Automatic welding method and system based on collaborative robot

Abstract

The invention discloses an automatic welding method and system based on a collaborative robot, and the method comprises the steps: collecting and preprocessing three-dimensional point cloud data of a to-be-welded steel part and the collaborative robot, and collecting thermal physical property data of to-be-welded steel; based on the three-dimensional point cloud data and the thermophysical property data of the steel, a welding simulation model is constructed through the digital twin technology; constructing a welding quality evaluation model based on a deep learning technology, and performing a virtual welding test by using the welding simulation model and the welding quality evaluation model to obtain an optimal welding route and optimal welding parameters; and based on the optimal welding route and the optimal welding parameters, the collaborative robot is started to conduct automatic welding. The invention relates to the technical field of automatic welding, and solves the technical problem that an existing automatic welding technology lacks welding parameter self-adaptive optimization and welding route intelligent planning through combination of a digital twinning technology and a deep learning technology.

Description

Technical Field

[0001] The present invention belongs to the field of automated welding and involves digital twin and artificial intelligence technologies, and specifically is an automated welding method and system based on collaborative robots. Background Art

[0002] In modern manufacturing, welding, as a basic and key connection technology, is widely used in many fields such as construction, aviation, automobiles and ships. With the continuous advancement of industrial automation and intelligence, traditional manual welding has been difficult to meet the needs of modern production due to its high labor intensity, low efficiency, unstable quality and other problems. Therefore, the automation and intelligent improvement of welding technology has become an inevitable trend. The emergence of collaborative robots has provided new opportunities for the innovative development of welding technology. Compared with traditional industrial robots, collaborative robots have higher flexibility and coordination, and they can work with human workers without safety fences. Especially in the welding process, collaborative robots can achieve complex welding tasks and improve welding quality and efficiency through advanced visual systems and sensitive force control.

[0003] However, in the welding technology based on collaborative robots, traditional teaching programming and offline programming methods can no longer meet the complex and changeable welding needs to a certain extent, especially in production scenarios with complex welding trajectories or diversified batches. For example, most collaborative robots can only weld according to a predetermined specific trajectory and lack intelligent optimization capabilities, which can easily lead to problems such as uneven weld quality or welding defects. In addition, the existing technology is mostly based on some simple rules or experience to determine the welding parameters, and it is impossible to achieve adaptive parameter optimization, making it difficult to achieve the best welding quality. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes an automated welding method and system based on a collaborative robot, which is used to solve the technical problem of lack of adaptive optimization of welding parameters and intelligent planning of welding routes in the existing automated welding technology.

[0005] To achieve the above objectives, a first aspect of the present invention provides an automated welding method based on a collaborative robot, comprising:

[0006] S100, collecting and preprocessing the three-dimensional point cloud data of the steel parts to be welded and the collaborative robot, collecting the thermophysical property data of the steel parts to be welded, and obtaining the preprocessed three-dimensional point cloud data and thermophysical property data;

[0007] S200, constructing a digital twin model of steel welding based on the three-dimensional point cloud data and the thermophysical property data to obtain a welding simulation model;

[0008] S300, performing a virtual welding test using the welding simulation model and the welding quality assessment model to obtain an optimal welding route and optimal welding parameters; wherein the welding quality assessment model is constructed based on deep learning technology;

[0009] S400: Based on the optimal welding route and optimal welding parameters, the collaborative robot is started to perform automated welding.

[0010] Based on the above technical solution, a digital twin model of steel welding is constructed by collecting and processing the three-dimensional point cloud data of the steel parts to be welded and the robot, as well as collecting the thermal physical properties data of the steel. This can simulate the welding process more accurately and provide a more reliable welding simulation model. By using the welding simulation model and the welding quality assessment model to conduct virtual welding tests, the optimal welding route and welding parameters can be obtained, which helps to guide the actual welding operations and realize the automated welding technology based on collaborative robots, reducing the test cost and time, and improving the efficiency and quality of welding.

[0011] Furthermore, the acquisition of three-dimensional point cloud data of the steel parts to be welded and the collaborative robot, and the collection of thermal physical property data of the steel parts to be welded, includes the following steps:

[0012] S101, data acquisition: using a 3D machine vision sensor to scan the welding area, acquiring three-dimensional point cloud data of the steel to be welded and the collaborative robot, and collecting thermophysical property data of the steel to be welded from an existing material database; wherein the thermophysical property data includes thermal conductivity, heat capacity, and thermal expansion coefficient;

[0013] S102, data preprocessing: performing denoising and filtering preprocessing on the three-dimensional point cloud data to obtain preprocessed three-dimensional point cloud data.

