the motion joint parameters (including the degree of freedom, the motion range and the dynamic response characteristics of each joint) and the welding process parameters (including the preset current, voltage, and feed rate). Secondly, the analyzed mechanical structure parameters are converted into a three-dimensional geometric model through parametric modeling tools. The kinematic model of the multi-welding robot twin is constructed by combining the motion joint parameters to simulate the joint motion range and linkage relationship. At the same time, the welding process parameters are bound to the virtual welding gun model, and the current and voltage data are associated with the virtual welding effect (such as the molten pool shape and heat input). Finally, the joint motion accuracy and welding process parameters of the twin are continuously calibrated through real-time dynamic data to ensure that the virtual model is completely consistent with the mechanical structure, motion characteristics and process settings of the physical robot, forming high-fidelity multi-welding robot twin model data.

[0065] M2. Based on the model data of the multi-welding robot twins, a three-dimensional geometric model and motion model of the multi-welding robot twins are established, and combined with the weld morphology data, a workpiece twin model consistent with the workpiece is constructed. According to the real-time dynamic data in the physical space module 1, the motion trajectory and welding process parameters of the multi-welding robot twins are continuously corrected, and a virtual simulation unit is preset to verify the consistency of the corrected motion trajectory and welding process parameters of the multi-welding robot twins with the real-time dynamic data in the physical space module 1. After successful verification, a complete virtual space module 2 is generated.

[0066] According to the real-time dynamic data in the physical space module 1, the motion trajectory and welding process parameters of the multi-welding robot twins are continuously corrected. The steps are as follows:

[0067] M201, the virtual space module 2 receives the robot posture data and welding parameter data uploaded by the physical space module 1 in real time, and dynamically compares them with the current motion trajectory and process parameters of the multi-welding robot twins to calculate the coordinate offset between the actual posture and the virtual model and the welding parameter deviation;

[0068] The method for calculating the coordinate offset and welding parameter deviation between the actual posture and the virtual model is as follows: the actual welding gun position coordinates (such as X / Y / Z axis coordinates) and joint angle data of the multiple welding robots in the physical space module (1) are obtained in real time, and the welding gun virtual coordinates and joint angles of the multiple welding robot twins in the virtual space module (2) are compared point by point to calculate the position difference (such as Euclidean distance) and joint angle deviation between the two in three-dimensional space; at the same time, the actual welding current and voltage values ​​collected in the physical space are calculated with the process parameters bound in the virtual model to obtain the current deviation and voltage deviation, and finally the coordinate offset and welding parameter deviation are generated by integrating the posture offset, current deviation and voltage deviation;

[0069] M202, based on the coordinate offset and welding parameter deviation, adjust the joint motion trajectory of the multi-welding robot twins through the interpolation algorithm to synchronize the virtual welding gun position with the actual multi-welding robot posture, and dynamically compensate the virtual values ​​of the welding current and voltage parameters to match the real-time process status of the physical equipment;

[0070] M203: Input the corrected motion trajectory and welding parameters of the multi-welding robot twins into the virtual simulation unit to simulate the welding process in the future time period, generate predicted weld formation and thermal deformation data, and verify consistency with the real-time collected data of the physical space module 1;

[0071] M204. If the verification result shows that the error between the predicted data and the actual data exceeds the preset threshold, the smoothing coefficient and welding process parameter weight of the multi-welding robot twin motion model are readjusted, and steps M21 to M23 are executed repeatedly until the error converges to within the threshold range.

[0072] Among them, the virtual simulation unit preset in step M2 verifies the consistency of the motion trajectory and welding process parameters of the modified multi-welding robot twins with the real-time dynamic data in the physical space module 1. The specific steps are:

[0073] M21. Based on the corrected motion trajectory and welding process parameters of the multi-welding robot twins, the virtual simulation unit drives the multi-welding robot twins and the workpiece twin model to simulate the welding process and generate virtual welding parameters, virtual weld formation quality data, and virtual thermal deformation state data.

