An intelligent tracking method and system for a robotic arm to track free-form curve weld seams

Abstract

The present invention relates to the technical field of intelligent tracking of weld seams, and particularly to an intelligent tracking method and system for a robotic arm to track free-form curve weld seams. The method includes the following steps: Deploy a laser profiler on the robotic arm to scan the weld seam, generate weld seam geometric feature data, and perform surface state mapping; then, analyze the distribution trend of weld porosity and estimate the loss of welding compactness, and further evaluate the weld fracture risk; according to the fracture risk data, adjust the welding heat output to optimize the welding quality; finally, construct an intelligent welding tracking model based on the random forest algorithm and upload it to the cloud platform to achieve intelligent tracking welding control of the robotic arm for free-form curve weld seams; the present invention makes the intelligent tracking technology of weld seams more perfect through the optimization of the intelligent tracking technology of weld seams.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent tracking of weld seams, and particularly to an intelligent tracking method and system for a robotic arm to free-form curve weld seams. Background Art

[0002] Robotic arms have high precision and high repeatability, and can effectively control the welding trajectory and posture when performing welding tasks. However, due to the complexity of free-form curve weld seams, robotic arms face challenges of dynamic changes when tracking these irregular paths. Therefore, it is particularly important to develop welding tracking methods based on intelligent technologies. This method not only requires high-precision sensors to monitor the geometric features of the weld seam in real time, but also needs to analyze the state changes of the weld seam with the help of data processing technologies in order to dynamically adjust welding parameters such as welding speed and heat input; through accurate real-time data feedback, the robotic arm can automatically correct the welding trajectory according to the changes of the weld seam, thereby maintaining the welding quality. However, there is a problem in a traditional intelligent tracking method for a robotic arm to free-form curve weld seams that the intelligent tracking welding regulation error for free-form curve weld seams is large, resulting in uneven welding quality among different weld seams. Summary of the Invention

[0003] Based on this, it is necessary to provide an intelligent tracking method and system for a robotic arm to free-form curve weld seams to solve at least one of the above technical problems.

[0004] To achieve the above object, an intelligent tracking method for a robotic arm to free-form curve weld seams, the method includes the following steps:

[0005] Step S1: Deploy a laser profiler on the robotic arm, and perform a curve weld seam profile scan on the free-form curve weld seam to obtain curve weld seam profile data; analyze the surface state of the weld seam for the curve weld seam profile data to obtain weld seam surface state data; perform a geometric trend weld seam surface state mapping process on the weld seam surface state data to obtain weld seam trend surface state mapping data;

[0006] Step S2: Analyze the weld porosity distribution trend for the weld seam trend surface state mapping data to obtain weld porosity distribution trend data; perform a simulation estimate of the welding compactness loss of the weld seam trend distribution based on the weld porosity distribution trend data to obtain weld seam trend compactness loss estimate data; perform a weld fracture risk calculation according to the weld seam trend compactness loss estimate data to obtain weld fracture risk data;

[0007] Step S3: Perform a robotic arm welding heat output constraint regulation matching between different weld seam trend surface states for the weld seam trend surface state mapping data according to the weld fracture risk data to obtain welding heat output constraint regulation data;

[0008] Step S4: Based on the random forest algorithm, construct an intelligent welding tracking model for free-form curve welds using the welding heat output constraint regulation data, and obtain the weld intelligent welding tracking model; send the weld intelligent welding tracking model to the cloud platform to execute the intelligent tracking welding control of the robotic arm for free-form curve welds.

[0009] Preferably, step S1 includes the following steps:

[0010] Step S11: Deploy a laser profiler on the robotic arm and scan the profile of the free-form curve weld to obtain the curve weld profile data;

[0011] Step S12: Clean the curve weld profile data to obtain the curve weld profile cleaning data;

[0012] Step S13: Analyze the geometric features of the curve weld from the curve weld profile cleaning data to generate the curve weld geometric data;

[0013] Step S14: Analyze the surface state of the weld from the curve weld profile cleaning data to obtain the weld surface state data;

[0014] Step S15: Perform a geometric trend weld surface state mapping process on the weld surface state data based on the curve weld geometric data to obtain the weld trend surface state mapping data.

[0015] Preferably, step S2 includes the following steps:

[0016] Step S21: Analyze the distribution trend of weld porosity in the weld trend surface state mapping data to obtain the weld porosity distribution trend data;

[0017] Step S22: Analyze the variation trend of weld width / height in the curve weld geometric data to obtain the weld width / height trend variation data;

[0018] Step S23: Based on the weld porosity distribution trend data and the weld width / height trend variation data, perform a simulation estimation of the welding compactness loss in the weld trend distribution to obtain the weld trend compactness loss estimation data;

[0019] Step S24: Calculate the weld distribution brittleness increment index from the weld trend compactness loss estimation data to obtain the weld distribution brittleness increment index;

[0020] Step S25: Based on the weld trend compactness loss estimation data and the weld distribution brittleness increment index, perform a weld fracture risk calculation on the weld trend surface state mapping data to obtain the weld fracture risk data.

[0021] Preferably, step S23 includes the following steps:

[0022] Step S231: Identify the trend distribution density of the weld porosity distribution trend data to obtain the weld porosity distribution density; based on the weld porosity distribution density, identify the linear distribution of porosity chains in the weld porosity distribution trend data to obtain the linear distribution data of porosity chains;

[0023] Step S232: Calculate the average difference of the effective load-bearing volume of the weld width / height trend change data to obtain the average difference of the effective load-bearing volume of the weld;

[0024] Step S233: Analyze the pore connectivity of the linear distribution data of porosity chains to obtain the pore connectivity data of the porosity distribution;

[0025] Step S234: Based on the average difference of the effective load-bearing volume of the weld and the pore connectivity data of the porosity distribution, perform deviation regression normal distribution sampling on the linear distribution data of porosity chains to obtain deviation regression normal sampling data of the pore structure;

[0026] Step S235: Based on the least squares estimation algorithm, perform simulation estimation of the welding compactness loss of the weld trend distribution on the deviation regression normal sampling data of the pore structure to obtain the estimation data of the weld trend compactness loss.

[0027] Preferably, step S24 includes the following steps:

[0028] Step S241: Identify the stress instability concentration degree of the weld trend compactness loss estimation data to obtain the stress instability concentration degree of the compactness loss;

[0029] Step S242: Based on the stress instability concentration degree of the compactness loss and the weld trend compactness loss estimation data, perform non-linear simulation calculation of the thermal energy plastic strain attenuation to obtain the fitting data of the thermal energy plastic strain attenuation;

[0030] Step S243: Calculate the recursive median absolute deviation of the attenuation of the thermal energy plastic strain attenuation fitting data to obtain the plastic recursive attenuation median absolute deviation;

[0031] Step S244: Based on the plastic recursive attenuation median absolute deviation and the divide-and-conquer algorithm, perform attenuation decomposition constraint analysis on the thermal energy plastic strain attenuation fitting data to generate the attenuation decomposition data of the thermal energy plastic strain attenuation;

[0032] Step S245: Calculate the brittle increment index of the weld distribution based on the attenuation decomposition data of the thermal energy plastic strain attenuation to obtain the brittle increment index of the weld distribution.

[0033] Preferably, step S3 includes the following steps:

[0034] Step S31: Normalize the weld fracture risk data to obtain the normalized weld fracture risk data;

[0035] Step S32: According to the normalized data of weld fracture risk and the estimated data of weld orientation tightness loss, perform mechanical arm welding heat output constraint regulation and matching between different weld orientation surface states on the weld orientation surface state mapping data to obtain welding heat output constraint regulation data.

[0036] Preferably, step S32 includes the following steps:

[0037] Step S321: Perform analysis on the welding heat energy limit values between different weld orientation surface states on the weld orientation surface state mapping data according to the normalized data of weld fracture risk and the estimated data of weld orientation tightness loss to obtain the welding heat energy limit values between different weld orientation surface states;

[0038] Step S322: Based on the welding heat energy limit values between different weld orientation surface states, perform mechanical arm contact radius matching to obtain mechanical arm contact radius matching data;

[0039] Step S323: According to the welding heat energy limit values and the mechanical arm contact radius matching data, perform mechanical arm welding force matching to obtain mechanical arm welding force matching data;

[0040] Step S324: Based on the mechanical arm contact radius matching data, the mechanical arm welding force matching data, and the welding heat energy limit values, perform mechanical arm welding heat output constraint regulation and matching between different weld orientation surface states to obtain welding heat output constraint regulation data.

[0041] Preferably, step S4 includes the following steps:

[0042] Step S41: Perform logical learning on the welding heat output constraint regulation data to obtain welding heat output constraint learning data;

[0043] Step S42: Based on the random forest algorithm, construct a free curve weld intelligent welding tracking model for the welding heat output constraint learning data to obtain a weld intelligent welding tracking model;

[0044] Step S43: Send the weld intelligent welding tracking model to the cloud platform to execute the intelligent tracking welding control of the mechanical arm for the free curve weld.

[0045] Preferably, the present invention also provides an intelligent tracking system for the mechanical arm to free curve welds, which is used to execute the intelligent tracking method of the mechanical arm to free curve welds as described above. The intelligent tracking system for the mechanical arm to free curve welds includes:

[0046] The weld surface state mapping module is used to deploy a laser profiler on the robotic arm, scan the profile of a free-form curve weld to obtain curve weld profile data; analyze the weld surface state of the curve weld profile data to obtain weld surface state data; perform geometric trend weld surface state mapping processing on the weld surface state data to obtain weld trend surface state mapping data;

[0047] The weld fracture risk calculation module is used to analyze the weld porosity distribution trend of the weld trend surface state mapping data to obtain weld porosity distribution trend data; simulate and estimate the welding compactness loss of the weld trend distribution based on the weld porosity distribution trend data to obtain weld trend compactness loss estimation data; calculate the weld fracture risk based on the weld trend compactness loss estimation data to obtain weld fracture risk data;

[0048] The welding heat output constraint regulation module is used to perform robotic arm welding heat output constraint regulation matching between different weld trend surface states on the weld trend surface state mapping data according to the weld fracture risk data to obtain welding heat output constraint regulation data;

[0049] The intelligent welding tracking model construction module is used to construct an intelligent welding tracking model for free-form curve welds based on the random forest algorithm for the welding heat output constraint regulation data to obtain a weld intelligent welding tracking model; send the weld intelligent welding tracking model to the cloud platform to execute intelligent tracking welding control of the robotic arm for free-form curve welds.

