Welding quality detection method and system

Abstract

The present invention discloses a welding quality inspection method and system. The method comprises acquiring temperature data and speed data; analyzing the temperature data to obtain a temperature gradient; analyzing the speed data and the temperature gradient to obtain a cooling rate; determining a cooling rate change trend based on the cooling rate; analyzing the temperature gradient and the cooling rate change trend to obtain an abnormal area; reading the abnormal area to obtain a first parameter set; analyzing the temperature gradient and the cooling rate based on the first parameter set to obtain a gradient weight and a rate weight; optimizing the first parameter set based on the gradient weight and the rate weight to obtain a second parameter set; adjusting different workpieces based on the second parameter set to obtain a third process parameter set; training an algorithm model based on the third process parameter set; and inputting the characteristics and shape information of the workpiece to be inspected into the algorithm model for analysis to obtain an optimization solution. The method can optimize parameters to construct a model that meets the welding inspection requirements of workpieces of different materials and shapes.

Description

Technical Field

[0001] The present invention relates to the technical field of welding detection, and in particular to a welding quality detection method and system. Background Art

[0002] In welding quality inspection, defects with similar characteristics can be categorized by analyzing large amounts of welding image data. However, in actual production, the stability of the welding process is a complex technical issue. Fluctuations in the temperature field distribution and welding speed during welding can cause changes in the temperature gradient and cooling rate within the weld area, which in turn affects the solidification process and microstructure of the weld, ultimately leading to welding defects.

[0003] In addition, factors such as the welding position, welding angle, and wire feed speed during the welding process will also affect the welding quality. How to find the key factors affecting welding quality among the complex welding process parameters and improve the stability of the welding process by optimizing these key factors is a technical problem that needs to be solved urgently. In actual production, factors such as the material, size, and shape of the welding workpiece will also affect the welding quality. The welding performance of different materials varies greatly, and different welding process parameters and welding methods need to be adopted. Changes in the size and shape of the workpiece will also change the heat conduction and cooling conditions during the welding process, thereby affecting the welding quality. How to optimize the welding process parameters for different welding workpieces to ensure the consistency and stability of the welding quality is also a challenging technical problem.

[0004] One existing technique involves collecting a large amount of welding image data. Using image processing and machine learning algorithms, the process extracts and analyzes defect features within the images. For example, edge detection algorithms identify defect edges, and clustering algorithms group defects with similar characteristics. Simultaneously, various welding process parameters are recorded, such as temperature distribution, welding speed, welding position, angle, wire feed speed, and other information, as well as the material, size, and shape of the workpiece. Correlation analysis is then used to identify key factors significantly impacting welding quality.

[0005] However, the existing technology cannot adapt to the welding inspection requirements of workpieces of different materials and shapes because it does not optimize the process parameters and adjust them in combination with the workpiece material properties and shape information. Summary of the Invention

[0006] The present invention provides a welding quality detection method and system to solve the problem in the prior art that the process parameters are not optimized and adjusted in combination with the workpiece material properties and shape information, making it impossible to adapt to the welding detection needs of workpieces of different materials and shapes.

[0007] In a first aspect, in order to solve the above technical problems, the present invention provides a welding quality detection method, comprising:

[0008] Obtain temperature field distribution data and welding speed fluctuation data;

[0009] Performing gradient analysis based on the temperature field distribution data to obtain a temperature gradient;

[0010] Performing rate analysis based on the welding speed fluctuation data and the temperature gradient to obtain a cooling rate;

[0011] Judging based on the distribution of the cooling rate, obtaining a cooling rate change trend;

[0012] Performing characteristic analysis based on the temperature gradient and the cooling rate change trend to obtain an abnormal area;

[0013] Reading data from the abnormal area to obtain a first process parameter set;

[0014] Performing a correlation analysis on the temperature gradient and the cooling rate according to the first process parameter set to obtain a gradient influence weight and a rate influence weight;

[0015] Optimizing the first process parameter set according to the gradient influence weight and the rate influence weight to obtain a second process parameter set;

[0016] Adjusting the second process parameter set for different workpiece materials to obtain a third process parameter set;

[0017] Using the third process parameter set as training data, constructing an initial model based on a support vector machine algorithm and performing training to obtain a welding parameter prediction model;

[0018] The material characteristics and shape information of the workpiece to be tested are input into the welding parameter prediction model for data analysis to obtain the optimal welding parameter combination, thereby generating an optimization plan.

[0019] In an implementation manner of the first aspect, performing gradient analysis based on the temperature field distribution data to obtain the temperature gradient includes:

[0020] Performing preprocessing on the temperature field distribution data to obtain normalized temperature field distribution data;

[0021] According to the normalized temperature field distribution data, a finite element analysis method is used to construct a model of the welding process to obtain a heat conduction model;

[0022] The heat conduction equation is solved according to the heat conduction model to obtain the temperature gradient.

[0023] In an implementation of the first aspect, performing rate analysis based on the welding speed fluctuation data and the temperature gradient to obtain the cooling rate includes:

[0024] Preprocessing is performed on the welding speed fluctuation data to obtain normalized speed fluctuation data;

[0025] Performing frequency extraction based on the normalized speed fluctuation data to obtain frequency features;

[0026] Judging the frequency components of the speed fluctuation according to the frequency characteristics to obtain the dominant factor of the speed fluctuation;

[0027] Performing correlation analysis based on the temperature gradient and the speed fluctuation dominant factor to obtain a first mapping relationship model;

[0028] The rate is mapped according to the mapping relationship model to obtain the cooling rate.