[0074] M22, compare the virtual welding parameters with the actual motion trajectory data of the multiple welding robots in the physical space module 1 in real time, and perform difference analysis on the virtual weld formation quality and virtual thermal deformation state data with the weld morphology data and ambient temperature data collected by the physical space module 1;

[0075] The specific contents of the difference analysis in step M22 include: comparing the weld width, height and geometric shape parameters in the virtual weld formation quality data with the measured values ​​of the actual weld morphology data collected by the physical space module 1, and calculating the dimensional error percentage between the two; at the same time, regionally matching the temperature distribution and workpiece deformation in the virtual thermal deformation state data with the ambient temperature data in the physical space, and analyzing the temperature difference and deformation deviation amplitude of each monitoring point; aligning the virtual and physical data based on the timestamp, and if there is a time delay, interpolating and compensating for it before comparison, and finally determining whether the weld formation error exceeds ±5%, whether the temperature deviation exceeds ±10°C, or whether the deformation error exceeds ±0.2mm. If any indicator exceeds the limit, a correction is triggered;

[0076] M23. If the difference analysis result exceeds the preset threshold, it is determined that the status of the virtual space module 2 is inconsistent with that of the physical space module 1, and the virtual space module 2 is triggered to re-correct the motion trajectory and welding parameters of the multi-welding robot twins; if the difference is within the threshold, it is confirmed that the consistency verification has passed, and the currently corrected virtual space module 2 is locked as a valid mapping.

[0077] The virtual space module 2 simulates the welding process in real time through multiple welding robot twins and workpiece twin models;

[0078] Twin data layer 3: Twin data layer 3 is used to connect the physical space module 1 and the virtual space module 2, fuse and synchronize the dynamic data with the motion trajectory, welding formation quality and thermal deformation state data of the multiple welding robot twins and the workpiece twin models, and predict the motion trajectory, weld formation quality and thermal deformation state of the multiple welding robots through a preset prediction unit;

[0079] Among them, in the twin data layer 3, the dynamic data is integrated and synchronized with the motion trajectory, welding forming quality and thermal deformation state data of the multiple welding robot twins and the workpiece twin model. The specific method includes:

[0080] Q1, the twin data layer 3 receives the dynamic data transmitted by the physical space module 1, and analyzes the robot posture data, welding parameter data, weld morphology data and ambient temperature data in the dynamic data through a preset parsing unit;

[0081] Q2: Match the analyzed robot pose data with the kinematic joint parameters of the multi-welding robot twins one by one, associate the welding parameter data with the virtual welding process parameters of the welding guns of the multi-welding robot twins, and map the weld topography data and ambient temperature data to the corresponding areas of the workpiece twin model.

[0082] Q3, Twin Data Layer 3, based on real-time analyzed dynamic data, drives the twin bodies of multiple welding robots to execute the same motion trajectory as the multiple welding robots in the virtual space, and updates the thermal deformation state and weld formation quality of the workpiece twin model to complete data fusion and synchronization.

[0083] Among them, the specific steps of predicting the motion trajectory, weld formation quality and thermal deformation state of multiple welding robots through the preset prediction unit in the twin data layer 3 include:

[0084] Q4. Based on the motion trajectory data of the multiple welding robot twins and the current thermal deformation state of the workpiece twin model, the prediction unit simulates the welding process in the future time period to generate predicted motion trajectory data, predicted weld formation size data, and predicted thermal deformation distribution data;

[0085] Q5. Perform real-time comparative analysis on the predicted motion trajectory data and the real-time motion trajectory of multiple welding robots in the physical space module 1. At the same time, perform difference calculation on the predicted weld formation size data and predicted thermal deformation distribution data, and the weld morphology data and ambient temperature data collected by the physical space module 1.

[0086] Q6. If the difference calculation result exceeds the preset threshold, the weight coefficient of the welding process parameters and the preset thermal deformation simulation algorithm in the prediction unit are adjusted according to the difference direction, and steps Q4 to Q5 are re-executed until the difference between the predicted motion trajectory data, the predicted weld formation size data and the predicted thermal deformation distribution data and the actual data collected by the physical space module 1 is within the threshold range; if the difference calculation result is less than the preset threshold, the predicted motion trajectory data, the predicted weld formation size data and the predicted thermal deformation distribution data are output to the control module 4, wherein the specific content of the adjustment according to the difference direction includes: if the weld formation size prediction value is greater than the actual value, the weight coefficient of the welding current parameter is reduced and the wire feeding speed weight is increased to reduce the simulated size of the virtual molten pool; if the thermal deformation prediction value exceeds the actual temperature distribution, the thermal conductivity coefficient of the heat input simulation model is reduced or the cooling rate parameter is increased; if the motion trajectory prediction offset direction is consistent with the actual deviation, the smoothing coefficient of the path planning is increased or the joint motion acceleration threshold is limited; at the same time, the iteration step size and the welding parameter association weight of the thermal deformation simulation algorithm are dynamically optimized based on the historical error data until the predicted data is consistent with the physical data trend.