[0050] The beneficial effects of the present invention are as follows. By deploying a laser profiler on the robotic arm, precise scanning of free-form curve welds can be achieved. In this process, high-precision contour data of the weld is obtained, providing basic data for subsequent geometric feature analysis. Through the geometric feature analysis of the weld contour data, detailed morphological and structural features of the weld can be extracted, including key information such as the weld path, curvature, width, etc. Using these geometric data, further surface state mapping processing is carried out to obtain weld surface state mapping data. Through this series of precise measurements and analyses, the robotic arm can clearly identify the geometric shape and surface state of the weld, providing an accurate basis for subsequent welding trajectory control and heat input adjustment. Therefore, step S1 not only improves the accuracy of the welding trajectory but also lays a solid foundation for the optimization of the entire intelligent welding process. The weld path surface state mapping data mainly includes the distribution of weld surface defects, such as pores, cracks, slag inclusions, surface unevenness, etc. The distribution, size, and quantity of these defects directly reflect the quality problems in the welding process and affect the strength and durability of the welded joint. By further analyzing the weld path surface state mapping data, especially the analysis of the distribution trend of weld pores, potential defects in the weld can be effectively identified. Pores are one of the common defects in the welding process and have a direct impact on welding quality. By analyzing the distribution trend of weld pores, the density of the weld can be evaluated in real time, and based on the simulation estimate of the loss of welding density, the quality problems that may occur in the welded joint can be accurately predicted. Then, based on this estimated data, the fracture risk of the weld is calculated to obtain the weld fracture risk data. By predicting these defects and assessing the risks in advance, quality problems that may occur during the welding process can be effectively avoided, and necessary data support can be provided for subsequent welding control. This step ensures the high quality and reliability of the welding process and effectively reduces production losses caused by defects. According to the weld fracture risk data, the robotic arm performs heat output constraint regulation and matching between different weld path surface states. This process ensures that the robotic arm can precisely adjust the welding heat input according to different weld states, thereby optimizing the welding effect. Different weld paths and surface states require different welding parameters and heat inputs to achieve the best welding quality. Therefore, precise heat output regulation can effectively avoid welding defects caused by uneven heat input, such as weld offset, hot cracks, etc. Through this process, the welding quality is further improved, and it is ensured that each weld can be completed under the most suitable heat conditions, avoiding joint weakening and performance degradation caused by improper heat input. By analyzing the welding heat output constraint regulation data based on the random forest algorithm, an intelligent welding tracking model for free-form curve welds is constructed. The random forest algorithm can effectively process complex multi-dimensional data and make accurate predictions by integrating the results of multiple decision trees. In the welding tracking model, this algorithm optimizes the heat output regulation data during the welding process, further improving the welding accuracy.The constructed intelligent welding tracking model can adapt to the changes in the weld shape and surface state during the welding process in real time, and intelligently adjust the welding trajectory and heat input to ensure the stability and accuracy of the welding process. After sending this model to the cloud platform, the robotic arm can perform intelligent tracking welding control for free-form curve welds based on real-time data, thus realizing a fully automatic and seamless docking welding process. This innovative intelligent control system not only improves the welding efficiency, but also greatly enhances the welding quality and consistency, promoting the development of the welding industry towards intelligence and automation. Therefore, the present invention is an optimization of the traditional intelligent tracking method of a robotic arm for free-form curve welds, solving the problem that the traditional intelligent tracking method of a robotic arm for free-form curve welds has a large error in intelligent tracking welding regulation for free-form curve welds, resulting in uneven welding quality among different welds, reducing the error of intelligent tracking welding regulation for free-form curve welds, and improving the welding quality among different welds. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a schematic diagram of the step flow of an intelligent tracking method of a robotic arm for free-form curve welds;

[0052] Figure 2 For Figure 1 it is a schematic diagram of the detailed implementation step flow of step S2 in

[0053] Figure 3 For Figure 1 it is a schematic diagram of the detailed implementation step flow of step S3 in DETAILED DESCRIPTION OF THE INVENTION

[0054] Please refer to Figures 1 to 3 An intelligent tracking method of a robotic arm for free-form curve welds, the method includes the following steps:

[0055] Step S1: Deploy a laser profiler on the robotic arm, and scan the profile of the free-form curve weld to obtain the curve weld profile data; analyze the surface state of the weld for the curve weld profile data to obtain the weld surface state data; perform geometric trend weld surface state mapping processing on the weld surface state data to obtain the weld trend surface state mapping data;

[0056] Step S2: Analyze the weld porosity distribution trend for the weld trend surface state mapping data to obtain the weld porosity distribution trend data; perform a simulation estimate of the welding densification loss of the weld trend distribution based on the weld porosity distribution trend data to obtain the weld trend densification loss estimate data; perform a weld fracture risk calculation based on the weld trend densification loss estimate data to obtain the weld fracture risk data;

[0057] Step S3: According to the weld fracture risk data, perform robotic arm welding heat output constraint regulation and matching among different weld surface state mapping data of the weld direction to obtain welding heat output constraint regulation data;

[0058] Step S4: Based on the random forest algorithm, construct an intelligent welding tracking model for free-form curve welds from the welding heat output constraint regulation data to obtain an intelligent welding tracking model for welds; send the intelligent welding tracking model for welds to the cloud platform to execute intelligent tracking welding control of the robotic arm for free-form curve welds.

[0059] In the embodiment of the present invention, referring to Figure 1 as described, it is a schematic diagram of the step flow of an intelligent tracking method for a robotic arm to free-form curve welds. In this example, the intelligent tracking method for a robotic arm to free-form curve welds includes the following steps:

[0060] Step S1: Deploy a laser profiler on the robotic arm, and perform curve weld profile scanning on the free-form curve weld to obtain curve weld profile data; perform weld surface state analysis on the curve weld profile data to obtain weld surface state data; perform geometric direction weld surface state mapping processing on the weld surface state data to obtain weld direction surface state mapping data;

[0061] In the embodiment of the present invention, when deploying a laser profiler on the robotic arm, first select a profile sensor with high-precision laser scanning ability, such as the KEYENCE LJ-V7000 series laser profiler, fix it on the end effector of the robotic arm, and ensure that the laser emission direction is perpendicular to the weld surface. During the scanning process, the robotic arm moves at a constant speed according to a preset trajectory, and the laser profiler real-time collects the height change data of the weld surface. The data is stored in the controller in the form of point cloud. Subsequently, use MATLAB to perform noise reduction processing on the collected point cloud data, adopt the mean filter algorithm to remove high-frequency noise, and then reconstruct the free-form curve profile data of the weld through the B-spline curve fitting method. The data is stored in the SQL database. In the weld surface state analysis stage, perform three-dimensional reconstruction on the point cloud data through OpenCV, use the Canny edge detection algorithm to extract the weld edge feature points, and then use the K-means clustering algorithm to classify the weld surface defects, including pores, slag inclusions, and cracks, etc. The data is stored in JSON format. Then, perform geometric direction weld surface state mapping processing, adopt the Bezier curve interpolation method to generate the weld geometric direction, and based on the surface state data, adopt the principal component analysis (PCA) method to extract the key feature points on the weld surface to generate the weld direction surface state mapping data, and the data is stored in an HDF5 format file.

[0062] In another embodiment, a laser profiler is installed at the end of the robotic arm. The laser profiler has the characteristics of a lateral resolution of 10 μm and a scanning frequency of 4 kHz, and can meet the high-precision real-time contour acquisition requirements for weld curves with large curvature changes. The robotic arm adopts a six-degree-of-freedom structure, model ABB IRB 2600, with a repeat positioning accuracy of ±0.02 mm. By synchronously controlling the path planning of the robotic arm and the scanning of the laser profiler, the robotic arm is driven to scan along the preset initial weld path. The laser profiler continuously collects three-dimensional contour data of the weld, and the data is transmitted in real time to the industrial computing platform via the EtherCAT bus for processing. The original amount of weld contour data obtained is approximately 12,000 point cloud data collected per meter of weld. Subsequently, outlier removal processing based on density clustering is performed on the point cloud data to filter out outliers caused by reflection and jitter, and the weld contour is continuously fitted using the B-spline curve reconstruction algorithm to construct a weld curve geometric model with an accuracy of 0.02 mm. The surface state information of the weld is extracted by combining the surface gray reflectivity change data, including penetration pits, uneven surface oxidation spot areas, splash attachment areas, etc. Through texture gradient direction recognition combined with gray-level co-occurrence matrix calculation, the weld surface state data vector is extracted. Using this vector, dimensionality reduction is performed by the principal component analysis (PCA) method and spatial registration is performed with the weld geometric model to form a mapping data matrix of the weld geometric orientation and surface state, which serves as the basic input for subsequent defect analysis.

[0063] Step S2: Analyze the weld porosity distribution trend of the weld orientation surface state mapping data to obtain the weld porosity distribution trend data; based on the weld porosity distribution trend data, simulate and estimate the welding compactness loss of the weld orientation distribution to obtain the weld orientation compactness loss estimation data; according to the weld orientation compactness loss estimation data, perform a weld fracture risk calculation to obtain the weld fracture risk data;

[0064] In the embodiments of the present invention, in the stage of analyzing the distribution trend of weld porosity, a deep learning method is used to analyze the mapped data of the surface state of the weld trend. First, the YOLOv5 object detection algorithm is used to label the porosity defects, and the main direction of the porosity distribution is calculated based on the weld trend information. The main direction calculation uses the PCA principal axis analysis to obtain the distribution trend data of the weld porosity. The data format is stored in CSV. Subsequently, based on the distribution trend data of the weld porosity, a simulation estimation of the welding densification loss of the weld trend distribution is performed. The finite element analysis (FEA) method is used to calculate the shrinkage stress of the weld metal based on the distribution of the welding energy input, and a prediction model of the weld densification loss is established in combination with the porosity distribution characteristics. A custom numerical solution algorithm written in Python is used to calculate the welding densification loss, and the results are stored as XML format files. Based on the weld trend densification loss estimation data, a weld fracture risk calculation is performed. The fracture mechanics analysis method is used to calculate the crack propagation rate caused by the weld porosity, and the Paris formula is used to evaluate the crack propagation life.

[0065] In another embodiment, the distribution trend of weld porosity is analyzed for the mapped data of the surface state of the weld trend. The density distribution of the weld porosity in different weld trend regions is calculated, and the distribution probability of the porosity is calculated using the histogram statistical method to determine the concentration degree of the porosity in different regions of the weld. For the spatial distribution characteristics of the weld porosity, the Voronoi diagram analysis method is used to calculate the relative distance between the porosities and identify the connectivity and chain-like distribution characteristics of the porosities. Based on the distribution trend data of the weld porosity and combined with the geometric characteristics of the weld surface, an estimation model of the welding densification loss is constructed. First, the effective load-bearing volume of the weld is calculated. The volume segmentation method is used to perform layered calculations on the weld area, and the effective volume of each layer of the weld is calculated. The finite element analysis method is used to simulate the distribution of the heat input during the welding process. The welding heat input is set to 350 J / mm, and the range of the heat affected zone of the weld is calculated. During the simulation, the cooling rate is set to 8 °C / s, the phase change situation during the cooling process of the weld is calculated, and the influence of the weld porosity on the welding densification during the cooling process is analyzed. For the estimation of the weld densification loss, the regression analysis method is used to calculate the seal loss amount of the weld and establish the mapping relationship between the weld seal loss and the porosity distribution. According to the weld trend densification loss estimation data, the fracture mechanics method is used to calculate the weld fracture risk, and the critical stress of the weld crack propagation is calculated based on the Griffith energy criterion. The finite element method is used to establish the weld stress distribution model. The Young's modulus of the weld material is set to 210 GPa, the Poisson's ratio is set to 0.3, and the stress concentration coefficient of the weld area is calculated. For the weld fracture risk calculation, the maximum principal stress on the weld surface is calculated and compared with the fracture toughness of the weld material to determine whether there is a fracture risk in the weld. Finally, the weld fracture risk data is obtained, providing basic data for subsequent weld tracking control.