[0029] In an implementation of the first aspect, performing characteristic analysis based on the temperature gradient and the cooling rate change trend to obtain the abnormal area includes:

[0030] Performing characteristic analysis based on the temperature gradient and cooling rate change trends to obtain dynamic characteristics; wherein the dynamic characteristics include the temperature gradient change rate and the cooling rate fluctuation amplitude;

[0031] A judgment is made based on the temperature gradient change rate and the cooling rate fluctuation amplitude; if the temperature gradient change rate exceeds a preset temperature gradient change rate threshold or the cooling rate fluctuation amplitude is greater than a predetermined fluctuation amplitude range, the welding area is determined to be an abnormal area.

[0032] In an implementation manner of the first aspect, performing a correlation analysis on the temperature gradient and the cooling rate according to the first process parameter set to obtain a gradient influence weight and a rate influence weight includes:

[0033] Performing parameter correlation analysis based on the first process parameter set to obtain a multi-parameter correlation model;

[0034] Calculating the correlation degree of each parameter in the first process parameter set according to the multi-parameter correlation model to obtain a parameter correlation degree;

[0035] A quantitative analysis is performed based on the parameter correlation to obtain the gradient influence weight and the rate influence weight.

[0036] In an implementation manner of the first aspect, optimizing the first process parameter set according to the gradient influence weight and the rate influence weight to obtain the second process parameter set includes:

[0037] Constructing a model based on the gradient influence weight and the rate influence weight to obtain a second mapping relationship model;

[0038] According to the second mapping relationship model, the first process parameter set is optimized with the minimization of the temperature gradient change rate and the minimization of the cooling rate fluctuation amplitude as optimization goals to obtain the second process parameter set.

[0039] In one implementation of the first aspect, the optimization scheme refers to predicting the optimal welding parameter combination of welding position, welding angle and wire feed speed for different workpieces based on the material, size and shape properties of the new welding workpiece.

[0040] In a second aspect, the present invention provides a welding quality detection system, comprising:

[0041] Data acquisition module, used to obtain temperature field distribution data and welding speed fluctuation data;

[0042] A gradient analysis module, configured to perform gradient analysis based on the temperature field distribution data to obtain a temperature gradient;

[0043] A rate analysis module, configured to perform rate analysis based on the welding speed fluctuation data and the temperature gradient to obtain a cooling rate;

[0044] A trend judgment module, configured to judge based on the distribution of the cooling rate and obtain a cooling rate change trend;

[0045] an abnormal region determination module, configured to perform characteristic analysis based on the temperature gradient and the cooling rate change trend to obtain an abnormal region;

[0046] a parameter reading module, configured to read data from the abnormal area to obtain a first process parameter set; wherein the first process parameter set includes: welding position, welding angle, and wire feeding speed;

[0047] a parameter analysis module, configured to perform a correlation analysis on the temperature gradient and the cooling rate according to the welding position, the welding angle, and the wire feed speed, to obtain a gradient influence weight and a rate influence weight;

[0048] a parameter optimization module, configured to optimize the first process parameter set according to the gradient influence weight and the rate influence weight to obtain a second process parameter set;

[0049] a parameter adjustment module, configured to adjust the second process parameter set for different workpiece materials to obtain a third process parameter set;

[0050] A model building module is used to use the third process parameter set as training data, build an initial model based on a support vector machine algorithm and perform training to obtain a welding parameter prediction model;

[0051] The output module is used to input the material characteristics and shape information of the workpiece to be detected into the welding parameter prediction model for data analysis to obtain the optimal welding parameter combination, thereby generating an optimization solution.

[0052] In an implementation manner of the second aspect, performing gradient analysis based on the temperature field distribution data to obtain the temperature gradient includes:

[0053] Performing preprocessing on the temperature field distribution data to obtain normalized temperature field distribution data;

[0054] According to the normalized temperature field distribution data, a finite element analysis method is used to construct a model of the welding process to obtain a heat conduction model;

[0055] The heat conduction equation is solved according to the heat conduction model to obtain the temperature gradient.

[0056] In an implementation of the second aspect, performing rate analysis based on the welding speed fluctuation data and the temperature gradient to obtain the cooling rate includes:

[0057] Preprocessing is performed on the welding speed fluctuation data to obtain normalized speed fluctuation data;

[0058] Performing frequency extraction based on the normalized speed fluctuation data to obtain frequency features;

[0059] Judging the frequency components of the speed fluctuation according to the frequency characteristics to obtain the dominant factor of the speed fluctuation;

[0060] Performing correlation analysis based on the temperature gradient and the speed fluctuation dominant factor to obtain a first mapping relationship model;

[0061] The rate is mapped according to the mapping relationship model to obtain the cooling rate.

[0062] In an implementation of the second aspect, performing characteristic analysis based on the temperature gradient and the cooling rate change trend to obtain the abnormal area includes:

[0063] Performing characteristic analysis based on the temperature gradient and cooling rate change trends to obtain dynamic characteristics; wherein the dynamic characteristics include the temperature gradient change rate and the cooling rate fluctuation amplitude;

[0064] A judgment is made based on the temperature gradient change rate and the cooling rate fluctuation amplitude; if the temperature gradient change rate exceeds a preset temperature gradient change rate threshold or the cooling rate fluctuation amplitude is greater than a predetermined fluctuation amplitude range, the welding area is determined to be an abnormal area.

[0065] In an implementation manner of the second aspect, performing a correlation analysis on the temperature gradient and the cooling rate according to the first process parameter set to obtain a gradient influence weight and a rate influence weight includes:

[0066] Performing parameter correlation analysis based on the first process parameter set to obtain a multi-parameter correlation model;

[0067] Calculating the correlation degree of each parameter in the first process parameter set according to the multi-parameter correlation model to obtain a parameter correlation degree;

[0068] A quantitative analysis is performed based on the parameter correlation to obtain the gradient influence weight and the rate influence weight.