[0087] The specific steps of calculating the difference in step Q5 include:

[0088] Q51. Align the predicted motion trajectory data, predicted weld size data, and predicted thermal deformation distribution data generated by the prediction unit with the actual motion trajectory data, actual weld morphology data, and actual ambient temperature data collected by the physical space module 1 during the same time period.

[0089] Timestamp alignment method: The acquisition time of the predicted data and the data in the physical space module are unified to the same clock source. For data segments with missing or delayed timestamps, virtual data points are inserted into the predicted time series through linear interpolation, or misaligned time periods in the physical data are truncated. A sliding time window (e.g., 0.1 seconds) is also set to ensure that the prediction and physical data match within the same time window.

[0090] Q52. Compare the coordinate deviations of the predicted motion trajectory data and the actual motion trajectory data at each time point, calculate the average offset and maximum offset of each robot joint position, and obtain the trajectory offset. Compare the weld width and height in the predicted weld formation size data with the measured values ​​of the weld morphology data, calculate the width error percentage and the height absolute error value, and obtain the weld error. Compare the predicted deformation state data with the actual ambient temperature data, calculate the difference between the predicted temperature and the actual temperature of the key area of ​​the workpiece, and the deviation amplitude of the deformation variable, and obtain the thermal deformation deviation. The trajectory offset, weld error, and thermal deformation deviation are the difference calculation results.

[0091] Calculation method of trajectory offset: Calculate the Euclidean distance between the predicted and actual coordinates at each time point, and count the average offset and maximum offset value of all robot joints;

[0092] Calculation method of weld error: Compare the predicted and actual weld width and height, and calculate the width error (predicted value - actual value / actual value × 100%) and the absolute height error (predicted value - actual value);

[0093] Calculation method of thermal deformation deviation: Select the key area of ​​the workpiece (such as the periphery of the weld), calculate the mean difference between the predicted temperature and the actual temperature, and the geometric deviation of the deformation variable (the mean vector modulus of the predicted deformation coordinate - the actual deformation coordinate).

[0094] Control module 4, based on the twin data layer 3, generates a welding path optimization plan, a multi-welding robot collaborative collision avoidance strategy and welding quality feedback control instructions, and sends the control instructions to the multi-welding robots in the physical space module 1 for execution through a preset logic controller.

[0095] The specific steps of the control module 4 are as follows:

[0096] H1. Generate a multi-robot collaborative collision avoidance path planning scheme based on trajectory offset, weld error, and thermal deformation deviation. At the same time, adjust the compensation values ​​of the welding current and voltage parameters of multiple welding robots according to the weld error, and optimize the welding sequence based on the thermal deformation deviation.

[0097] H2, the generated multi-robot collaborative collision avoidance path planning scheme, compensation value and welding sequence are sent to the multiple welding robots in the physical space module 1 in real time through the logic controller, driving the multiple welding robots to perform the corrected welding action;

[0098] H3. During the welding process, the multi-welding robot continuously receives the updated prediction data from the twin data layer 3 and the dynamic data from the physical space module 1. If it is detected that the error of the robot posture data or the weld morphology data exceeds the preset threshold, steps H2 to H3 are re-executed until the welding task is completed.

[0099] The above are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. The multi-welding robot collaborative operation system based on digital twin is characterized by: include: Physical space module (1), virtual space module (2), twin data layer (3), control module (4); A physical space module (1), the physical space module (1) comprising multiple welding robots, a sensor group and a data transmission network, the physical space module (1) performing welding tasks through the multiple welding robots, collecting dynamic data in the welding process in real time through the sensor group, and then transmitting the dynamic data through the data transmission network; A virtual space module (2), the virtual space module (2) including multiple welding robot twins and a workpiece twin model, the virtual space module (2) being completely mapped by the physical space module (1); the virtual space module (2) simulating the welding process in real time through the multiple welding robot twins and the workpiece twin model; A twin data layer (3), the twin data layer (3) is used to connect the physical space module (1) and the virtual space module (2), fuse and synchronize the dynamic data with the motion trajectory, welding forming quality and thermal deformation state data of the multi-welding robot twin and the workpiece twin model, and predict the motion trajectory, weld forming quality and thermal deformation state of the multi-welding robot through a preset prediction unit; A control module (4) generates control instructions based on the twin data layer (3), and sends the control instructions to the multi-welding robot for execution.