[0066] Step S3: According to the weld fracture risk data, perform robotic welding heat output constraint regulation and matching among different weld surface states for the weld surface state mapping data of the weld path to obtain welding heat output constraint regulation data;

[0067] In the embodiment of the present invention, according to the weld fracture risk data, perform robotic welding heat output constraint regulation and matching among different weld surface states. First, establish a welding heat input mathematical model. Based on the weld geometry, material properties, and welding process parameters, use the heat conduction equation to calculate the temperature distribution of the weld, perform welding heat conduction simulation using COMSOL Multiphysics software, and combine the weld fracture risk data to optimize the welding heat input for different weld paths. The optimization method uses the Particle Swarm Optimization (PSO) algorithm, with the minimum welding heat input as the objective function to ensure that the weld joint has uniform metallurgical structure and mechanical properties. The optimized welding heat output parameters are stored as a JSON file and finally uploaded to the welding control system of the robotic arm to achieve intelligent matching of welding heat output.

[0068] In another embodiment, input the weld fracture risk data into the welding heat output regulation module. First, perform maximum-minimum normalization processing on the fracture risk data to map all risk values to the [0, 1] interval. Then, combine the weld geometry path information to classify the state of each weld segment. Use the K-means clustering algorithm to divide the weld surface state mapping data of the weld path into five state segments, including high-curvature high-risk area, low-curvature medium-risk area, wide-weld low-risk area, etc. For each type of state segment, calculate the upper and lower limits of heat energy input according to its normalized fracture risk value and weld width-to-height ratio, use the fuzzy control method to set the heat output regulation rules, generate the corresponding welding heat output power regulation value, and the control logic is that the higher the risk level, the lower the heat input. At the same time, adjust the welding torch movement speed and angle to minimize the heat-affected zone. Finally, form a welding heat output constraint regulation data table, which records parameters such as the regulation power value, welding torch angle, and welding speed for each 1 mm weld segment.

[0069] Step S4: Based on the random forest algorithm, construct an intelligent welding tracking model for free-form curve welds for the welding heat output constraint regulation data to obtain an intelligent welding tracking model for welds; send the intelligent welding tracking model for welds to the cloud platform to perform intelligent tracking welding control of the robotic arm for free-form curve welds.

[0070] In the embodiments of the present invention, based on the random forest algorithm, an intelligent welding tracking model for free-form curve welds is constructed for the welding heat output constraint regulation data. First, data on different weld geometries, welding heat inputs, and weld quality are collected to construct a weld feature dataset in the CSV format. Subsequently, the Scikit-learn library in Python is used to train a random forest model with 100 decision trees, and the maximum depth of each tree is set to 10. Five-fold cross-validation is used during the training process, and the mean squared error (MSE) is used as the evaluation metric. After the model training is completed, the XGBoost algorithm is used to optimize the random forest model, and finally, an intelligent welding tracking model for welds is obtained. The model is stored in the ONNX format. During the model deployment phase, the intelligent welding tracking model for welds is uploaded to the AWS cloud platform and communicates with the robotic arm control system through the MQTT protocol to achieve remote intelligent tracking welding control.

[0071] In another embodiment, the welding heat output constraint regulation data is input into the intelligent tracking model construction module, and a model is constructed using the random forest algorithm with a tree structure depth of 10. The training data includes a six-dimensional feature combination of weld seam direction, surface state characteristics, pore distribution parameters, densification loss rate, fracture risk level, and heat output parameters. The number of training samples is 100,000. The Gini coefficient is used as the splitting criterion during training, and the forest structure is optimized through 5-fold cross-validation. Finally, an intelligent welding tracking model for welds is generated. This model takes the weld geometry direction and surface state as inputs and outputs the corresponding heat output power, welding speed, and path correction angle. After the model training is completed, the model file is uploaded to the cloud platform management system through the MQTT protocol, and the cloud platform automatically distributes it to the robotic arm controller. The controller reads the weld state data fed back by the laser profiler in real time during the actual welding process and performs path prediction and heat output regulation by calling the model to achieve dynamic intelligent tracking welding control of free-form curve welds.

[0072] Step S1 includes the following steps:

[0073] Step S11: Deploy a laser profiler on the robotic arm and scan the profile of the free-form curve weld to obtain the curve weld profile data;

[0074] Step S12: Clean the curve weld profile data to obtain the cleaned curve weld profile data;

[0075] Step S13: Analyze the geometric features of the curve weld for the cleaned curve weld profile data to generate curve weld geometric data;

[0076] Step S14: Analyze the surface state of the weld for the cleaned curve weld profile data to obtain the weld surface state data;

[0077] Step S15: Perform geometric trend weld surface state mapping processing on the weld surface state data according to the geometric data of the curved weld to obtain weld trend surface state mapping data.

[0078] In the embodiment of the present invention, when deploying a laser profiler on the robotic arm, the KEYENCE LJ-V7000 series laser profiler is selected and fixed on the end effector of the robotic arm. A customized aluminum alloy bracket is used to ensure the installation stability of the laser profiler. A six-axis force sensor is used to calibrate the installation angle so that the laser beam is vertically projected onto the weld surface. The robotic arm moves at a constant speed along the set path. The laser profiler emits high-frequency laser beams, receives the optical signals reflected from the weld surface, and calculates the three-dimensional contour data of the weld surface through the triangulation method. The data point interval is controlled within 0.05 mm. The collected point cloud data is transmitted to the industrial control computer through the EtherCAT protocol and stored as a PLY format file. When cleaning the curved weld contour data, first, the radius filtering algorithm is used to remove the outlier points in the point cloud. The radius threshold is set to 0.1 mm, and only the data points that meet the density conditions within the local neighborhood are retained. Subsequently, the statistical filtering method is applied to calculate the average distance of each point in the point cloud, and the abnormal data points exceeding 3 times the standard deviation are removed. The MLS (Moving Least Squares) surface reconstruction method is used to smooth the curved weld contour to eliminate the high-frequency noise during the acquisition process. The cleaned curved weld contour data is stored in the PCL (Point Cloud Library) format and a data backup is performed.

[0079] When analyzing the geometric features of a curved weld seam from the cleaned data of the curved weld seam profile, the RANSAC (Random Sample Consensus) algorithm is used to perform plane fitting on the point cloud data. With the weld center line as the reference coordinate system, the width, height, and groove angle of the weld seam are calculated. The DBSCAN (Density-Based Spatial Clustering of Applications with Noise) method is used to cluster the boundary points of the weld seam, calibrate the boundary line of the weld seam, extract the discontinuous feature points of the weld seam by combining the second derivative change rate, identify the regions of sudden change in weld seam curvature, and use the least squares method to fit a B-spline curve to represent the geometric trend of the weld seam. Finally, the weld seam geometric data is generated, stored in JSON format, and uploaded to the welding control system. When analyzing the surface state of the curved weld seam from the cleaned data of the curved weld seam profile, first, the point cloud data is meshed, and the weld seam surface is divided into unit grids of 0.1 mm × 0.1 mm. The Canny edge detection algorithm of OpenCV is used to extract the main feature contours of the weld seam surface. The gray-level co-occurrence matrix (GLCM) is used to calculate the texture features of the weld seam surface to distinguish the molten pool region, heat-affected zone, and base metal region. The ResNet-50 deep learning model is used to classify the defects on the weld seam surface, and the defects are divided into three categories: pores, slag inclusions, and undercut. The K-means clustering algorithm is used to analyze the surface roughness data, and the root mean square value (Rq) of the surface roughness is calculated. The weld seam surface state data is stored in HDF5 format and synchronized to the welding quality monitoring system.

[0080] When performing geometric trend weld seam surface state mapping processing on the weld seam surface state data based on the curved weld seam geometric data, the Bezier curve interpolation method is used to construct the weld seam trend curve. Combining the PCA (Principal Component Analysis) method, the dimensionality reduction processing of the weld seam surface state data is carried out, and the key feature points on the weld seam surface are extracted. The Poisson surface reconstruction algorithm is used to generate a three-dimensional weld seam surface state model. The weld seam geometric information and surface state data are fused to establish a mapping relationship. Finally, the weld seam trend surface state mapping data is generated, stored in an SQLite database, and imported into the weld seam intelligent tracking control module.

[0081] In another embodiment, by customizing the installation of a laser profiler, the installation angle of the laser profiler is controlled to be at an angle of 30 degrees with the normal of the weld surface, ensuring that the laser beam irradiates the weld surface in an inclined manner to enhance the reflection contrast of the uneven features of the weld. When the profiler is working, it performs line scanning at a frequency of 4 kHz through a built-in blue laser source. The scanning line width is set to 25 mm, and the resolution is set to 10 μm. The robotic arm moves along the free-form curve weld trajectory to be welded at a speed of 100 mm / s according to a preset path. The path planning uses an offline trajectory point set generated based on a CAD model, and a trajectory control point is set every 10 mm. The profiler transmits the scanned two-dimensional profile data to the industrial computing unit in the control cabinet in real time through an EtherCAT high-speed communication interface. The three-dimensional point cloud reconstruction of the weld is realized by the coordinated control of the scanning frequency and the robotic arm speed, obtaining a weld profile data set composed of approximately 12,000 contour lines per meter of weld. The data set includes spatial information such as the height of the weld cross-section, the position of the weld boundary, the reinforcement height, and the groove shape.

[0082] After obtaining the weld profile data, it is necessary to clean the original point cloud data to eliminate the outliers and noise points generated by surface reflection or impurities interference during the scanning process. The DBSCAN density clustering algorithm is used to cluster and identify the point cloud on each contour line during the cleaning process. The clustering radius is set to 0.05 mm, and the minimum number of samples is 4. After clustering, the isolated points and small low-density noise clusters are removed, and only the main contour point set is retained. At the same time, local smoothing processing is performed on the point set. The Gaussian weighted sliding window method is used to smooth each contour line, with the window size set to 5 and the weight standard deviation to 1.2. Further, the height mutation between the contour lines is checked. If the height difference between the corresponding points of adjacent contour lines exceeds 0.2 mm, it is determined as an abnormal jump, and cubic spline interpolation is used for transition reconstruction. Finally, the three-dimensional weld profile cleaning data with good continuity and outliers removed is generated, and the data retention rate after cleaning is about 92%. When analyzing the geometric features of the cleaned weld profile data, first, the geometric parameters of each cross-section contour are extracted, including weld reinforcement, weld width, groove angle, weld center offset, etc. The outermost point pairs of the contour line are used as the width boundary, and their horizontal distance is calculated as the weld width. The distance from the highest point in the vertical direction to the plate reference plane is defined as the reinforcement. The groove angle is obtained by calculating the tangent value through fitting the inclined section of the contour edge. The weld center offset is the horizontal offset distance between the contour centroid and the preset center line. The collected geometric parameters are stored in units of every 10 mm weld to form a curve weld geometric data sequence. At the same time, the weld direction is analyzed. The least squares method is used to fit the connection line of the weld centers of each sampling point, and the curvature change of the fitted curve is extracted as the weld geometric direction parameter. Finally, a geometric feature data set containing weld reinforcement, weld width, groove angle, center offset, and direction curvature is formed. When analyzing the weld surface state of the weld profile cleaning data, the reflection intensity value of each contour point is obtained by using the gray reflection measurement function of the profiler, and the weld surface defect area is identified in combination with image texture analysis technology. First, the texture features, including energy, contrast, uniformity, and correlation, are extracted by using the gray-level co-occurrence matrix. The window size is set to 5×5, and the step size is 1. The whole weld surface is slid and calculated to identify the area with abnormal texture features as the surface state abnormal area. At the same time, the morphology of the weld surface is analyzed, and the Ra value of the weld surface roughness is extracted, which is obtained by calculating the average value of the height difference between adjacent points. Each weld section takes 100 mm in length as an analysis unit, and its state features such as surface melt penetration depression area, splash adhesion area, oxidation discoloration area, etc. are counted to form a weld surface state data set containing gray reflection value, texture parameters, surface roughness, and surface abnormal types. This data set corresponds one-to-one with the geometric feature data for subsequent mapping processing.