[0069] In an implementation manner of the second aspect, optimizing the first process parameter set according to the gradient influence weight and the rate influence weight to obtain the second process parameter set includes:

[0070] Constructing a model based on the gradient influence weight and the rate influence weight to obtain a second mapping relationship model;

[0071] According to the second mapping relationship model, the first process parameter set is optimized with the minimization of the temperature gradient change rate and the minimization of the cooling rate fluctuation amplitude as optimization goals to obtain the second process parameter set.

[0072] In one implementation of the second aspect, the optimization scheme refers to predicting the optimal welding parameter combination of welding position, welding angle and wire feed speed for different workpieces based on the material, size and shape properties of the new welding workpiece.

[0073] In a third aspect, the present invention further provides an electronic device comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements any one of the above-mentioned welding quality detection methods when executing the computer program.

[0074] In a fourth aspect, the present invention also provides a computer-readable storage medium, which includes a stored computer program, wherein when the computer program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the welding quality detection methods described above.

[0075] Compared with the prior art, the present invention has the following beneficial effects:

[0076] The present invention discloses a welding quality detection method, comprising acquiring temperature field distribution data and welding speed fluctuation data; performing gradient analysis according to the temperature field distribution data to obtain a temperature gradient; performing rate analysis according to the welding speed fluctuation data and the temperature gradient to obtain a cooling rate; performing judgment according to the distribution of the cooling rate to obtain a cooling rate change trend; performing feature analysis according to the temperature gradient and the cooling rate change trend to obtain an abnormal area; performing data reading according to the abnormal area to obtain a first process parameter set; performing correlation analysis on the temperature gradient and the cooling rate according to the first process parameter set to obtain a gradient influence weight and a rate influence weight; optimizing the first process parameter set according to the gradient influence weight and the rate influence weight to obtain a second process parameter set; adjusting the second process parameter set for different workpiece materials to obtain a third process parameter set; training according to the third process parameter set to obtain a welding parameter prediction model; inputting the material characteristics and shape information of the workpiece to be detected into the welding parameter prediction model for data analysis to obtain an optimization solution. The present invention obtains temperature field distribution and velocity fluctuation data during the welding process, uses finite element analysis to calculate temperature gradients and cooling rates, and analyzes grain growth and solute redistribution characteristics during solidification. For abnormal areas, the present invention obtains real-time welding parameter data, establishes a multi-parameter correlation model, and determines the weight of each parameter's influence on the temperature gradient and cooling rate. Based on this, the present invention uses a genetic algorithm to optimize welding process parameters and adjusts them in combination with workpiece material properties and shape information. Finally, the present invention uses a support vector machine algorithm to predict the optimal welding parameter combination for different workpieces, generates an optimization plan, and outputs it to the welding control system. The present invention can effectively improve welding quality, reduce defects, adapt to the welding requirements of workpieces of different materials and shapes, and enhance the intelligence and adaptability of the welding process. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 1 is a flow chart of a welding quality detection method provided by a first embodiment of the present invention;

[0078] Figure 2 2 is a schematic structural diagram of a welding quality detection system provided by the second embodiment of the present invention. DETAILED DESCRIPTION

[0079] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0080] Reference Figure 1 The first embodiment of the present invention provides a welding quality detection method, comprising the following steps:

[0081] S1, obtain temperature field distribution data and welding speed fluctuation data;

[0082] S2, performing gradient analysis based on the temperature field distribution data to obtain a temperature gradient;

[0083] S3, performing rate analysis based on the welding speed fluctuation data and the temperature gradient to obtain a cooling rate;

[0084] S4, determining the cooling rate change trend based on the cooling rate distribution;

[0085] S5, performing characteristic analysis based on the temperature gradient and the cooling rate change trend to obtain an abnormal area;

[0086] S6, reading data from the abnormal area to obtain a first process parameter set;

[0087] S7, performing a correlation analysis on the temperature gradient and the cooling rate according to the first process parameter set to obtain a gradient influence weight and a rate influence weight;

[0088] S8, optimizing the first process parameter set according to the gradient influence weight and the rate influence weight to obtain a second process parameter set;

[0089] S9, adjusting the second process parameter set for different workpiece materials to obtain a third process parameter set;

[0090] S10, using the third process parameter set as training data, constructing an initial model based on a support vector machine algorithm and performing training to obtain a welding parameter prediction model;

[0091] S11, inputting the material characteristics and shape information of the workpiece to be detected into the welding parameter prediction model for data analysis to obtain the optimal welding parameter combination, thereby generating an optimization solution.

[0092] In step S1 , it is necessary to obtain temperature field distribution data and welding speed fluctuation data.

[0093] Alternatively, the temperature field distribution and speed fluctuation during the welding process are crucial to the welding quality. Acquiring this data through sensors can provide a deep understanding of the dynamic changes in the welding process.

[0094] In step S2, a gradient analysis is performed based on the temperature field distribution data to obtain a temperature gradient.

[0095] In the above step S2, the gradient analysis is performed based on the temperature field distribution data to obtain the temperature gradient, which specifically includes the following steps:

[0096] S21, performing preprocessing on the temperature field distribution data to obtain normalized temperature field distribution data;

[0097] S22, constructing a model of the welding process using a finite element analysis method based on the normalized temperature field distribution data to obtain a heat conduction model;

[0098] S23, solving the heat conduction equation according to the heat conduction model to obtain the temperature gradient.