2. The multi-welding robot collaborative operation system based on digital twin according to claim 1 is characterized in that: Communication connections are established between the physical space module (1), the virtual space module (2), the twin data layer (3) and the control module (4); The sensor group in the physical space module (1) includes a visual sensor, a force sensor, a temperature sensor, and a current and voltage sensor; The dynamic data in the physical space module (1) includes welding parameter data, robot posture data, weld morphology data and ambient temperature data.

3. The multi-welding robot collaborative operation system based on digital twin according to claim 2 is characterized in that: The dynamic data collection method includes: S1. Scanning the area to be welded of the workpiece by the visual sensor to obtain a first visual image of the weld position, shape, and size, and scanning the welding posture of the multi-welding robot by the visual sensor to generate a second visual image, and performing denoising and feature enhancement on the first and second visual images by using an image processing algorithm to generate weld morphology data and robot posture data; S2. Collecting temperature distribution data of the welding area of ​​the multi-welding robot through the temperature sensor, recording the heat input change trend, and generating ambient temperature data; S3. Continuously collect the current value, voltage value and wire feeding speed data of the preset welding power source through the current and voltage sensors. At the same time, monitor the force changes of the welding gun of the multi-welding robot when it contacts the workpiece in real time through the force sensor, judge the collision risk during the welding process, and record it in real time to generate welding parameter data. The welding parameter data, the robot posture data, the weld morphology data and the ambient temperature data are packaged to generate dynamic data, and the dynamic data is uploaded to the twin data layer (3) and the virtual space module (2) through the data transmission network.

4. The multi-welding robot collaborative operation system based on digital twin according to claim 3 is characterized in that: The specific steps of completely mapping the virtual space module (2) from the physical space module (1) are as follows: M1, the virtual space module (2) receives the dynamic data, and extracts the mechanical structure parameters, motion joint parameters and welding process parameters of the multi-welding robot based on the dynamic data, and generates model data of the multi-welding robot twin corresponding to the multi-welding robot; M2. Based on the model data of the multi-welding robot twins, a three-dimensional geometric model and a motion model of the multi-welding robot twins are established, and combined with the weld morphology data, a workpiece twin model consistent with the workpiece is constructed, and based on the real-time dynamic data in the physical space module (1), the motion trajectory and welding process parameters of the multi-welding robot twins are continuously corrected, and a virtual simulation unit is preset to verify the consistency of the corrected motion trajectory and welding process parameters of the multi-welding robot twins with the real-time dynamic data in the physical space module (1). After successful verification, a complete virtual space module (2) is generated.

5. The multi-welding robot collaborative operation system based on digital twin according to claim 4 is characterized in that: In the step M2, a virtual simulation unit is preset to verify the consistency of the corrected motion trajectory and welding process parameters of the multi-welding robot twins with the real-time dynamic data in the physical space module (1), and the specific steps are as follows: M21. Based on the corrected motion trajectory and welding process parameters of the multi-welding robot twins, the virtual simulation unit drives the multi-welding robot twins and the workpiece twin model to simulate the welding process and generates virtual welding parameters, virtual weld formation quality data, and virtual thermal deformation state data; M22, comparing the virtual welding parameters with the actual motion trajectory data of the multiple welding robots in the physical space module (1) in real time, and performing a difference analysis on the virtual weld formation quality and the virtual thermal deformation state data and the weld morphology data and the ambient temperature data collected by the physical space module (1); M23. If the difference analysis result exceeds a preset threshold, it is determined that the virtual space module (2) and the physical space module (1) are in inconsistent states, and the virtual space module (2) is triggered to re-correct the motion trajectory and welding parameters of the multi-welding robot twins; if the difference is within the threshold, it is confirmed that the consistency verification has passed, and the currently corrected virtual space module (2) is locked as a valid mapping.