[0014] Furthermore, the construction of the digital twin model of steel welding comprises the following steps:

[0015] S201, using the preprocessed three-dimensional point cloud data to generate a three-dimensional geometric model of the weldment and the collaborative robot through a point cloud reconstruction algorithm;

[0016] S202, based on the thermophysical property data, the preset welding process and welding parameters, establish a welding simulation model using a numerical method on the three-dimensional geometric model; wherein the welding parameters include welding current, welding voltage, preheating temperature, welding temperature, welding speed, welding angle, welding posture, welding time, pre-pressing time and holding time;

[0017] S203, solving the welding simulation model by using a numerical solution method, calculating the temperature field, thermal stress and deformation parameters in the welding process, and obtaining a simulation result;

[0018] S204, performing difference calculation between the simulation results and the temperature field, thermal stress and deformation parameters in the actual welding process, and determining whether the obtained differences are all within the threshold range; if yes, obtaining a welding simulation model; if not, resetting the welding parameters and jumping to S203 for further simulation and optimization.

[0019] By collecting three-dimensional point cloud data and thermal physical properties data of the steel to be welded, a basis is provided for subsequent path planning, simulation and optimization, which helps to predict key parameters such as temperature field, thermal stress and deformation during welding; reconstruction and preprocessing based on point cloud data improves the quality and accuracy of the data, providing reliable input for subsequent model construction and simulation calculations; by simulating the welding process through digital twin technology and performing difference calculations with the actual welding process, the accuracy of the simulation model can be verified and adjusted, thereby improving the credibility and accuracy of the model.

[0020] Furthermore, the welding quality assessment model is constructed based on deep learning technology, including the following steps:

[0021] S301, collecting data: obtaining weld area, root gap area, maximum weld width, average weld width and residual stress data according to the historical welding database, and calculating the weld integrity, taking the weld area, average weld width, residual stress and weld integrity as evaluation indicators of welding quality, and constructing corresponding label data for each weldment, to obtain a number of evaluation indicators and a number of label data;

[0022] S302, constructing a data set: dividing a data set consisting of a plurality of indicator data and a plurality of corresponding label data into a training set and a validation set; wherein the data of the training set and the validation set are in the form of indicator-label pairs;

[0023] S303, training model: constructing a welding quality assessment model based on deep learning technology, using the training set to train the assessment model, using the validation set to obtain the validation accuracy of the assessment model, and obtaining a pre-trained welding quality assessment model.

[0024] Furthermore, the weld integrity is calculated by the formula Calculated; where W is the weld integrity, μ1 is the area coefficient, μ2 is the width coefficient, S1 is the weld area, S2 is the root gap area, L1 is the maximum weld width, and L2 is the average weld width.

[0025] By establishing a welding quality assessment model based on historical data and deep learning technology, we can learn features and patterns from a large amount of data and automatically assess welding quality, thereby reducing the subjectivity and uncertainty of manual assessment and improving the accuracy and consistency of the assessment.

[0026] Furthermore, a virtual welding test is performed using the welding simulation model and the welding quality assessment model, comprising the following steps:

[0027] S401, determining the welding sequence and welding method of the steel parts to be welded according to the welding drawings, and setting the initial welding route and initial welding parameters;

[0028] S402, performing a virtual welding test on the welding simulation model using the initial welding route, the initial welding parameters and the thermophysical property data to obtain weld area, average weld width, residual stress and weld integrity data during the test;

[0029] S403, inputting the weld area, average weld width, residual stress and weld integrity data in the test process into a welding quality assessment model to obtain an assessment result; wherein the assessment result is divided into qualified and unqualified;

[0030] S404, when the evaluation result is qualified, outputting the current welding parameters and welding route to obtain the optimal welding parameters and the optimal welding route;

[0031] When the evaluation result is unqualified, a genetic algorithm is used to search for the best parameter combination, and S402 and S403 are repeated until the evaluation result is qualified, and the current welding parameters and welding route are output to obtain the optimal welding parameters and the optimal welding route.

[0032] Virtual welding tests are carried out using welding simulation models and welding quality assessment models, so that comprehensive simulation and evaluation can be carried out before actual welding, potential problems and defects can be discovered in advance, thereby reducing the risks and costs of the welding process; through automated parameter search and optimization processes, the optimal welding parameters and welding routes can be obtained, thereby improving the efficiency and quality of actual welding.