[0099] It should be noted that in steps S21 to S23 above, temperature field distribution data and welding speed fluctuation data during the welding process are acquired via sensors. The acquired data is preprocessed to remove outliers and noise interference, thereby obtaining normalized temperature field distribution data and welding speed fluctuation data. Based on the preprocessed temperature field distribution data, a finite element analysis method is used to establish a heat conduction model for the welding process. By solving the heat conduction equation, the temperature gradient during the welding process is calculated. For example, data preprocessing is a key step. Median filtering can be used to remove outliers, and wavelet transform can be used to eliminate noise interference, thereby obtaining reliable temperature field and speed fluctuation data. Finite element analysis is an effective tool for establishing welding heat conduction models. By dividing the weldment into a finite number of elements and considering factors such as material properties and boundary conditions, the heat conduction equation can be solved. For example, for a steel plate welding process, it can be divided into thousands of tetrahedral elements, a heat source movement path is set, and the temperature gradient around the weld is calculated.

[0100] S3, performing rate analysis based on the welding speed fluctuation data and the temperature gradient to obtain a cooling rate.

[0101] In the above step S3, the rate analysis is performed based on the welding speed fluctuation data and the temperature gradient to obtain the cooling rate, which specifically includes the following steps:

[0102] S31, preprocessing the welding speed fluctuation data to obtain normalized speed fluctuation data;

[0103] S32, performing frequency extraction based on the normalized speed fluctuation data to obtain frequency features;

[0104] S33, judging the frequency component of the speed fluctuation according to the frequency characteristics to obtain the dominant factor of the speed fluctuation;

[0105] S34, performing correlation analysis based on the temperature gradient and the speed fluctuation dominant factor to obtain a first mapping relationship model;

[0106] S35, mapping the rate according to the mapping relationship model to obtain a cooling rate.

[0107] It should be noted that in steps S31 to S35 above, based on the preprocessed welding speed fluctuation data, a time-frequency analysis method is used to extract the frequency characteristics of the speed fluctuation, identify the primary frequency components of the speed fluctuation, and determine the dominant factors of the speed fluctuation. The calculated temperature gradient distribution is then correlated with the dominant factors of the speed fluctuation. A support vector machine algorithm is then used to establish a mapping model between the temperature gradient and speed fluctuation, enabling mutual prediction of the temperature gradient and speed fluctuation. Based on the established heat conduction model, the cooling rate of the welding process is calculated by solving the inverse problem of the heat conduction equation. For example, a short-time Fourier transform (SFT) can be used for time-frequency analysis of welding speed fluctuation. By performing a time-frequency transformation on the speed data, the primary frequency components can be extracted. For example, the analysis results may reveal the presence of two primary frequencies, 0.5 Hz and 2 Hz, corresponding to the effects of mechanical vibration and power supply fluctuation, respectively. A support vector machine algorithm can be used to establish a mapping between the temperature gradient and speed fluctuation. By selecting an appropriate kernel function, such as a radial basis function, the nonlinear relationship between the temperature gradient and speed fluctuation can be captured. This mapping model can predict the impact of speed fluctuation on the temperature field or infer speed fluctuation based on temperature gradient changes. Solving the inverse problem of the heat conduction equation yields the cooling rate distribution. This requires the use of an iterative algorithm, such as the conjugate gradient method, to infer the heat source parameters and boundary conditions from the known temperature field distribution. For example, for a thick plate welding process, the time-varying cooling rate curve along the weld centerline can be calculated to determine whether there is a risk of hardening due to rapid cooling. Finally, by integrating temperature gradients, velocity fluctuations, and cooling rate information, a comprehensive predictive model can be constructed. Convolutional neural networks are ideal for processing this multidimensional data. By designing multiple convolutional and pooling layers, features can be automatically extracted, enabling dynamic prediction of the welding process. For example, given a temperature field image and velocity data at the current moment, the model can predict the temperature distribution and cooling rate at the next moment. This comprehensive analysis method provides a comprehensive view of the welding process, helping to optimize welding parameters and improve welding quality and efficiency. Through real-time monitoring and prediction, welding processes can be adjusted promptly to avoid defects and ensure the mechanical properties and reliability of the welded joint.

[0108] S4, judging based on the distribution of the cooling rate to obtain a cooling rate change trend.

[0109] For example, the cooling rate is calculated by solving the inverse problem of the heat conduction equation. This process usually requires the use of iterative algorithms such as the conjugate gradient method, which uses the known temperature field distribution to infer the heat source parameters and boundary conditions, and then determine the cooling rate change trend.

[0110] S5. Perform characteristic analysis based on the temperature gradient and the cooling rate change trend to obtain an abnormal area.

[0111] It should be noted that, in the above step S5, the characteristic analysis based on the temperature gradient and the cooling rate change trend to obtain the abnormal area specifically includes the following steps:

[0112] S51, performing characteristic analysis based on the temperature gradient and cooling rate change trends to obtain dynamic characteristics; wherein the dynamic characteristics include the temperature gradient change rate and the cooling rate fluctuation amplitude;

[0113] S52, making a judgment based on the temperature gradient change rate and the cooling rate fluctuation amplitude; if the temperature gradient change rate exceeds a preset temperature gradient change rate threshold or the cooling rate fluctuation amplitude is greater than a predetermined fluctuation amplitude range, determining that the welding area is an abnormal area at this time.

[0114] Optionally, real-time temperature data and cooling rate data are acquired during the weld solidification process, and data is collected and recorded according to a preset sampling frequency. The collected temperature data is filtered to remove noise interference and obtain a smooth temperature curve. The temperature gradient change rate is calculated based on the temperature curve to determine whether the temperature gradient exceeds a preset threshold range. The collected cooling rate data is statistically analyzed to calculate the fluctuation amplitude and frequency of the cooling rate. If the fluctuation amplitude exceeds the preset range or the frequency is abnormal, it is determined that there are abnormal fluctuations in the cooling rate. A machine learning algorithm is used to establish a correlation model between the temperature gradient, cooling rate, grain growth, and solute redistribution. A support vector machine or decision tree algorithm is used to train historical data to obtain model parameters. The real-time collected temperature gradient change rate and cooling rate fluctuation data are input into the trained machine learning model to predict the dynamic change trends of grain size and solute distribution. Based on the grain growth and solute redistribution prediction results, it is determined whether there are abnormalities in the solidification process. If the predicted grain size or solute distribution exceeds the preset range, the area is determined to be a solidification abnormality area.