6. The multi-welding robot collaborative operation system based on digital twin according to claim 5 is characterized in that: The twin data layer (3) fuses and synchronizes the dynamic data with the motion trajectory, welding forming quality and thermal deformation state data of the multi-welding robot twin and the workpiece twin model, and the specific method includes: Q1, the twin data layer (3) receives the dynamic data transmitted by the physical space module (1), and analyzes the robot posture data, welding parameter data, weld morphology data and ambient temperature data in the dynamic data through a preset analysis unit; Q2. Match the analyzed robot pose data with the motion joint parameters of the multi-welding robot twins one by one, associate the welding parameter data with the virtual welding process parameters of the welding guns of the multi-welding robot twins, and map the weld topography data and ambient temperature data to corresponding areas of the workpiece twin model; Q3. The twin data layer (3) drives the multi-welding robot twins to execute the same motion trajectory as the multi-welding robots in the virtual space based on the real-time analyzed dynamic data, and updates the thermal deformation state and weld formation quality of the workpiece twin model to complete data fusion and synchronization.

7. The multi-welding robot collaborative operation system based on digital twin according to claim 6 is characterized in that: The specific steps of predicting the motion trajectory, weld formation quality and thermal deformation state of the multi-welding robot by a preset prediction unit in the twin data layer (3) include: Q4. Based on the motion trajectory data of the multiple welding robot twins and the current thermal deformation state of the workpiece twin model, the prediction unit simulates the welding process in the future time period to generate predicted motion trajectory data, predicted weld formation size data, and predicted thermal deformation distribution data; Q5. Performing real-time comparative analysis on the predicted motion trajectory data and the real-time motion trajectory of the multiple welding robots in the physical space module (1), and performing difference calculation on the predicted weld formation size data and the predicted thermal deformation distribution data, and the weld morphology data and the ambient temperature data collected by the physical space module (1); Q6. If the difference calculation result exceeds a preset threshold, the weight coefficient of the welding process parameters in the prediction unit and the preset thermal deformation simulation algorithm are adjusted according to the difference direction, and steps Q4 to Q5 are re-executed until the difference between the predicted motion trajectory data, the predicted weld formation size data and the predicted thermal deformation distribution data and the actual data collected by the physical space module (1) is within the threshold range; if the difference calculation result is less than the preset threshold, the predicted motion trajectory data, the predicted weld formation size data and the predicted thermal deformation distribution data are output to the control module (4).

8. The multi-welding robot collaborative operation system based on digital twin according to claim 7 is characterized in that: The specific steps of difference calculation in step Q5 include: Q51, aligning the predicted motion trajectory data, the predicted weld formation size data, and the predicted thermal deformation distribution data generated by the prediction unit with the actual motion trajectory data, the actual weld morphology data, and the actual ambient temperature data collected by the physical space module (1) within the same time period; Q52. Compare the coordinate deviations of the predicted motion trajectory data and the actual motion trajectory data at each time point, calculate the average offset and maximum offset of each robot joint position, obtain the trajectory offset, compare the weld width and height in the predicted weld formation size data with the measured values ​​of the weld morphology data, calculate the width error percentage and the height absolute error value, obtain the weld error, compare the predicted deformation state data with the actual ambient temperature data, calculate the difference between the predicted temperature and the actual temperature of the key area of ​​the workpiece, and the deviation amplitude of the deformation variable, obtain the thermal deformation deviation, the trajectory offset, the weld error and the thermal deformation deviation are the difference calculation results.

9. The multi-welding robot collaborative operation system based on digital twin according to claim 8 is characterized in that: The specific steps of the control module (4) are as follows: H1. Based on the trajectory offset, the weld error, and the thermal deformation deviation, a multi-robot collaborative collision avoidance path planning scheme is generated. Meanwhile, compensation values ​​of welding current and voltage parameters of the multiple welding robots are adjusted according to the weld error, and a welding sequence is optimized based on the thermal deformation deviation. H2, sending the generated multi-robot collaborative collision avoidance path planning scheme, compensation value and welding sequence to the multi-welding robots in the physical space module (1) in real time through the logic controller, and driving the multi-welding robots to perform the corrected welding action; H3. During the welding process, the multi-welding robot continuously receives the updated prediction data of the twin data layer (3) and the dynamic data of the physical space module (1). If it is detected that the error of the robot posture data or the weld morphology data exceeds a preset threshold, steps H2 to H3 are re-executed until the welding task is completed.

Images