[0033] The second aspect of the present invention provides an automated welding system based on a collaborative robot, comprising: a data acquisition module, a model building module, a data processing module and an automated welding module connected to each other; wherein:

[0034] The data acquisition module is used to acquire and preprocess the three-dimensional point cloud data of the steel parts to be welded and the collaborative robot, collect the thermophysical property data of the steel parts to be welded, and obtain the preprocessed three-dimensional point cloud data and thermophysical property data;

[0035] The model building module is used to build a welding simulation model based on the preprocessed three-dimensional point cloud data and the thermal physical property data; and a welding quality assessment model based on deep learning technology;

[0036] The data processing module is used to perform a virtual welding test using the welding simulation model and the welding quality assessment model to obtain an optimal welding route and optimal welding parameters;

[0037] The automated welding module is used to start the collaborative robot to perform automated welding using the optimal welding route and optimal welding parameters.

[0038] Compared with the prior art, the present invention has the following beneficial effects:

[0039] By accurately collecting and preprocessing the three-dimensional point cloud data of the steel parts to be welded and the collaborative robots, as well as collecting the thermal physical properties data of the steel, an accurate digital twin model is constructed, which can simulate the actual welding environment and conditions in detail, and predict and optimize the key parameters in the welding process in advance; the welding quality assessment model constructed using deep learning technology can learn complex features and patterns from historical data, realize automatic assessment of welding quality, significantly reduce the subjectivity and uncertainty of manual assessment, and further improve the accuracy and consistency of welding quality assessment; based on the welding simulation model and the welding quality assessment model, a mechanism for virtual testing and optimization of welding is provided, which can test multiple welding strategies without consuming physical materials and resources, so as to find the optimal welding route and welding parameters, realize the automation and intelligent adjustment of welding routes and welding parameters, reduce dependence on expert experience, and improve the quality and efficiency of automated welding. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0041] Figure 1 A schematic flow chart of an automated welding method based on a collaborative robot provided by the present invention;

[0042] Figure 2 A schematic flow chart of adaptively optimizing welding routes and welding parameters based on digital twin technology provided by the present invention;

[0043] Figure 3 A schematic block diagram of an automated welding system based on a collaborative robot provided by the present invention. DETAILED DESCRIPTION

[0044] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0045] See also Figure 1-Figure 2 The first embodiment of the present invention provides an automated welding method based on a collaborative robot, comprising:

[0046] S100, collecting and preprocessing the three-dimensional point cloud data of the steel parts to be welded and the collaborative robot, collecting the thermophysical property data of the steel parts to be welded, and obtaining the preprocessed three-dimensional point cloud data and thermophysical property data;

[0047] S200, based on 3D point cloud data and thermal properties data of steel, builds a digital twin model of steel welding to obtain a welding simulation model;

[0048] S300, using the welding simulation model and the welding quality assessment model to conduct a virtual welding test to obtain the optimal welding route and optimal welding parameters; wherein the welding quality assessment model is constructed based on deep learning technology;

[0049] S400, based on the optimal welding route and optimal welding parameters, starts the collaborative robot for automated welding.

[0050] In this embodiment, in order to construct a reasonable and accurate welding simulation model, it is necessary to first obtain the three-dimensional point cloud data of the steel to be welded and the collaborative robot and the thermal physical properties data of the steel; use a 3D machine vision sensor to scan the welding area, record the shape, size and surface topology information of the steel to be welded, obtain the three-dimensional point cloud data of the steel to be welded and the collaborative robot, and collect the thermal physical properties data of the steel to be welded from the existing material database, including thermal conductivity, heat capacity, and thermal expansion coefficient, which are used to simulate the factors affecting heat conduction and thermal deformation during welding.

[0051] It is worth noting that in order to improve the accuracy of three-dimensional point cloud data, preprocessing operations are required. In this embodiment, a Gaussian filter is first used for denoising; specifically, the Gaussian filter calculates the mean and standard deviation by considering each point and its neighboring points, and then uses these statistical data to adjust the coordinates of the point to remove noise or outliers in the data; then the voxel downsampling method is used to smooth the point cloud data; specifically, the voxel downsampling method divides the point cloud data into small cubic voxels, and then replaces all the points in each voxel with the centroid of all the points in the voxel, which can effectively reduce the number of point clouds and make the data smoother.