[0115] For example, a real-time data acquisition system can record temperature changes at a preset sampling frequency (e.g., 10 times per second). The collected raw data often contains noise, requiring filtering to obtain a smooth temperature curve. Common filtering methods include moving average and low-pass filtering, which can effectively remove high-frequency noise. The temperature gradient rate reflects the intensity of weld cooling and has a significant impact on grain growth. For example, a temperature gradient rate exceeding 100°C / s·mm can lead to coarsening of grains. A set threshold range (e.g., 50-80°C / s·mm) can promptly detect anomalies. Fluctuations in cooling rate can affect the uniformity of the solidified structure. Statistical analysis can calculate the standard deviation and main frequency components of the cooling rate. An anomaly is identified if the standard deviation exceeds a preset value (e.g., 20%) or if abnormal high-frequency fluctuations (e.g., greater than 1 Hz) occur. Such anomalies arise from unstable coolant flow or external interference. Machine learning algorithms can model the relationship between temperature gradient, cooling rate, and microstructure. Support vector machines are suitable for handling nonlinear relationships and can map input features into a high-dimensional space using kernel functions (e.g., radial basis functions). The decision tree algorithm can intuitively reflect the influence weight of each factor. Model training requires a large amount of historical data, including temperature curves, cooling rates, and corresponding metallographic analysis results under different process parameters. Using the trained model, grain size and solute distribution can be predicted in real time. For example, when the temperature gradient change rate is 60°C / s·mm and the cooling rate is 15°C / s, the model may predict an average grain size of 50μm and a primary α phase volume fraction of 60%. If the predicted results deviate from the target values ​​(e.g., grain size >100μm or primary α phase <40%), it is determined to be a solidification anomaly zone.

[0116] S6: Read data based on the abnormal area to obtain a first process parameter set.

[0117] For example, the first process parameter set may include real-time data of the welding process, such as welding position, welding angle, and wire feed speed.

[0118] S7: Perform a correlation analysis on the temperature gradient and the cooling rate according to the first process parameter set to obtain a gradient influence weight and a rate influence weight.

[0119] In the above step S7, the correlation analysis of the temperature gradient and the cooling rate is performed according to the first process parameter set to obtain the gradient influence weight and the rate influence weight, which specifically includes the following steps:

[0120] S71, performing parameter correlation analysis based on the first process parameter set to obtain a multi-parameter correlation model;

[0121] S72, calculating the correlation degree of each parameter in the first process parameter set according to the multi-parameter correlation model to obtain a parameter correlation degree;

[0122] S73, performing quantitative analysis based on the parameter correlation to obtain a gradient influence weight and a rate influence weight.

[0123] It should be noted that in steps S71 to S73 above, real-time data on welding position, welding angle, and wire feed speed is acquired via sensors and transmitted to the data processing module. This acquired real-time data undergoes preprocessing, including data cleaning and normalization, to improve data quality and usability. Based on this preprocessed data, a multi-parameter correlation model is constructed using grey correlation analysis to quantify the correlation between each parameter and the temperature gradient and cooling rate. The correlation calculation results are used to determine the influence weights of the welding position, welding angle, and wire feed speed on the gradient and rate.

[0124] For example, during the welding process, real-time data acquisition and processing are crucial for controlling weld quality. The sensor system can include laser displacement sensors, angle sensors, and encoders, respectively, for measuring weld position, weld angle, and wire feed speed. These sensors convert their signals into digital quantities using a high-speed data acquisition card, which is then transmitted to the data processing module via industrial Ethernet. Data preprocessing is a key step in improving data quality. Data cleaning can remove outliers, such as sudden data jumps caused by electromagnetic interference. For example, weld angle data may exhibit large, transient fluctuations, which may be caused by sensor failure or external interference. By setting appropriate thresholds, these outliers can be identified and removed. Normalization can unify parameters of different dimensions to the same scale, facilitating subsequent analysis. For example, weld position may be measured in millimeters, while wire feed speed may be measured in meters per minute. Normalization can map these to a range of 0–1. Grey relational analysis is an effective multi-parameter correlation modeling method. It quantifies the strength of the correlation between parameters by calculating the grey relational degree between a reference sequence (such as temperature gradient) and a comparison sequence (such as weld position). For example, consider a set of welding position data and corresponding temperature gradient data. First, the two data sets are dimensionlessly processed, and then the difference between the two series at each time point is calculated. Smaller differences indicate a higher degree of correlation. By setting a resolution factor (typically 0.5), the final correlation can be determined. Determining the influence weights is crucial for optimizing welding parameters. For example, if the analysis results show a correlation of 0.85 between welding angle and temperature gradient, and 0.65 between welding angle and cooling rate, this means that adjusting welding angle has a more significant effect on controlling the temperature gradient.

[0125] S8. Optimize the first process parameter set according to the gradient influence weight and the rate influence weight to obtain a second process parameter set.

[0126] In the above step S8, the first process parameter set is optimized according to the gradient influence weight and the rate influence weight to obtain the second process parameter set, which specifically further includes the following steps:

[0127] S81, constructing a model based on the gradient influence weight and the rate influence weight to obtain a second mapping relationship model;

[0128] S82 , optimizing the first process parameter set according to the second mapping relationship model with the optimization objectives of minimizing the temperature gradient change rate and the cooling rate fluctuation amplitude to obtain a second process parameter set.