[0083] Perform geometric trend weld surface state mapping processing on the weld surface state data based on the extracted weld geometric data. Use the spatial repositioning method to remap the surface state data to the corresponding position of the geometric model in the three-dimensional space. First, construct a spatial trend reference coordinate system through the weld geometric center line, with the weld trend as the X-axis and the normal direction as the Z-axis. Use local coordinate transformation to convert the surface state data points from the scanning coordinate system to the weld trend coordinate system. Subsequently, bind the surface state data according to the weld geometry segmentation, establish a point-to-point data mapping matrix. Each data point contains the trend curvature value, width, reinforcement, groove angle of the corresponding section and its corresponding surface state parameters. Finally, generate the weld trend surface state mapping data. The storage format of this data is a multi-dimensional array structure, which includes spatial position index, geometric parameter subset, surface state subset and mapping index relationship. The mapping data is used as the basic input for subsequent weld structure defect assessment and welding heat control model construction.

[0084] Step S2 includes the following steps:

[0085] Step S21: Analyze the weld porosity distribution trend of the weld trend surface state mapping data to obtain the weld porosity distribution trend data;

[0086] Step S22: Analyze the weld width / height trend change of the curved weld geometric data to obtain the weld width / height trend change data;

[0087] Step S23: Based on the weld porosity distribution trend data and the weld width / height trend change data, perform a simulation estimate of the weld trend distribution welding densification loss to obtain the weld trend densification loss estimate data;

[0088] Step S24: Calculate the weld distribution brittleness increment index for the weld trend densification loss estimate data to obtain the weld distribution brittleness increment index;

[0089] Step S25: Perform a weld fracture risk calculation on the weld trend surface state mapping data according to the weld trend densification loss estimate data and the weld distribution brittleness increment index to obtain the weld fracture risk data.

[0090] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes:

[0091] Step S21: Analyze the weld porosity distribution trend of the weld trend surface state mapping data to obtain the weld porosity distribution trend data;

[0092] In the embodiment of the present invention, when analyzing the distribution trend of weld porosity from the weld surface state mapping data of the weld seam direction, first, the weld surface state mapping data stored in the SQLite database is called, and the data format is parsed using the pandas library of Python to extract the coordinates of the surface feature points of the weld seam and the porosity annotation information. The Open3D tool is used to perform voxelization processing on the three-dimensional point cloud data, divide the weld surface into grid cells of 0.1 mm × 0.1 mm, calculate the distribution density of porosity in each cell, use the DBSCAN density clustering algorithm to identify the main direction of the porosity distribution, calculate its direction angle, combine with the PCA principal axis analysis method to obtain the main direction of the overall porosity distribution of the weld seam, generate the weld porosity distribution trend data, store the data in the HDF5 format, and synchronize it to the weld quality monitoring system.

[0093] In another embodiment, when analyzing the distribution trend of weld porosity from the weld surface state mapping data of the weld seam direction, first, a three-dimensional weld model is constructed based on the weld surface point cloud data, and the YOLOv5 deep learning object detection algorithm is used to identify the porosity defects. The gray value of the pixel points in the porosity area is at least 15% lower than the average gray value of the surrounding weld metal. Morphological operations are used to refine the porosity edge, and the center coordinates and size information of the porosity are identified through Hough transform. The center coordinates of all pores are mapped to the weld geometry curve, the DBSCAN density clustering algorithm is used to classify the pores, the distribution direction of the pores on the weld seam is calculated according to the clustering results, the principal component analysis (PCA) method is used to extract the main distribution axis of the pores, generate the weld porosity distribution trend data, and store the data in the JSON format.

[0094] Step S22: Analyze the variation trend of the weld width / height direction of the curved weld geometry data to obtain the weld width / height direction variation data;

[0095] In the embodiment of the present invention, when analyzing the variation trend of the weld width / height direction of the curved weld geometry data, first, the curved weld geometry data is read, the matplotlib library is called to draw the cross-sectional contour of the weld seam, the variation trends of the weld width and height are calculated, the Sobel edge detection algorithm is used to extract the weld boundary points, the least squares method is used to fit the weld width and height curves, the geometric change rate of the weld seam in different welding directions is calculated, the second derivative analysis method is used to calculate the fluctuation degree of the weld width / height, and the Fourier transform method is used to analyze its frequency characteristics, generate the weld width / height direction variation data, store the data in the JSON format, and import it into the welding process monitoring system.

[0096] In another embodiment, when analyzing the variation trend of the weld width / height along the weld line for the geometric data of the curved weld, based on the weld point cloud data, the least squares method is used to fit the weld center line, and the width and height of the weld along the center line direction are calculated. The moving window method is adopted to analyze the variation trend of the weld width / height. The window length is set to 2 mm, and the sliding step is 0.5 mm. The mean and standard deviation of the weld width and height within each window are calculated. The wavelet transform is used to perform frequency domain analysis on the variation signals of the weld width and height, and the high-frequency components are extracted to detect the local mutation regions of the weld. The Fourier transform is used to calculate the periodic variation trend of the weld width and height along the weld line. The data of the variation trend of the weld width / height along the weld line are stored in the SQLite database and synchronized to the welding process control system.

[0097] Step S23: Based on the weld porosity distribution trend data and the weld width / height trend variation data, perform a simulation estimation of the weld line distribution welding densification loss to obtain the weld line densification loss estimation data;

[0098] In the embodiment of the present invention, during the simulation estimation of the weld line distribution welding densification loss, first, a finite element analysis model needs to be established for the weld area, and appropriate material parameters are selected, such as the thermal conductivity, specific heat capacity, and density of the weld base material, to ensure that the model can accurately reflect the heat conduction and phase change behaviors during the welding process. Subsequently, based on the surface porosity distribution data of the weld, pores with different scales are implanted in the finite element mesh, the morphological characteristics of the pores (such as elliptical, circular, or irregular shapes) are defined, and a lower thermal conductivity is given to the pore area to reflect the shielding effect of the pores on the welding heat transfer. The steady-state and transient heat conduction coupling analysis is adopted, a welding heat source model (such as a Gaussian distribution heat source) is applied, and the range of the heat input power density is set to 350–420 J / mm. The moving heat source method is used to simulate the temperature field evolution process of the weld area. During the cooling stage, combined with the trend variation data of the weld width and height, the dynamic deformation process of the weld geometric boundary is defined, and the shrinkage stress distribution of the weld area is calculated based on the thermal stress analysis method. Through the material microstructure simulation, the density change of the weld area is calculated to quantify the weld densification loss caused by the pores and the weld geometric non-uniformity. Using the effective density ratio of the weld area (i.e., the ratio of the actual density to the theoretical density of the weld area after welding), combined with the densification loss rate calculation formula, the welding densification loss estimation data for different regions are obtained, and finally, the weld line densification loss distribution data are formed to provide basic data support for the subsequent weld fracture risk assessment.

[0099] In another embodiment, when simulating and estimating the loss of welding compactness due to the distribution trend of weld porosity and the change trend of weld width / height during the welding process, a weld microstructure model is first established. The phase field method is used to simulate the solidification process of weld metal. The distribution information of pores and the data of weld width / height change are input into the finite element simulation software Abaqus. The thermal-structural coupling analysis is used to calculate the stress distribution inside the weld. Based on the thermal cycle curve inside the weld and combined with the metallographic structure characteristics of the welded joint, the hardness gradient of the heat affected zone (HAZ) of the weld is calculated. The numerical integration method is used to calculate the proportion of the loss of compactness along the weld direction, generating the estimation data of the loss of weld compactness along the direction. The data is stored in the MAT format and imported into the weld quality analysis system to simulate and estimate the loss of welding compactness caused by the defect distribution of the weld direction.

[0100] Step S24: Calculate the brittle increment index of the weld distribution based on the estimation data of the loss of weld compactness along the direction, obtaining the brittle increment index of the weld distribution;

[0101] In the embodiments of the present invention, when calculating the brittle increment index of weld distribution for the weld orientation compactness loss estimation data, it is first necessary to call the weld orientation compactness loss estimation data, which is stored in the database and stores the metal compactness distribution of the weld area in HDF5 format. The h5py library in Python is used to read the compactness loss values of each area of the weld, and the data is standardized to remove outliers and then converted into an input format that can be parsed by ABAQUS. In the ABAQUS finite element software, first establish a finite element model of the weld area, use the C3D8R (8-node reduced integration element) element type for mesh division, and set the minimum element size of the mesh to 0.1 mm to ensure the accuracy of the simulation results. Import the welding material parameters, where the weld metal selects ER70S-6 low-carbon steel filler metal and the base metal selects Q235 steel, and assign their elastic modulus E = 210 GPa and Poisson's ratio ν = 0.3 respectively, and define the Johnson-Cook constitutive model to describe the strain hardening effect of the weld area, and use the bilinear hardening model to describe the plastic deformation characteristics of the weld under the high-temperature welding state. Set the initial yield strength of the material to y = 350 MPa and the maximum tensile strength to u = 500 MPa. During the simulation analysis of residual stress, set the heat source movement path during the welding process, use the Goldak double ellipsoid heat source model to simulate the heat input during the welding process, set the heat source power to P = 6 kW, the welding speed to v = 5 mm / s, and the heat source semi-axis parameters to af = 3 mm, ar = 6 mm, b = 4 mm. Calculate the heat cycle curve of the weld metal during the welding process, use the thermal-structural coupling analysis method to calculate the evolution of the internal temperature field and stress field of the weld, apply the thermal boundary conditions, set the convective heat transfer coefficient to h = 50 W / (m²·K), and the ambient temperature to 293 K. Finally, solve the weld residual stress distribution and export the equivalent stress values of each node. Process the obtained weld residual stress data using the SciPy library in Python, use the np.gradient function to calculate the stress gradient, and then solve the stress concentration factor SCF (Stress Concentration Factor) of the weld area. Classify the stress distribution in the weld area by the K-means clustering method, set the K value to 3, and divide the weld area into three regions: high brittleness, medium brittleness, and low brittleness. Calculate the brittle increment index of each region using fracture mechanics theory. The specific calculation is based on the Griffith energy release rate criterion, using to calculate the brittle increment of each region, where is the maximum principal stress in the stress concentration area, is the elastic modulus of the material. The calculated brittle increment index is stored in the MAT file format and synchronized to the welding quality control system for subsequent welding fracture risk assessment.

[0102] Step S25: Perform a weld fracture risk calculation on the weld surface state mapping data based on the weld path tightness loss estimation data and the weld distribution brittleness increment index to obtain weld fracture risk data.