[0052] Furthermore, constructing the digital twin model of steel welding includes the following steps: first, applying a point cloud reconstruction algorithm based on Poisson reconstruction to convert the preprocessed point cloud data into an accurate 3D geometric model to reflect the physical size and shape of the weldment and the robot in detail; then, dividing the 3D geometric model into a series of grids, and further refining the grids in areas where accurate results are required, such as near the welding line, to improve the local resolution and accuracy of the simulation;

[0053] Then, the thermophysical property data of the steel is assigned to the geometric model, and then appropriate thermal boundary conditions are set for the model according to the heat input model, convection and radiation conditions of the actual welding process, and appropriate mechanical boundary conditions are set for the model according to the fixture constraints and external forces; then, a detailed welding route is defined in the three-dimensional geometric model, including the welding starting point, focus and direction of travel, and then the initial welding parameters are configured, including welding current, welding voltage, preheating temperature, welding temperature, welding speed, welding angle, welding posture, welding time, pre-pressing time and holding time; welding parameters can be initialized according to the technician's experience;

[0054] Next, a suitable numerical solution method should be selected, such as the finite element method, the finite difference method or the finite volume method. In this embodiment, the finite element method is selected as the solution method to calculate the temperature field, thermal stress and deformation parameters in the welding process, and perform difference calculation with the results in the actual welding process to determine whether the difference is within the threshold range. If yes, a welding simulation model is obtained; if not, the welding parameters in the simulation model are adjusted, and the simulation is re-run for further simulation and optimization.

[0055] The setting of thermal boundary conditions and mechanical boundary conditions can be reflected by an example: suppose that the welding process of a steel structure bridge joint is being simulated. This joint consists of two steel plates and is connected together by arc welding.

[0056] First, the thermal boundary conditions are set: in order to accurately reflect the distribution of heat flux in arc welding, a double ellipsoid heat source model is selected to simulate the heat input of arc welding. The parameters of the double ellipsoid model are adjusted according to the actual welding parameters to ensure that the simulated heat input is consistent with the actual situation; then, convection boundary conditions are applied to the outer surface of the steel plate to simulate the cooling effect of air on the steel plate, and the appropriate convection heat exchange coefficient is set according to the welding environment and the state of the steel plate surface; at the same time, the radiation heat loss of the steel plate surface at high temperature must also be considered, and the radiation heat flux is calculated using the Stefan-Boltzmann law, and the appropriate radiation coefficient is set according to the surface characteristics and temperature range of the steel;

[0057] Next is the setting of mechanical boundary conditions: in the simulation, first fix the position of the steel plate and apply displacement constraints to the non-welding area of ​​the steel plate to simulate the clamping conditions in actual welding, and apply the clamp to constrain the stress and deformation generated during the welding process; if there are additional external forces in the welding process, such as the preset force applied to control the deformation, these forces are applied in an appropriate manner in the model; then define the contact pair on the contact surface between the two steel plates, and set appropriate contact properties, such as friction coefficient and thermal conductivity, while considering the heat transfer and mechanical interaction between the contact surfaces to accurately simulate the behavior of the welded joint;

[0058] Finally, according to the sequence of actual welding operations, heat input and fixture constraints are gradually applied to simulate the movement of heat input during welding, as well as time-varying boundary conditions to capture the temperature and stress changes during welding.

[0059] Based on the above process, a reasonable and accurate welding simulation model can be obtained, which provides strong support for the optimization and quality control of the welding process.

[0060] When applying welding simulation models, a reasonable feedback mechanism is very important. In this embodiment, a quality assessment model based on deep learning is used to judge the quality of welding simulation results and guide the welding simulation model to optimize welding parameters and welding routes; the following are the specific implementation steps:

[0061] Step 1: Determine the welding sequence and method: Set the initial welding route and parameters: According to the welding drawings provided, analyze the design of the steel parts to be welded, determine the welding sequence and the welding method to be used, such as arc welding or laser welding; for example, you can weld the center part of the structure first and then gradually expand outwards; then, set the initial welding route, that is, the trajectory of the welding gun moving on the steel part, and the initial welding parameters;

[0062] Step 2: Conduct a virtual welding test: Use the set initial welding parameters and known thermophysical property data to run a welding simulation model on a computer to conduct a virtual welding test. During this process, the simulation software will simulate the actual welding process and calculate the process of weld formation, including key data such as weld area, average weld width, residual stress distribution, and weld integrity;