[0129] It should be noted that in steps S81 to S82, a mapping relationship model between welding process parameters and the temperature gradient change rate and cooling rate fluctuation amplitude is established based on the influence weights. A genetic algorithm is used to encode a first set of process parameters, including welding position, welding angle, and wire feed speed, with the optimization objectives of minimizing the temperature gradient change rate and cooling rate fluctuation amplitude. This encodes the initial population. Based on the fitness function, the fitness value of each individual is calculated. A higher fitness value indicates a lower temperature gradient change rate and cooling rate fluctuation amplitude for that process parameter combination. Through genetic operations such as selection, crossover, and mutation, the process parameter combination is continuously optimized until the termination condition is met, resulting in the optimal welding process parameter set. The optimal welding process parameter set is then decoded to obtain a second process parameter set consisting of the corresponding welding position, welding angle, and wire feed speed parameter values.

[0130] S9: Adjust the second process parameter set for different workpiece materials to obtain a third process parameter set.

[0131] For example, according to the type of workpiece material, the thermal conductivity, specific heat capacity and other thermophysical parameters of the material are obtained from the material thermophysical property database. Using three-dimensional modeling software, a three-dimensional geometric model of the workpiece is established according to the size and shape information of the workpiece. The obtained material thermophysical parameters are input into the finite element analysis software, and the three-dimensional geometric model of the workpiece is imported. In the finite element analysis software, the welding heat source model and boundary conditions are set, and the second process parameters are initialized. Through finite element analysis, the transient temperature field distribution during the welding process is simulated to obtain the temperature field cloud map under different welding process parameters. According to the temperature field distribution range and peak temperature, it is judged whether the current welding process parameters meet the requirements. If not, the welding speed and heat input are adjusted, and the finite element analysis is performed again. Through repeated iterative optimization, the appropriate welding speed range and temperature field distribution range are finally determined to obtain the optimized third process parameter set.

[0132] S10, using the third process parameter set as training data, constructing an initial model based on a support vector machine algorithm and performing training to obtain a welding parameter prediction model; inputting the material characteristics and shape information of the workpiece to be detected into the welding parameter prediction model for data analysis to obtain an optimal welding parameter combination, thereby generating an optimization solution.

[0133] In one achievable manner, the optimization solution refers to predicting the optimal welding parameter combination of welding position, welding angle and wire feed speed for different workpieces based on the material, size and shape properties of the new welding workpiece.

[0134] For example, a third set of process parameters for workpieces of varying materials, sizes, and shapes is obtained as a training dataset for the support vector machine algorithm. This training dataset is preprocessed to extract key parameters related to welding quality as feature vectors. These feature vectors are then normalized, normalizing features of varying dimensions to the [0, 1] range. For example, wire feed speed (unit: m / min) and thermal conductivity (unit: W / m·K) need to be scaled uniformly. An importance analysis is then performed using the Pearson correlation coefficient or random forest features to remove features with low correlation to the output label. The normalized data is then divided into training and test sets in a ratio (e.g., 7:3).

[0135] During the support vector machine model construction process, the kernel function is selected based on the degree of nonlinearity of the data features. For example, the radial basis function (RBF) is suitable for nonlinear relationships and requires optimization of the hyperparameters C (penalty coefficient) and gamma (kernel function bandwidth). Grid search or Bayesian optimization is used to iterate over hyperparameter combinations, and cross-validation (such as 5-fold cross-validation) is used to evaluate model performance. The parameter combination with the lowest mean squared error (MSE) of the validation set is selected. During model training, the preprocessed training set feature matrix (X_train) and label vector (y_train) are input. The trained SVM regression model can predict welding parameters based on the input features. Through the SVM model, the multi-dimensional process parameters in the third process parameter set are effectively mapped to material and shape characteristics, achieving accurate prediction of welding parameters. This method significantly improves the adaptability to different workpieces, solves the problem of disconnection between process parameters and workpiece characteristics in previous technologies, and ultimately improves the stability of welding quality.

[0136] For example, a new welded workpiece is classified into corresponding workpiece attribute combination categories based on its material, size, and shape properties. The key welding parameters of the new workpiece are extracted and input as feature vectors into the corresponding category's welding parameter prediction model to predict the optimal welding parameter combination. For example, the workpiece's physical properties are obtained from a material database or sensor, including thermal conductivity (unit: W / m·K), specific heat capacity (J / kg·K), melting point (°C), and material type code (e.g., mild steel = 1, aluminum alloy = 2). Shape parameters are extracted through 3D modeling or image processing techniques, including geometric complexity (calculated based on the surface area to volume ratio, with a larger ratio indicating a more complex shape); key dimensions such as thickness, radius of curvature, and weld length; and symmetry (normalized to a 0-1 scale using principal component analysis (PCA) to determine the degree of shape symmetry).

[0137] Taking a welding workpiece made of 8mm thick aluminum alloy (material code = 2) and a curved, irregularly shaped part (geometric complexity = 0.85, radius of curvature = 50mm) as an example, the parameters of different dimensions are normalized to the interval [0, 1]. The material properties and shape parameters are combined into an input vector: [0.2, 0.8, 0.6, 0.85, 0.3]. The features in this vector correspond to the normalized thermal conductivity, melting point, thickness, geometric complexity, and radius of curvature, respectively. The preprocessed feature vector is then fed into a trained welding parameter prediction model. The model uses a kernel function (such as RBF) to map the input to a high-dimensional space and find the optimal hyperplane for regression prediction. The model output is a multi-target welding parameter combination, including the welding position coordinates (x, y, z), welding angle (°), and wire feed speed (m / min).