[0103] In the embodiments of the present invention, when performing the calculation of the weld fracture risk, first, the weld orientation density loss estimation data and the weld distribution brittleness increment index data stored in the database are read. The weld orientation density loss estimation data is stored in the HDF5 format and contains the density distribution information of each region of the weld. The weld distribution brittleness increment index data is in the MAT format and covers the brittleness increment index of different regions of the weld. All data is imported and processed through the h5py library and scipy library of Python. Based on these data, then the extended finite element method (XFEM) is used to predict the crack propagation path of the weld. The XFEM method can effectively capture the tip singularity of the crack and the dynamic behavior of crack propagation. The crack propagation simulation of the weld region is realized by using the ABAQUS finite element software. For the division of the weld region, a grid with high precision is used, and the minimum element size is set to 0.1 mm. The C3D8R element type is used for division. Considering the different material properties of the weld, the heat-affected zone and the base metal, the material of the weld region is selected as ER70S-6 low-carbon steel, and the base metal is selected as Q235 steel. The fracture toughness parameters of the welding material are input, including KIC (critical stress intensity factor) and GIC (critical energy release rate). These parameters are KIC = 80 MPa√m and GIC = 0.1 kJ / m² respectively. Then, by using the Cohesive Zone Model (CZM) to simulate the crack propagation in different regions of the weld, the CZM describes the transition zone of the material from crack-free to fracture by introducing the behavior of the cohesive zone, and considers the stress and strain distribution characteristics at the crack tip and the non-uniformity of the welding region. The maximum strength of the cohesive zone in the model is defined as 70 MPa, and the maximum energy release rate is 0.2 kJ / m². For the fatigue crack propagation behavior of the weld, the Paris-Erdogan crack propagation equation is used to calculate the crack propagation rate of the weld under fatigue load. The form of this equation is da / dN = C(ΔK)^m, where da / dN is the crack propagation rate, C = 1.0×10^-11 (unit: mm / cycle), m = 3.0 is the empirical coefficient of the material, and ΔK is the change in the stress intensity factor. During the calculation process, the stress amplitude and load frequency are input based on the actual welding conditions. The fatigue load frequency of the welding conditions is set to f = 1 Hz, and the load amplitude is set to ΔP = 200 N. The crack propagation life of the weld under fatigue load is calculated through these parameters. After the crack propagation path of the weld is simulated, the fracture risk of each region is further calculated, and the occurrence probability of weld fracture is evaluated based on the time history of crack propagation. Combining the weld orientation density loss data, the weld distribution brittleness increment index data and the fatigue crack propagation data, the stochastic simulation of the fracture risk is carried out through the Monte Carlo method, and finally the weld fracture risk data is obtained.The data includes the fracture occurrence probability of the weld seam, crack propagation time, regional fracture risk, etc. All calculation results are stored in CSV format, including the fracture risk values of different regions of the weld seam and the corresponding crack propagation life information. Finally, all weld fracture risk data is uploaded to the welding quality prediction system through the API interface. The welding quality prediction system further analyzes and processes the data, and generates a corresponding quality assessment report for subsequent welding process optimization and risk control.

[0104] Step S23 includes the following steps:

[0105] Step S231: Identify the trend distribution density of the weld porosity distribution trend data to obtain the weld porosity distribution density; based on the weld porosity distribution density, identify the pore chain linear distribution of the weld porosity distribution trend data to obtain the pore chain linear distribution data;

[0106] Step S232: Calculate the average difference of the effective load volume of the weld width / height trend change data to obtain the average difference of the weld effective load volume;

[0107] Step S233: Analyze the pore connectivity of the pore chain linear distribution data to obtain the pore distribution pore connectivity data;

[0108] Step S234: Perform deviation regression normal distribution sampling of the pore chain linear distribution data according to the average difference of the weld effective load volume and the pore distribution pore connectivity data to obtain the deviation regression normal sampling data of the pore structure;

[0109] Step S235: Based on the least squares estimation algorithm, perform simulation estimation of the weld trend distribution welding densification loss on the deviation regression normal sampling data of the pore structure to obtain the weld trend densification loss estimation data.

[0110] In the embodiment of the present invention, first, the spatial density of the weld surface porosity distribution trend data is calculated. Through rasterization processing, the weld surface area is divided into multiple grid units with equal spacing. The number of pores is counted in each grid unit, and the pore density value per unit area is calculated. Then, the kernel density estimation algorithm is used to continuously process the weld surface porosity distribution density, so that the density data can more accurately reflect the distribution characteristics of pores in the weld area. After obtaining the weld porosity distribution density, the connected region analysis method is used to cluster and identify the spatial arrangement of pores, and the pore regions with continuous distribution and small spacing are divided into chain structures. The minimum spanning tree algorithm is used to calculate the shortest connection path between pores, identify the linear pore chain structure, and extract its length, direction angle, average spacing between pores and other parameters to form the pore chain linear distribution data.

[0111] Analyze the geometric characteristics of the weld cross-section and extract the spatial variation data of the weld width and height. Using the curve fitting method, smooth the changes in the width and height of the weld along its trend to eliminate the interference of local noise on data analysis. Subsequently, based on the weld cross-sectional area calculation method, determine the bearing volume of the weld at different positions, and use statistical analysis methods to calculate the mean difference of the weld bearing volume. The calculation of the mean difference in bearing volume involves the volume deviation of different cross-sections in the weld area. The discrete integral method can be used to numerically calculate the trend of weld volume change to obtain the mean difference of the effective bearing volume of the weld, providing basic data for subsequent simulation of welding densification loss. Through the topological analysis method, evaluate the connectivity of the linear distribution data of the pore chains obtained in the previous step. First, based on the three-dimensional reconstruction method, map the pore structure on the weld surface to the three-dimensional space to construct a topological network of pore distribution. Use the Dijkstra algorithm to calculate the shortest path between pores, judge the connectivity between different pores, and evaluate the permeability of pores by calculating the length of the connectivity path and the channel impedance. Based on the fluid dynamics simulation method, analyze the flow characteristics of the pore connectivity channels, calculate the effective permeability coefficient between pores, and combine with the densification parameters of the weld material to form pore distribution pore connectivity data. Based on the statistical regression method, conduct a modeling analysis of the deviation characteristics of the weld pore structure.

[0112] First, use the least squares regression algorithm to establish a non-linear regression model between the average difference in the effective load-bearing volume of the weld seam and the pore connectivity data of the pore distribution, and calculate the regression coefficient of the weld seam pore structure deviation. Then, based on the error distribution of the regression model, use the normal distribution sampling method to randomly sample the deviation of the weld seam pore structure, generating pore structure deviation data that conforms to the characteristics of the normal distribution. This data can be used for the simulation calculation of the loss of weld seam tightness to ensure the consistency between the pore distribution characteristics of the weld seam and the welding quality assessment model. Use the least squares estimation algorithm to simulate the loss of weld seam tightness for the normal sampling data of the deviation regression pore structure obtained in the previous stage. First, based on the finite element analysis method, establish a heat conduction model for the weld seam area, and define the thermal and mechanical parameters of the weld seam material, including the thermal conductivity, specific heat capacity, and density of the weld seam base material. Then, use the moving heat source model to simulate the heat input to the weld seam area, set the welding heat input power range to 380–420 J / mm, and calculate the phase change behavior during the cooling process of the weld seam. Combine the pore structure data to evaluate the tightness of the weld seam area. Use the material volume fraction analysis method to calculate the proportion of the density loss of the weld seam caused by the pore distribution, and use the least squares method to fit the distribution curve of the loss of weld seam tightness. Finally, obtain the estimated data of the loss of weld seam tightness along the weld seam direction, providing data support for the assessment of the weld seam fracture risk. Because pores themselves are void defects in the weld seam, directly destroying the continuity and tightness of the weld metal. These pores provide channels for gases or liquids. Even when pressure is applied to the weld seam surface or in an environment with sealing requirements, the medium can penetrate through the pores, resulting in the loss of tightness of the weld seam. In addition, due to the irregular shape around the pores, stress concentration is likely to occur, triggering microcracks. These microcracks will further expand and connect with each other, further deteriorating the tightness of the weld seam.

[0113] In another embodiment, when identifying the trend distribution density of the weld seam pore distribution trend data, first divide the weld seam trend path into equally spaced analysis units, with each unit length set to 10 mm. Count the center points of the pores identified in each unit, and calculate the number of pores per unit length, which is the local pore distribution density. Set the density threshold to 0.3 pores / mm. When the density value exceeds this threshold, it is marked as a high-density section. At the same time, perform a linear fit on the spatial distribution direction of the pores in each section. Use the least squares linear fit method to obtain the main direction of the pore distribution spindle, and calculate the angle between this spindle and the main direction of the weld seam trend. If the angle is less than 15 degrees and the fitting residual is less than 0.05 mm, it is identified as a pore chain-like linear distribution area. Record parameters such as the start and end positions, fitting direction, average density, chain length, and linear fitting residual of this area as pore chain-like linear distribution data. This data is organized in an array structure, containing the spatial index and geometric statistical characteristics of each chain-like area, and is used for subsequent connectivity and tightness modeling.

[0114] When calculating the average difference of the effective bearing volume for the data on the variation of the weld width and height trends, first extract the cross-sectional profiles of each weld segment in the 3D weld model. The contour extraction interval is set to 10 mm. Calculate the weld cross-sectional area of each cross-sectional profile as the local bearing volume of that segment. Use the triangulation method to discretize the contour enclosed area and calculate the integral area. Subsequently, perform statistical processing on the bearing volume data of all weld segments, calculate the overall average bearing volume, and calculate the average difference for each segment. The average difference is defined as the absolute value of the difference between the volume of that segment and the overall mean. The obtained average difference of the weld effective bearing volume is used to measure the consistency and stability of the weld structure. Bind the average difference value of each segment to its corresponding spatial index and output it as the weld effective bearing volume average difference data, which provides a geometric constraint basis for subsequent pore sampling deviation modeling. When performing pore connectivity analysis on the data of the chain-like linear distribution of pores, use the 3D voxelization method to divide the weld space into cubic units with a side length of 0.1 mm. Perform voxel mapping on the center points of the pores contained in each pore chain-like region, mark the voxels where the pores are located as 1, and mark the non-pore voxels as 0. Subsequently, use the 3D connected region recognition algorithm to perform a 26-neighborhood search on the voxel group marked as 1 to identify the interconnected pore structures. Calculate the volume, surface area, maximum connected path length, and connectivity coefficient of each connected region. The connectivity coefficient is defined as the product of the connected path length and the volume divided by the surface area. The larger the value, the easier it is for the pores to form a connected channel. Number all the identified connected regions and record their positions in the weld space. The output pore distribution pore connectivity data includes the connected region number, start and end positions, connected path length, volume, connectivity, and the number of the chain-like structure where it is located, which is used for subsequent pore structure sampling modeling.