[0063] Step 3: Evaluate welding quality: input the obtained weld area, average weld width, residual stress and weld integrity data into the welding quality evaluation model to evaluate whether the welding quality is qualified;

[0064] Step 4: Optimize welding parameters and paths: When the evaluation result is qualified, the welding parameters and paths currently used are considered to be optimal and can be directly output as the final result;

[0065] When the evaluation result is unqualified, a genetic algorithm is used to search for a better combination of welding parameters. The genetic algorithm simulates the process of natural selection and continuously iterates to find the optimal solution. After each iteration, steps 2 to 4 are repeated with a new parameter combination until the welding quality evaluation result is qualified. The current welding parameters and path can be output as the optimal result.

[0066] Based on the above process, not only can the appropriate welding parameters and paths be found, but also the welding quality can be ensured to meet the predetermined standards, thereby reducing the trial and error costs in actual production and improving production efficiency and product quality.

[0067] After obtaining the optimal welding route and optimal welding parameters, the collaborative robot is finally started to perform automated welding according to the welding route and welding parameters.

[0068] See also Figure 3 The second aspect of the present invention provides an automated welding method based on a collaborative robot, comprising: a data acquisition module, a model building module, a data processing module and an automated welding module that are interconnected; wherein,

[0069] A data acquisition module is used to acquire and preprocess the three-dimensional point cloud data of the steel parts to be welded and the collaborative robot, collect the thermophysical property data of the steel parts to be welded, and obtain the preprocessed three-dimensional point cloud data and thermophysical property data;

[0070] Model building module, used to build welding simulation model based on pre-processed 3D point cloud data and thermophysical property data; and welding quality assessment model based on deep learning technology;

[0071] A data processing module is used to perform virtual welding tests using a welding simulation model and a welding quality assessment model to obtain an optimal welding route and optimal welding parameters;

[0072] The automated welding module is used to start the collaborative robot for automated welding using the optimal welding route and optimal welding parameters.

[0073] Part of the data in the above formula is calculated by removing the dimension and taking its numerical value. The formula is a formula closest to the actual situation obtained by software simulation of a large amount of collected data; the preset parameters and preset thresholds in the formula are set by technical personnel in this field according to actual conditions or obtained through simulation of a large amount of data.

[0074] Working principle of the present invention:

[0075] Use 3D machine vision sensors to scan the steel parts to be welded and the welding area of ​​the collaborative robot to obtain three-dimensional point cloud data. At the same time, collect the thermophysical property data of the steel from the material database, and perform denoising and filtering on the collected three-dimensional point cloud data to improve data quality; use the pre-processed three-dimensional point cloud data to generate a three-dimensional geometric model of the welded parts and the collaborative robot through the point cloud reconstruction algorithm; based on the three-dimensional geometric model, combine the thermophysical property data and the preset welding parameters and welding routes to build a welding simulation model; based on historical welding data, use deep learning technology to build a welding quality assessment model; use the welding simulation model to conduct welding simulation tests, and input the simulation results into the welding quality assessment model, adjust the welding parameters according to the assessment results, and obtain the optimal welding parameters and path; based on the obtained optimal welding route and parameters, start the collaborative robot to perform automated welding operations.

[0076] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An automated welding method based on a collaborative robot, characterized in that: include: S100, collecting and preprocessing the three-dimensional point cloud data of the steel parts to be welded and the collaborative robot, collecting the thermophysical property data of the steel parts to be welded, and obtaining the preprocessed three-dimensional point cloud data and thermophysical property data; S200, constructing a digital twin model of steel welding based on the three-dimensional point cloud data and the thermophysical property data to obtain a welding simulation model; S300, performing a virtual welding test using the welding simulation model and the welding quality assessment model to obtain an optimal welding route and optimal welding parameters; wherein the welding quality assessment model is constructed based on deep learning technology; S400: Based on the optimal welding route and optimal welding parameters, the collaborative robot is started to perform automated welding.

2. The automated welding method based on a collaborative robot according to claim 1, characterized in that: The method of collecting the three-dimensional point cloud data of the steel parts to be welded and the collaborative robot and collecting the thermal physical property data of the steel parts to be welded includes the following steps: S101, data acquisition: Scan the welding area using a 3D machine vision sensor to acquire 3D point cloud data of the steel to be welded and the collaborative robot, and collect the thermophysical property data of the steel to be welded from the existing material database; S102, data preprocessing: performing denoising and filtering preprocessing on the three-dimensional point cloud data to obtain preprocessed three-dimensional point cloud data.