[0138] In summary, the present invention discloses a welding quality detection method, including obtaining temperature field distribution data and welding speed fluctuation data; performing gradient analysis according to the temperature field distribution data to obtain a temperature gradient; performing rate analysis according to the welding speed fluctuation data and the temperature gradient to obtain a cooling rate; making a judgment according to the distribution of the cooling rate to obtain a cooling rate change trend; performing feature analysis according to the temperature gradient and the cooling rate change trend to obtain an abnormal area; reading data according to the abnormal area to obtain a first process parameter set; performing a correlation analysis on the temperature gradient and the cooling rate according to the first process parameter set to obtain a gradient influence weight and a rate influence weight; optimizing the first process parameter set according to the gradient influence weight and the rate influence weight to obtain a second process parameter set; adjusting the second process parameter set for different workpiece materials to obtain a third process parameter set; training according to the third process parameter set to obtain a welding parameter prediction model; inputting the material characteristics and shape information of the workpiece to be detected into the welding parameter prediction model for data analysis to obtain an optimization solution. The present invention obtains temperature field distribution and velocity fluctuation data during the welding process, uses finite element analysis to calculate temperature gradients and cooling rates, and analyzes grain growth and solute redistribution characteristics during solidification. For abnormal areas, the present invention obtains real-time welding parameter data, establishes a multi-parameter correlation model, and determines the weight of each parameter's influence on the temperature gradient and cooling rate. Based on this, the present invention uses a genetic algorithm to optimize welding process parameters and adjusts them in combination with workpiece material properties and shape information. Finally, the present invention uses a support vector machine algorithm to predict the optimal welding parameter combination for different workpieces, generates an optimization plan, and outputs it to the welding control system. The present invention can effectively improve welding quality, reduce defects, adapt to the welding requirements of workpieces of different materials and shapes, and enhance the intelligence and adaptability of the welding process.

[0139] Reference Figure 2 A second embodiment of the present invention provides a welding quality detection system, comprising:

[0140] Data acquisition module 101, used to acquire temperature field distribution data and welding speed fluctuation data;

[0141] A gradient analysis module 102 is configured to perform gradient analysis based on the temperature field distribution data to obtain a temperature gradient;

[0142] A rate analysis module 103 is configured to perform rate analysis based on the welding speed fluctuation data and the temperature gradient to obtain a cooling rate;

[0143] A trend judgment module 104 is used to judge based on the distribution of the cooling rate to obtain a cooling rate change trend;

[0144] An abnormal region determination module 105 is configured to perform feature analysis based on the temperature gradient and the cooling rate change trend to obtain an abnormal region;

[0145] A parameter reading module 106 is configured to read data from the abnormal area to obtain a first process parameter set, wherein the first process parameter set includes: welding position, welding angle, and wire feeding speed;

[0146] a parameter analysis module 107 for performing a correlation analysis on the temperature gradient and the cooling rate according to the welding position, the welding angle, and the wire feeding speed, to obtain a gradient influence weight and a rate influence weight;

[0147] a parameter optimization module 108, configured to optimize the first process parameter set according to the gradient influence weight and the rate influence weight to obtain a second process parameter set;

[0148] a parameter adjustment module 109, configured to adjust the second process parameter set for different workpiece materials to obtain a third process parameter set;

[0149] A model building module 110 is configured to use the third process parameter set as training data, build an initial model based on a support vector machine algorithm, and perform training to obtain a welding parameter prediction model;

[0150] The output module 111 is used to input the material characteristics and shape information of the workpiece to be detected into the welding parameter prediction model for data analysis to obtain the optimal welding parameter combination, thereby generating an optimization solution.

[0151] It should be noted that a welding quality inspection system provided in an embodiment of the present invention is used to execute all process steps of a welding quality inspection method in the above embodiment. The working principles and beneficial effects of the two correspond one to one, and thus will not be described in detail.

[0152] An embodiment of the present invention further provides an electronic device. The electronic device includes: a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a welding quality detection program. When the processor executes the computer program, the steps in the above-mentioned various welding quality detection method embodiments are implemented, such as Figure 1 Alternatively, when the processor executes the computer program, the functions of the modules / units in the above-mentioned device embodiments are realized, such as the model building module.

[0153] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, and the instruction segments are used to describe the execution process of the computer program in the electronic device.

[0154] The electronic device may be a computing device such as a desktop computer, notebook, PDA, or smart tablet. The electronic device may include, but is not limited to, a processor and memory. Those skilled in the art will appreciate that the aforementioned components are merely examples of electronic devices and do not constitute a limitation of the electronic device. The electronic device may include more or fewer components than those described above, or a combination of certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, and the like.

[0155] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor. The processor is the control center of the electronic device and connects various parts of the entire electronic device using various interfaces and lines.

[0156] The memory can be used to store the computer programs and / or modules. The processor implements the various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and accessing the data stored in the memory. The memory may primarily include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function (such as a sound playback function or an image playback function); the data storage area may store data generated based on the use of the mobile phone (such as audio data, a phone book, etc.). Furthermore, the memory may include high-speed random access memory and non-volatile memory, such as a hard disk, internal memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0157] If the module / unit integrated into the electronic device is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention can implement all or part of the process steps in the above-mentioned method embodiments by using a computer program to instruct the relevant hardware. The computer program can be stored in a computer-readable storage medium. When executed by a processor, the computer program can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal, and software distribution medium. It should be noted that the content of the computer-readable medium can be appropriately increased or decreased based on the requirements of legislation and patent practice in a jurisdiction. For example, in some jurisdictions, based on legislation and patent practice, computer-readable media does not include electric carrier signals and telecommunication signals.

[0158] It should be noted that the device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed across multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment. In addition, in the drawings of the device embodiments provided by the present invention, the connection relationship between the modules indicates that there is a communication connection between them, which may be specifically implemented as one or more communication buses or signal lines. A person of ordinary skill in the art can understand and implement the present invention without inventive effort.