[0115] When performing deviation regression normal distribution sampling on the linear distribution data of pore chains according to the average difference in the effective load-bearing volume of welds and the pore connectivity data of pore distribution, first construct a pore parameter distribution model. Take the connectivity coefficient of each chain-shaped pore region as the main variable and the average difference in the load-bearing volume of the corresponding region as the regression offset. Use a normal distribution sampling model with a deviation term to simulate the sampling of pore sizes. Set the average pore diameter to 0.2 mm and the standard deviation to 0.05 mm. The regression offset adjusts and corrects the sampling mean. If the average difference in the load-bearing volume of a certain region is large, the corresponding sampling mean shifts upward to increase the pore size. At the same time, increase the pore sampling density in regions with high connectivity. Conduct 100 samplings in each chain-shaped structure, record the pore position, diameter, volume, and distance to adjacent pores for each sampling. The sampling results are output in the form of a structure array, named deviation regression normal sampling data of pore structure, for subsequent simulation estimation of densification loss. When performing simulation estimation of welding densification loss in the weld direction distribution based on the least squares estimation algorithm for the deviation regression normal sampling data of pore structure, map the sampling data into the three-dimensional weld model. Taking the weld direction segments as units, calculate the ratio of the total volume of all sampled pores in each segment to the local load-bearing volume of the weld segment, which is defined as the densification deficiency ratio. Use this ratio as the target value to construct a least squares objective function. With pore size, pore number, and connectivity path length as independent variables, perform least squares regression fitting to obtain the densification deficiency prediction value for each weld segment. Add a regularization term during the fitting process to prevent overfitting. Each weld segment in the fitting result contains an estimated densification value and a confidence interval. The output weld direction densification loss estimation data is organized in the form of a one-dimensional array, and each element contains a spatial position index, densification loss rate, fitting residual, and lower and upper limits of the confidence interval, which are used as data inputs for subsequent brittle increment calculation and fracture risk analysis.

[0116] Step S24 includes the following steps:

[0117] Step S241: Identify the stress instability concentration degree of the weld direction densification loss estimation data to obtain the densification loss stress instability concentration degree;

[0118] Step S242: Based on the densification loss stress instability concentration degree and the weld direction densification loss estimation data, perform non-linear simulation calculation of thermal energy plastic strain attenuation to obtain thermal energy plastic strain attenuation fitting data;

[0119] Step S243: Perform attenuation recursive median absolute deviation calculation on the thermal energy plastic strain attenuation fitting data to obtain the plastic recursive attenuation median absolute deviation;

[0120] Step S244: According to the plastic recursive attenuation median absolute deviation and the divide-and-conquer algorithm, perform attenuation decomposition constraint analysis on the thermal energy plastic strain attenuation fitting data to generate thermal energy plastic strain attenuation decomposition data;

[0121] Step S245: Calculate the weld distribution brittleness increment index based on the thermoplastic strain attenuation decomposition data to obtain the weld distribution brittleness increment index.

[0122] In the embodiments of the present invention, for the identification of stress instability concentration degree of weld seam orientation densification loss estimation data, first, it is necessary to construct a local stress gradient change function based on the stress distribution of the weld cross-section, numerically solve the stress states of different regions inside the weld using the finite element method, calculate the equivalent stress distribution using the Von Mises criterion, select the maximum stress point in the weld region as the starting point for stress concentration degree calculation, set multiple monitoring nodes inside the weld, each monitoring node is arranged along the normal direction of the weld curve, judge the range of the stress concentration region by calculating the local stress gradient change rate of each monitoring point, calculate its instability risk index for the region with a sharp change in stress gradient, classify and cluster the instability region using the Gaussian mixture model (GMM), and calibrate the high stress concentration points on the weld seam orientation with different categories of stress instability regions, so as to obtain the stress instability concentration degree data of densification loss. Based on the stress instability concentration degree of densification loss and the weld seam orientation densification loss estimation data, perform a non-linear simulation calculation of thermal energy plastic strain attenuation. First, extract the temperature field data in the heat-affected zone (HAZ) of the weld, establish a temperature gradient change model using the heat conduction equation, calculate the thermal expansion strain in combination with the thermal expansion coefficient of the weld material, introduce the thermoplastic constitutive relationship to simulate the plastic strain evolution of the material in a high-temperature environment, calculate the material yield strength at different temperatures using the Johnson-Cook model, establish a thermal energy plastic strain attenuation function according to the non-linear change law of the yield strength, use the finite element iterative solution method to fit the thermal energy plastic strain attenuation, calculate the fitting residual using the least squares method, and adjust the model parameters to make the thermal energy plastic strain attenuation data converge within the error range of the experimental measurement value, so as to obtain the thermal energy plastic strain attenuation fitting data. Perform a decay recursive median absolute deviation calculation on the thermal energy plastic strain attenuation fitting data. First, define the time series X(t) to represent the plastic strain attenuation values at different time steps, calculate the median M of the time series, calculate the absolute deviation of all observation points from the median, and obtain the median of these deviations. Use the median of the deviations as a measure of non-linear decay, and further use the recursive regression analysis method to iteratively predict the change trend of the median of the deviations in the time series. Calculate the prediction error for each iteration and update the weight parameters of the recursive model to minimize the prediction error, and finally obtain the plastic recursive decay median absolute deviation data. According to the plastic recursive decay median absolute deviation and the divide-and-conquer algorithm, perform a decay decomposition constraint analysis on the thermal energy plastic strain attenuation fitting data. First, perform a Fourier transform on the thermal energy plastic strain attenuation data, extract the main attenuation components in the frequency domain, decompose the attenuation curve into multiple independent attenuation modes, group different attenuation modes using the divide-and-conquer algorithm, calculate the attenuation contribution degree between each group, and adjust the boundary conditions of the attenuation grouping based on the constraint optimization method to minimize the overall error of the attenuation grouping, and finally generate the thermal energy plastic strain attenuation decomposition data.Based on the thermoplastic strain attenuation decomposition data, the brittle increment index of weld distribution is calculated. First, the brittle increment index of the weld is defined. Its calculation is based on the thermal strain rate, yield strength change rate, and stress concentration in different regions of the weld. The principal component analysis (PCA) method is used to optimize the weights of different influencing factors, and the three most important influencing factors are selected as the core variables for brittle increment calculation. A brittle increment index calculation model is constructed, and the gradient descent method is used to optimize the parameters of the calculation model, so as to obtain the brittle increment index of weld distribution.

[0123] In another embodiment, first, the densification loss values of different regions of the weld seam are extracted from the densification loss estimation data along the weld seam direction, and a grid data structure is established according to the coordinate distribution of the weld seam path. The finite element mesh of the weld seam region is constructed by triangulation. The equivalent stress within the grid elements is calculated using the Von Mises yield criterion. According to the stress-strain relationship curve of the welding material, the equivalent stress of each grid element is converted into a plastic strain increment. Through the fracture instability determination method based on the maximum principal stress criterion, the stress concentration positions in the weld seam region are screened, and the instability degree of the stress concentration region is calculated. Finally, the Kriging interpolation method is used to interpolate and fit the stress instability data, generating the stress instability concentration degree of the weld seam densification loss, and the result is stored as a structured data file. Using the stress instability concentration degree of the densification loss and the densification loss estimation data along the weld seam direction as input variables, the Johnson-Cook constitutive model is used to calculate the thermoplastic deformation behavior of the welding material at high temperatures. The strain rate effect formula in continuum mechanics is used to calculate the decay trend of the plastic strain over time, and a non-linear plastic strain decay function based on time and temperature is constructed. The weld seam region is divided into multiple calculation units according to the temperature gradient. The plastic strain decay function is numerically solved within each calculation unit, and the Runge-Kutta method is used for integral calculation. Finally, the thermal energy plastic strain decay fitting data is generated, and the calculation result is stored as a data table in matrix format. The thermal energy plastic strain decay fitting data is screened, and after abnormal data points are removed, the median absolute deviation (MAD) method is used to calculate the deviation distribution of the plastic strain decay data. First, the median of the data sample is calculated, then the absolute deviation value between each data point and the median is calculated, and the median of these absolute deviation values is calculated again to obtain the plastic recursive decay median absolute deviation. Subsequently, the piecewise linear interpolation method is used to recursively fit the plastic decay data, generating a set of plastic recursive decay data with smooth characteristics, which is stored in the data analysis platform. Based on the plastic recursive decay median absolute deviation, the divide-and-conquer algorithm is used to perform regional analysis and processing on the thermal energy plastic strain decay fitting data. The weld seam region is divided into multiple independent calculation sub-domains, and independent plastic strain decay calculations are performed within each sub-domain. The Lagrange interpolation method is used to construct the thermal energy plastic strain decay curve, and the plastic strain decay gradient of different regions of the weld seam is calculated through the optimal sub-structure recursive solution method. Finally, the thermal energy plastic strain decay decomposition data is generated and stored as a hierarchical data structure for subsequent calculations.According to the thermal plastic strain attenuation decomposition data, a microscopic structure evolution model of welding materials is used to calculate the brittleness increment index of different regions of the weld. First, the grain size of the weld microstructure is statistically analyzed, and the Hall-Petch relationship is used to calculate the influence degree of grain size on material brittleness. Then, combined with the cumulative damage theory of plastic deformation, the brittleness increment of the weld under thermal cycling is calculated. The brittleness data of the weld area is dimensionally reduced by the principal component analysis (PCA) method, and the least squares regression method is used to construct a prediction model for the brittleness increment of the weld, and the brittleness increment index of the weld distribution is calculated. The calculation formula is as follows: , where represents the brittleness increment index of the weld distribution, is the number of grid cells in the weld area, is the index, is the yield strength of the th grid cell, is the initial yield strength, is the predicted value of the grain size of the th grid cell, and the finally calculated brittleness increment index data of the weld distribution.

[0124] Step S25 includes the following steps:

[0125] Step S251: Estimate the welding heat input loss of the weld surface state mapping data according to the weld orientation compactness loss estimation data to obtain the weld orientation heat input loss data;

[0126] Step S252: Analyze the internal residual stress of the weld orientation based on the weld orientation heat input loss data to obtain the internal residual stress data of the weld orientation;

[0127] Step S253: Perform a strain gradient difference calculation of the weld interface layer orientation on the weld surface state mapping data according to the internal residual stress data of the weld orientation and the brittleness increment index of the weld distribution to obtain the interface layer strain gradient difference data;

[0128] Step S254: Deduce the critical extreme value interval of the weld according to the interface layer strain gradient difference data and the brittleness increment index of the weld distribution to obtain the critical extreme value interval of the weld;

[0129] Step S255: Calculate the weld fracture risk on the weld surface state mapping data based on the critical extreme value interval of the weld to obtain the weld fracture risk data.