3. The automated welding method based on a collaborative robot according to claim 1, characterized in that: The construction of the digital twin model of steel welding comprises the following steps: S201, using the preprocessed three-dimensional point cloud data to generate a three-dimensional geometric model of the weldment and the collaborative robot through a point cloud reconstruction algorithm; S202, based on the thermophysical property data, the preset welding process and welding parameters, establish a welding simulation model using a numerical method on the three-dimensional geometric model; wherein the welding parameters include welding current, welding voltage, preheating temperature, welding temperature, welding speed, welding angle, welding posture, welding time, pre-pressing time and holding time; S203, solving the welding simulation model by using a numerical solution method, calculating the temperature field, thermal stress and deformation parameters in the welding process, and obtaining a simulation result; S204, performing difference calculation between the simulation results and the temperature field, thermal stress and deformation parameters in the actual welding process, and determining whether the obtained differences are all within a threshold range; if yes, obtaining a welding simulation model; if not, resetting the welding parameters and jumping to S203 for further simulation and optimization.

4. The automated welding method based on a collaborative robot according to claim 1, characterized in that: The welding quality assessment model is constructed based on deep learning technology and includes the following steps: S301, collecting data: obtaining weld area, root gap area, maximum weld width, average weld width and residual stress data according to the historical welding database, and calculating the weld integrity, taking the weld area, average weld width, residual stress and weld integrity as evaluation indicators of welding quality, and constructing corresponding label data for each weldment, to obtain a number of evaluation indicators and a number of label data; S302, constructing a data set: dividing a data set consisting of a plurality of indicator data and a plurality of corresponding label data into a training set and a validation set; wherein the data of the training set and the validation set are in the form of indicator-label pairs; S303, training model: constructing a welding quality assessment model based on deep learning technology, using the training set to train the assessment model, using the validation set to obtain the validation accuracy of the assessment model, and obtaining a pre-trained welding quality assessment model.

5. The automated welding method based on a collaborative robot according to claim 4, characterized in that: The weld integrity is given by the formula Calculated; where W is the weld integrity, μ1 is the area coefficient, μ2 is the width coefficient, S1 is the weld area, S2 is the root gap area, L1 is the maximum weld width, and L2 is the average weld width.

6. The automated welding method based on a collaborative robot according to claim 1, characterized in that: The virtual welding test is performed using the welding simulation model and the welding quality assessment model, comprising the following steps: S401, determining the welding sequence and welding method of the steel parts to be welded according to the welding drawings, and setting the initial welding route and initial welding parameters; S402, performing a virtual welding test on the welding simulation model using the initial welding route, the initial welding parameters and the thermophysical property data to obtain weld area, average weld width, residual stress and weld integrity data during the test; S403, inputting the weld area, average weld width, residual stress and weld integrity data in the test process into a welding quality assessment model to obtain an assessment result; wherein the assessment result is divided into qualified and unqualified; S404, when the evaluation result is qualified, outputting the current welding parameters and welding route to obtain the optimal welding parameters and the optimal welding route; When the evaluation result is unqualified, a genetic algorithm is used to search for the best parameter combination, and S402 and S403 are repeated until the evaluation result is qualified, and the current welding parameters and welding route are output to obtain the optimal welding parameters and the optimal welding route.

7. An automated welding system based on a collaborative robot, characterized in that: include: A data collection module (100), a model building module (200), a data processing module (300) and an automated welding module (400).

8. The automated welding system based on a collaborative robot according to claim 7, characterized in that: The data acquisition module (100) is used to acquire and pre-process the three-dimensional point cloud data of the steel parts to be welded and the collaborative robot, collect the thermophysical property data of the steel parts to be welded, and obtain the pre-processed three-dimensional point cloud data and thermophysical property data; The model building module (200) is used to build a welding simulation model based on the preprocessed three-dimensional point cloud data and the thermal physical property data; and a welding quality assessment model based on deep learning technology; The data processing module (300) is used to perform a virtual welding test using the welding simulation model and the welding quality assessment model to obtain an optimal welding route and optimal welding parameters; The automated welding module (400) is used to start the collaborative robot to perform automated welding using the optimal welding route and optimal welding parameters.

Images