[0159] The specific embodiments described above further illustrate the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A welding quality detection method, characterized in that: Executed by a computer, including: Obtain temperature field distribution data and welding speed fluctuation data; Performing gradient analysis based on the temperature field distribution data to obtain a temperature gradient; Performing rate analysis based on the welding speed fluctuation data and the temperature gradient to obtain a cooling rate; Judging based on the distribution of the cooling rate, obtaining a cooling rate change trend; Performing characteristic analysis based on the temperature gradient and the cooling rate change trend to obtain an abnormal area; Reading data from the abnormal area to obtain a first process parameter set; Performing a correlation analysis on the temperature gradient and the cooling rate according to the first process parameter set to obtain a gradient influence weight and a rate influence weight; Optimizing the first process parameter set according to the gradient influence weight and the rate influence weight to obtain a second process parameter set; Adjusting the second process parameter set for different workpiece materials to obtain a third process parameter set; Using the third process parameter set as training data, constructing an initial model based on a support vector machine algorithm and performing training to obtain a welding parameter prediction model; Inputting the material characteristics and shape information of the workpiece to be tested into the welding parameter prediction model for data analysis to obtain the optimal welding parameter combination, thereby generating an optimization solution; The step of optimizing the first process parameter set according to the gradient influence weight and the rate influence weight to obtain the second process parameter set includes: Constructing a model based on the gradient influence weight and the rate influence weight to obtain a second mapping relationship model; According to the second mapping relationship model, the first process parameter set is optimized with minimizing the temperature gradient change rate and the cooling rate fluctuation amplitude as optimization objectives to obtain a second process parameter set; The second mapping relationship model is a mapping relationship model between welding process parameters and temperature gradient change rate and cooling rate fluctuation amplitude.

2. The welding quality detection method according to claim 1, characterized in that: The step of performing gradient analysis based on the temperature field distribution data to obtain a temperature gradient includes: Performing preprocessing on the temperature field distribution data to obtain normalized temperature field distribution data; According to the normalized temperature field distribution data, a finite element analysis method is used to construct a model of the welding process to obtain a heat conduction model; The heat conduction equation is solved according to the heat conduction model to obtain the temperature gradient.

3. The welding quality detection method according to claim 1, characterized in that: The performing rate analysis based on the welding speed fluctuation data and the temperature gradient to obtain the cooling rate includes: Preprocessing is performed on the welding speed fluctuation data to obtain normalized speed fluctuation data; Performing frequency extraction based on the normalized speed fluctuation data to obtain frequency features; Judging the frequency components of the speed fluctuation according to the frequency characteristics to obtain the dominant factor of the speed fluctuation; Performing correlation analysis based on the temperature gradient and the speed fluctuation dominant factor to obtain a first mapping relationship model; The rate is mapped according to the first mapping relationship model to obtain a cooling rate.

4. The welding quality detection method according to claim 1, characterized in that: The characteristic analysis based on the temperature gradient and the cooling rate change trend is performed to obtain the abnormal area, including: Performing characteristic analysis based on the temperature gradient and cooling rate change trends to obtain dynamic characteristics; wherein the dynamic characteristics include the temperature gradient change rate and the cooling rate fluctuation amplitude; A judgment is made based on the temperature gradient change rate and the cooling rate fluctuation amplitude; if the temperature gradient change rate exceeds a preset temperature gradient change rate threshold or the cooling rate fluctuation amplitude is greater than a predetermined fluctuation amplitude range, the welding area is determined to be an abnormal area.

5. The welding quality detection method according to claim 1, characterized in that: The performing correlation analysis on the temperature gradient and the cooling rate according to the first process parameter set to obtain a gradient influence weight and a rate influence weight includes: Performing parameter correlation analysis based on the first process parameter set to obtain a multi-parameter correlation model; Calculating the correlation degree of each parameter in the first process parameter set according to the multi-parameter correlation model to obtain a parameter correlation degree; A quantitative analysis is performed based on the parameter correlation to obtain the gradient influence weight and the rate influence weight.

6. The welding quality detection method according to claim 1, characterized in that: The optimization scheme refers to predicting the optimal welding parameter combination of welding position, welding angle and wire feed speed for different workpieces based on the material, size and shape properties of the new welding workpiece.

7. A welding quality detection system, characterized in that: A welding quality detection method for implementing any one of claims 1 to 6, comprising: Data acquisition module, used to obtain temperature field distribution data and welding speed fluctuation data; A gradient analysis module, configured to perform gradient analysis based on the temperature field distribution data to obtain a temperature gradient; A rate analysis module, configured to perform rate analysis based on the welding speed fluctuation data and the temperature gradient to obtain a cooling rate; A trend judgment module, configured to judge based on the distribution of the cooling rate and obtain a cooling rate change trend; an abnormal region determination module, configured to perform characteristic analysis based on the temperature gradient and the cooling rate change trend to obtain an abnormal region; a parameter reading module, configured to read data from the abnormal area to obtain a first process parameter set; wherein the first process parameter set includes: welding position, welding angle, and wire feeding speed; a parameter analysis module, configured to perform a correlation analysis on the temperature gradient and the cooling rate according to the welding position, the welding angle, and the wire feed speed, to obtain a gradient influence weight and a rate influence weight; a parameter optimization module, configured to optimize the first process parameter set according to the gradient influence weight and the rate influence weight to obtain a second process parameter set; a parameter adjustment module, configured to adjust the second process parameter set for different workpiece materials to obtain a third process parameter set; a model building module, configured to use the third process parameter set as training data, build an initial model based on a support vector machine algorithm, and perform training to obtain a welding parameter prediction model; The output module is used to input the material characteristics and shape information of the workpiece to be detected into the welding parameter prediction model for data analysis to obtain the optimal welding parameter combination, thereby generating an optimization solution.

8. An electronic device, characterized in that: The device comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the welding quality detection method according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the welding quality detection method according to any one of claims 1 to 6.

Images