[0130] In the embodiments of the present invention, based on the weld orientation compactness loss estimation data, a simulation estimation of the welding heat input loss for the weld orientation surface state mapping data is carried out. First, based on the heat source model of the welding process, a three-dimensional finite element model for welding heat transfer is constructed. The welding current, voltage, and welding speed parameters are input into the model to calculate the heat input distribution under standard welding conditions. Based on the weld orientation compactness loss estimation data, the positions and distribution characteristics of the weld defect regions are extracted, and the geometric shape parameters of the weld defects are input into the finite element model. The changes in the material thermal physical property parameters of the defect regions are defined to simulate the local absorption and thermal resistance effects of the defects on the welding heat input. The temperature field evolution along the weld orientation direction is calculated through the unsteady heat conduction equation, and the local loss amount of the welding heat energy is calculated based on the principle of heat flux conservation. An adaptive mesh division strategy is adopted to refine the mesh in the high-temperature gradient region of the weld defect region to improve the calculation accuracy, and the implicit time integration method is used to solve the temperature field to obtain the heat energy loss distribution data at each time step during the welding process. For weld defects with different orientations, the spatial variation of the heat input loss rate is calculated, and the interpolation method is used to fit the global heat energy loss data of the weld to establish a regression model for the heat energy input loss along the weld orientation. By comparing the loss data with the standard welding heat input, the percentage of the overall heat energy input loss of the weld is calculated, and finally, the heat energy input loss data along the weld orientation are obtained. Based on the heat energy input loss data along the weld orientation, an analysis of the internal residual stress along the weld orientation is carried out. The welding thermo-elastoplastic finite element analysis method is used to numerically simulate the residual stress inside the weld. The welding residual stress distribution function is defined, and the stress gradient at each depth layer inside the weld is calculated. The stress concentration degree of the high-stress regions in the heat-affected zone and the fusion zone is evaluated. The inverse solution method is used to analyze the distribution trend of the residual stress along the weld orientation. Through the fitting error analysis of the residual stress distribution curve, the boundary conditions are adjusted to obtain the internal residual stress data along the weld orientation. According to the internal residual stress data along the weld orientation and the weld distribution brittleness increment index, a differential calculation of the strain gradient along the weld orientation for the weld orientation surface state mapping data is carried out. First, the strain gradient distribution model of the weld interface layer is defined, local strain monitoring points of the interface layer are set, the residual stress data are extracted, and the elastic strain distribution of the weld interface layer is calculated based on Hooke's law. The non-linear change characteristics of the plastic strain are calculated in combination with the weld distribution brittleness increment index. The central difference method is used to calculate the strain gradient along the weld orientation direction of the weld interface layer, and the high-order difference method is used to calculate the change rate of the strain gradient to identify the regions where the strain gradient changes violently. Finally, the differential data of the strain gradient of the interface layer are obtained.Deduce the critical extreme value interval of the weld seam based on the strain gradient difference data of the interface layer and the brittle increment index of the weld seam distribution. First, construct a critical strain model for weld seam fracture. Input the strain gradient difference data of the interface layer into the critical strain model to calculate the local plastic limit at different positions of the weld seam. Determine the high-risk areas where fractures occur in the weld seam through the distribution of critical plastic strain. Combine the brittle increment index of the weld seam distribution to analyze the brittle evolution trend of different areas of the weld seam. Use the piecewise regression method to partition the strain distribution along the weld seam, define the upper and lower boundaries of the critical extreme value interval of the weld seam, and use the numerical iteration method to adjust the boundary parameters to make the critical condition of weld seam fracture meet the safety margin range of the material. Finally, obtain the critical extreme value interval of the weld seam. Based on the critical extreme value interval of the weld seam, perform a weld seam fracture risk calculation on the mapped data of the surface state along the weld seam. First, establish a weld seam fracture probability model, define the fracture toughness parameters of the weld seam material, extract the strain and stress data of the critical extreme value interval of the weld seam, use the fracture mechanics method to calculate the fracture safety factor at different positions of the weld seam, calculate the fracture failure probability for the high-risk points within the critical extreme value interval of the weld seam, use the Monte Carlo method to numerically simulate the weld seam fracture probability, generate a weld seam fracture risk distribution curve, and calculate the overall fracture risk index of the weld seam through the extreme value statistics method. Finally, obtain the weld seam fracture risk data.

[0131] Step S3 includes the following steps:

[0132] Step S31: Normalize the weld seam fracture risk data to obtain normalized weld seam fracture risk data;

[0133] Step S32: Based on the normalized weld seam fracture risk data and the estimated data of the tightness loss along the weld seam, perform a mechanical arm welding heat output constraint regulation and matching between different surface states along the weld seam for the mapped data of the surface state along the weld seam to obtain welding heat output constraint regulation data.

[0134] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes:

[0135] Step S31: Normalize the weld seam fracture risk data to obtain normalized weld seam fracture risk data;

[0136] In an embodiment of the present invention, the weld fracture risk data is normalized. First, according to the numerical distribution of the weld fracture risk data, the maximum value and the minimum value are selected as the normalization boundaries, and the minimum-maximum normalization method is used to convert the weld fracture risk data, and all the data are mapped to the interval 0, 1. For the fracture risk data of different weld directions, the mean and standard deviation are calculated respectively, and the Z-score standardization method is used to adjust the data scale so that the mean of all data is zero and the standard deviation is one, thereby eliminating the numerical deviation between different weld directions and improving the comparability of the data. In order to prevent outliers from affecting the normalization results, the box plot analysis method is used to detect outliers, calculate the quartile range, eliminate abnormal fracture risk data points that exceed the normal distribution range, and recalculate the normalization parameters to ensure that the normalized data is evenly distributed, and finally obtain the normalized data of weld fracture risk.

[0137] Step S32: matching the welding heat output constraint control of the robot arm between different weld direction surface states with the weld direction surface state mapping data according to the weld fracture risk normalization data and the weld direction density loss estimation data to obtain the welding heat output constraint control data.

[0138] In the embodiment of the present invention, the welding heat output constraint control matching of the manipulator between different weld direction surface states is performed on the weld direction surface state mapping data according to the weld fracture risk normalization data and the weld direction density loss estimation data. First, a welding heat output control database is constructed, which includes the standard heat input range under different weld directions, and the heat energy compensation value required for different weld areas is calculated in combination with the weld direction density loss estimation data. According to the weld fracture risk normalization data, the risk weight of each weld section is calculated, and the welding heat input target value of each weld direction is calculated using the weighted average method. The dynamic constraint optimization algorithm is used to match and control the welding heat input parameters of the manipulator, and the objective function is constructed. The minimization of the deviation of the welding heat input is taken as the optimization target, and the weld surface state mapping data is introduced as the constraint condition to calculate the heat input adjustment coefficient between different weld directions. The adaptive step size adjustment strategy is adopted to dynamically adjust the welding current, voltage and moving speed of the manipulator, so that the welding heat input gradually approaches the target value, and the heat input power is adjusted in combination with the real-time temperature monitoring feedback, and finally the welding heat output constraint control data is obtained.

[0139] Step S32 includes the following steps:

[0140] Step S321: analyzing the welding heat energy limit values ​​between different weld direction surface states on the weld direction surface state mapping data according to the weld fracture risk normalization data and the weld direction density loss estimation data, and obtaining the welding heat energy limit values ​​between different weld direction surface states;

[0141] Step S322: Based on the welding heat energy limit values between the surface states of different weld bead orientations, perform matching of the robotic arm contact radius to obtain the robotic arm contact radius matching data;

[0142] Step S323: Based on the welding heat energy limit values and the robotic arm contact radius matching data, perform matching of the robotic arm welding force to obtain the robotic arm welding force matching data;

[0143] Step S324: Based on the robotic arm contact radius matching data, the robotic arm welding force matching data, and the welding heat energy limit values, perform matching of the robotic arm welding heat output constraint regulation between the surface states of different weld bead orientations to obtain the welding heat output constraint regulation data.

[0144] In the embodiments of the present invention, based on the normalized data of weld fracture risk and the estimated data of weld tightness loss along the weld direction, the welding heat energy limit values between different surface states of the weld direction are analyzed for the surface state mapping data of the weld direction. First, the normalized data of weld fracture risk is partitioned and calculated. Taking the weld direction as the classification criterion, multiple sub-intervals are divided, and the mean and variance of the normalized data of weld fracture risk within each sub-interval are calculated. Combining the estimated data of weld tightness loss along the weld direction, the Lagrange interpolation method is used to calculate the change trend of weld tightness loss along the weld direction, and the numerical integration method is used to calculate the cumulative tightness loss of each weld area. Based on the tightness loss data and the normalized data of weld fracture risk in different weld areas, a welding heat energy input demand equation is constructed, a welding penetration threshold is set, and the welding heat energy limit value is calculated in combination with the heat input-penetration response curve, so as to obtain the welding heat energy limit values between different surface states of the weld direction. Based on the welding heat energy limit values between different surface states of the weld direction, the contact radius of the robotic arm is matched. First, the maximum welding temperature rise in the weld area is calculated according to the welding heat energy limit value, the heat diffusion radius of the welding area is calculated using Fourier's law of heat conduction, and the thermal field of the weld area is numerically simulated using a heat conduction simulation software to determine the spatial range of heat diffusion. Combining the geometric dimensions and welding angles of the welding torch at the end of the robotic arm, the contact area of the welding torch at the end of the robotic arm in different weld directions is calculated, and the optimal contact radius of the robotic arm is calculated according to the expansion radius of the welding pool. The gradient descent method is used to iteratively optimize the contact radius of the robotic arm to match the welding area determined by the welding heat energy limit value, and finally the robotic arm contact radius matching data is obtained. Based on the welding heat energy limit value and the robotic arm contact radius matching data, the robotic arm welding force matching is carried out. First, the surface tension of the molten metal in the welding area is calculated using the welding heat energy limit value, and the welding pressure distribution applied by the robotic arm is determined in combination with the robotic arm contact radius matching data. The finite element method is used to simulate the stress situation in the welding contact area, calculate the stress uniformity in different weld areas, and optimize the welding pressure parameters of the robotic arm using the stress distribution analysis method. Combining the curvature change situation of the weld direction, the torque on the robotic arm during welding is calculated, and the Newton-Raphson method is used to optimize the force matching parameters of the robotic arm to meet the stability requirements of the welding pool, and finally the robotic arm welding force matching data is obtained. Based on the robotic arm contact radius matching data, the robotic arm welding force matching data, and the welding heat energy limit value, the robotic arm welding heat output constraint regulation matching between different surface states of the weld direction is carried out. First, a welding heat output regulation matrix is constructed, with the welding heat energy limit value as the target variable and the robotic arm contact radius matching data and the robotic arm welding force matching data as the constraint conditions, and the quadratic programming optimization algorithm is used to calculate the welding heat input optimization coefficients in different weld directions.Combined with the real-time temperature monitoring data during the welding process, the Kalman filtering method is used to dynamically adjust the welding heat input, and a feedback control system is introduced to optimize the welding current, voltage and moving speed of the robotic arm in real time, so that the welding heat input conforms to the regulation range of the welding heat energy limit value, and finally the regulated data of the welding heat output constraint is obtained.

[0145] Step S4 includes the following steps:

[0146] Step S41: Conduct logical learning on the regulated data of the welding heat output constraint to obtain the learning data of the welding heat output constraint;

[0147] Step S42: Based on the random forest algorithm, construct an intelligent welding tracking model for free-form curve welds from the learning data of the welding heat output constraint to obtain the intelligent welding tracking model for welds;

[0148] Step S43: Send the intelligent welding tracking model for welds to the cloud platform to execute the intelligent tracking welding control of the robotic arm for free-form curve welds.

[0149] In the embodiments of the present invention, during the welding process, the welding heat input directly affects the forming quality and microstructure properties of the weld seam. Therefore, it is necessary to perform logical learning on the welding heat output constraint regulation data to optimize the heat input parameters. First, parameters such as welding heat input, arc voltage, current, welding speed, and wire feeding speed are extracted from the welding data acquisition system, and the sampling frequency is set to 1000 Hz to ensure the continuity and high resolution of the data. Secondly, based on the welding process constraint conditions, the data is normalized so that the numerical range of each parameter is standardized to the interval [0,1] for subsequent modeling. Then, a time series analysis method is used to calculate the correlation between different welding parameters, and parameter pairs with a Pearson correlation coefficient greater than 0.7 are selected as the main research objects. For example, in the experiment, it is found that the correlation between the welding current and the welding heat input reaches 0.85. Therefore, it is necessary to focus on analyzing the influence of current fluctuations on the stability of the heat input. Then, a logical regression model is used to learn the welding heat input constraint conditions, and the loss function is set as the cross-entropy loss to minimize the prediction error. Through iterative optimization, the optimal heat input control data under different welding parameters is obtained. Finally, the output welding heat output constraint learning data includes the welding process parameter range, the optimized value of the welding heat input, and the welding stability evaluation index, providing basic data for the construction of the subsequent welding tracking model. Based on the welding heat output constraint learning data, a random forest algorithm is used to construct an intelligent welding tracking model for free-form curve welds. First, using the data of welding heat input, arc voltage, current, welding speed, wire feeding speed, etc. extracted in the previous step, a training data set is constructed, and the size of the data set is set to 50,000 groups to ensure the generalization ability of the model. All data is divided according to the ratio of 80% for training and 20% for testing, and the data is normalized so that all feature values are normalized to the interval [0,1] to improve the training efficiency and stability of the model. Then, a random forest regression model is constructed, with the number of decision trees set to 100 and the maximum depth set to 10 to prevent overfitting, and the mean square error (MSE) is used as the loss function. The model prediction accuracy is optimized through the Bagging ensemble learning method. During the training process, the model calculates the influence of parameters such as welding heat input, current, and welding speed on the weld forming quality through multiple iterations, and establishes a mapping relationship between the input parameters and the weld forming based on historical welding data. After the training is completed, the performance of the model is evaluated using the test data set, and the root mean square error (RMSE) is calculated as the model accuracy evaluation index. The RMSE obtained from the experimental test is 0.015 mm, indicating that the model can accurately predict the influence of welding parameters on the weld quality. In the model optimization stage, the grid search method is used to adjust the hyperparameters, and the prediction accuracy of the model is further improved by optimizing parameters such as the number of decision trees, the maximum depth, and the minimum number of samples in the leaf nodes.In the final model testing stage, welding process parameters are input, and weld seam trajectory tracking adjustment data is output to ensure that the model can predict key control parameters such as the offset of the weld seam trajectory and the adjustment value of the welding speed in real time during the welding process, providing high-precision path planning data for subsequent intelligent weld seam tracking control. After the intelligent welding tracking model of the weld seam is constructed, it needs to be deployed to the cloud platform to achieve intelligent tracking welding control of the robotic arm for free-form curve weld seams. First, the trained model is converted into the TensorFlow Serving format to support real-time inference in the cloud. Then, the MQTT protocol is used to establish a communication channel between the model and the robotic arm control system, and the message transmission rate is set to 50 Hz to ensure the real-time nature of data transmission. Next, a RESTful API service is deployed in the cloud to support remote access to the weld path prediction data. The cloud server uses an NVIDIA A100 GPU for inference calculation, and the single inference time is controlled within 10 ms to meet the real-time control requirements of the welding process. On the robotic arm side, the welding control system requests the cloud prediction data once per second and dynamically adjusts the movement trajectory of the welding torch according to parameters such as the weld offset and the welding speed adjustment value output by the model. Finally, the correction of the welding path based on the intelligent tracking model is achieved, enabling the robotic arm to accurately track the welding trajectory of the free-form curve weld seam and improving the welding quality and weld seam consistency.

[0150] The present invention also provides an intelligent tracking system for a robotic arm to free-form curve weld seams, which is used to execute the intelligent tracking method of the robotic arm to free-form curve weld seams as described above. The intelligent tracking system for the robotic arm to free-form curve weld seams includes:

[0151] A weld surface state mapping module, which is used to deploy a laser profiler on the robotic arm, scan the curve weld seam profile of the free-form curve weld seam to obtain curve weld seam profile data; analyze the weld surface state of the curve weld seam profile data to obtain weld surface state data; perform geometric trend weld surface state mapping processing on the weld surface state data to obtain weld trend surface state mapping data;

[0152] A weld fracture risk calculation module, which is used to analyze the weld porosity distribution trend of the weld trend surface state mapping data to obtain weld porosity distribution trend data; perform simulation estimation of the welding compactness loss of the weld trend distribution based on the weld porosity distribution trend data to obtain weld trend compactness loss estimation data; perform weld fracture risk calculation according to the weld trend compactness loss estimation data to obtain weld fracture risk data;

[0153] A welding heat output constraint regulation module, which is used to perform mechanical arm welding heat output constraint regulation matching between different weld trend surface states on the weld trend surface state mapping data according to the weld fracture risk data to obtain welding heat output constraint regulation data;

[0154] An intelligent welding tracking model construction module is used to construct an intelligent welding tracking model for free-form curve welds based on the random forest algorithm for the welding heat output constraint regulation data, and obtain the intelligent welding tracking model for the welds; send the intelligent welding tracking model for the welds to the cloud platform to execute the intelligent tracking welding control of the free-form curve welds by the robotic arm.

[0155] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather to the widest scope consistent with the principles and novel features invented herein.

Claims

1. An intelligent tracking method for a free-curve weld by a robotic arm, characterized in that: The following steps are involved: Step S1: deploying a laser profiler on the robot arm, and scanning the free-curve weld profile to obtain the curve weld profile data; Perform weld surface state analysis on the curved weld profile data to obtain weld surface state data; Performing geometric trend weld surface state mapping processing on the weld surface state data to obtain weld trend surface state mapping data; Wherein, step S2 comprises: Step S21: analyzing the weld pore distribution trend on the weld surface state mapping data to obtain weld pore distribution trend data; Step S22: analyzing the trend change of the weld width / height on the curved weld geometry data to obtain the trend change data of the weld width / height; Step S23: performing a simulation estimation of the weld density loss of the weld direction distribution based on the weld porosity distribution direction data and the weld width / height trend change data to obtain weld density loss estimation data; Step S24: Calculate the weld distribution brittleness increment index for the weld direction density loss estimation data to obtain the weld distribution brittleness increment index; wherein step S24 includes: Step S241: performing stress instability concentration identification on the weld direction density loss estimation data to obtain the density loss stress instability concentration; Step S242: performing nonlinear simulation calculation of thermal plastic strain attenuation based on the density loss stress instability concentration and the weld direction density loss estimation data to obtain thermal plastic strain attenuation fitting data; Step S243: performing attenuation recursive median absolute deviation calculation on the thermal plastic strain attenuation fitting data to obtain the plastic recursive attenuation median absolute deviation; Step S244: performing attenuation decomposition constraint analysis on the thermal energy plastic strain attenuation fitting data according to the plastic recursive attenuation median absolute deviation and divide-and-conquer algorithm to generate thermal energy plastic strain attenuation decomposition data; Step S245: calculating the weld distribution brittleness increment index according to the thermal plastic strain attenuation decomposition data to obtain the weld distribution brittleness increment index; Step S25: performing weld fracture risk calculation on the weld trend surface state mapping data according to the weld trend density loss estimation data and the weld distribution brittleness increment index to obtain weld fracture risk data; Wherein, step S3 comprises: Step S31: normalizing the weld fracture risk data to obtain weld fracture risk normalized data; Step S32: matching the welding heat output constraint control of the robot arm between different weld direction surface states with the weld direction surface state mapping data according to the weld fracture risk normalization data and the weld direction density loss estimation data, to obtain welding heat output constraint control data; wherein, step S32 includes: Step S321: analyzing the welding heat energy limit values ​​between different weld direction surface states on the weld direction surface state mapping data according to the weld fracture risk normalization data and the weld direction density loss estimation data, and obtaining the welding heat energy limit values ​​between different weld direction surface states; Step S322: performing robot arm contact radius matching based on welding heat energy limit values ​​between different weld seam surface states to obtain robot arm contact radius matching data; Step S323: performing robot arm welding force matching according to the welding heat energy limit value and the robot arm contact radius matching data to obtain robot arm welding force matching data; Step S324: performing welding heat output constraint control matching of the robot arm between different weld direction surface states based on the robot arm contact radius matching data, the robot arm welding force matching data and the welding heat energy limit value to obtain welding heat output constraint control data; Step S4: Based on the random forest algorithm, the free curve weld intelligent welding tracking model is constructed for the welding heat output constraint control data to obtain the weld intelligent welding tracking model; the weld intelligent welding tracking model is sent to the cloud platform to execute the intelligent tracking welding control of the free curve weld by the robot arm.

2. The intelligent tracking method for free curve welds by a robot arm according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: deploying a laser profiler on the robot arm, and scanning the free-curve weld profile to obtain the curve weld profile data; Step S12: cleaning the curved weld profile data to obtain the curved weld profile cleaning data; Step S13: performing a curve weld geometric feature analysis on the curve weld profile cleaning data to generate curve weld geometric data; Step S14: performing weld surface state analysis on the curved weld profile cleaning data to obtain weld surface state data; Step S15: performing geometric trend weld surface state mapping processing on the weld surface state data according to the curved weld geometry data to obtain weld trend surface state mapping data.

3. The intelligent tracking method for free curve welds by a robot arm according to claim 1, characterized in that: Step S23 includes the following steps: Step S231: performing direction distribution density identification on the weld pore distribution direction data to obtain the weld pore distribution density; performing pore chain linear distribution identification on the weld pore distribution direction data based on the weld pore distribution density to obtain pore chain linear distribution data; Step S232: calculating the effective bearing volume mean difference of the weld width / height trend change data to obtain the effective bearing volume mean difference of the weld; Step S233: performing pore connectivity analysis on the pore chain linear distribution data to obtain pore distribution pore connectivity data; Step S234: performing deviation regression pore structure normal distribution sampling on the pore chain linear distribution data according to the mean difference of the effective bearing volume of the weld and the pore connectivity data of the pore distribution to obtain deviation regression pore structure normal sampling data; Step S235: Based on the least squares estimation algorithm, the deviation regression pore structure normal sampling data is used to simulate and estimate the welding density loss of the weld trend distribution to obtain the weld trend density loss estimation data.

4. The intelligent tracking method for free curve welds by a robot arm according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: performing logic learning on the welding heat output constraint control data to obtain welding heat output constraint learning data; Step S42: constructing a free curve weld intelligent welding tracking model for the welding heat output constraint learning data based on a random forest algorithm to obtain a weld intelligent welding tracking model; Step S43: Send the weld intelligent welding tracking model to the cloud platform to execute the intelligent tracking welding control of the free curve weld by the robot arm.

5. An intelligent tracking system for free-curve welds by a robotic arm, characterized in that: The method for intelligently tracking a free-curve weld by a robot arm according to claim 1 comprises: The weld surface state mapping module is used to deploy a laser profiler on the robot arm, and perform a curved weld profile scan on the free-curve weld to obtain curved weld profile data; perform weld surface state analysis on the curved weld profile data to obtain weld surface state data; perform geometric trend weld surface state mapping processing on the weld surface state data to obtain weld trend surface state mapping data; The weld fracture risk calculation module is used to analyze the weld porosity distribution trend based on the weld surface state mapping data to obtain the weld porosity distribution trend data; simulate and estimate the weld density loss based on the weld porosity distribution trend data to obtain the weld density loss estimation data; perform weld fracture risk calculation based on the weld density loss estimation data to obtain the weld fracture risk data; A welding heat output constraint control module is used to match the welding heat output constraint control of the robot arm between different weld direction surface states according to the weld fracture risk data and the weld direction surface state mapping data, so as to obtain the welding heat output constraint control data; The intelligent welding tracking model building module is used to build a free curve weld intelligent welding tracking model based on the welding heat output constraint control data based on the random forest algorithm to obtain the weld intelligent welding tracking model; the weld intelligent welding tracking model is sent to the cloud platform to execute the intelligent tracking welding control of the free curve weld by the robot arm.

Images