Weld joint strength prediction method and welding parameter optimization method

Abstract

The present application relates to the technical field of welding, and discloses a welding joint strength prediction method and a welding parameter optimization method, the welding joint strength prediction method comprising: collecting real welding parameters and corresponding welding joint strength data; establishing a microstructure model according to the welding parameters, performing thermal-mechanical coupling simulation, and obtaining corresponding welding joint strength simulation values; establishing a relationship model between the welding joint strength simulation values and the welding parameters; and predicting corresponding welding joint strength based on the relationship model and new welding parameters. The optimization method comprises: setting a value range of the welding joint strength; substituting all values of each welding parameter into the welding joint strength prediction method to obtain corresponding welding joint prediction values; and selecting welding parameters corresponding to the welding joint prediction values that meet the value range, which are the optimized welding parameters. The method can improve the prediction accuracy of the welding joint strength and obtain welding parameters corresponding to the required welding joint strength.

Description

Technical Field

[0001] This invention relates to the field of welding technology, specifically to a method for predicting the strength of welded joints and a method for optimizing welding parameters. Background Technology

[0002] With the rapid development of my country's economy, the welding industry has also experienced rapid growth. Welding technology is an indispensable process for connecting metal structures in various engineering constructions, and the mechanical properties of welded joints directly determine structural safety. However, the welding process involves thermo-mechanical coupling and complex microstructure evolution, leading to large fluctuations in the mechanical properties of the joints and poor controllability of product quality. Currently, the optimization design of welding process parameters usually relies on experience, which is time-consuming and costly. Therefore, researching methods for predicting the mechanical properties of welded joints is of great significance for optimizing welding structure processes and improving product performance.

[0003] The mechanical properties of welded joints are significantly affected by their microstructure, primarily including grain size, distribution of hard phases, and distribution of soft phases. Currently, traditional methods for analyzing the microstructure of joints mainly employ experimental testing techniques, including fracture surface analysis, electrolytic corrosion methods, and X-ray diffraction. These methods are limited by factors such as experimental equipment, cost, and manpower, and are only suitable for small-scale experimental research and process verification.

[0004] Machine learning technology offers an effective approach to solving this problem. Methods such as multiple regression analysis, decision trees, neural networks, and support vector machines can be used to predict microstructure parameters (such as grain size and hard phase volume fraction). However, these methods all establish a direct relationship between output and input using statistical methods, neglecting the physical processes of welding. The accuracy of coupling welding process parameters, microstructure characteristics, and the mechanical properties of welded joints in prediction is limited. Summary of the Invention

[0005] This invention was made to solve the above-mentioned technical problems, and one of its objectives is to provide a method for predicting the strength of welded joints, which can improve the accuracy of welded joint strength prediction.

[0006] Another objective of this invention is to provide a welding parameter optimization method that can obtain the welding parameters corresponding to the required weld joint strength.

[0007] According to one embodiment of the present invention, a method for predicting the strength of a welded joint is provided, comprising the following steps:

[0008] S1. Collect actual welding parameters and corresponding weld joint strength data;

[0009] S2. Based on the welding parameters, establish a microstructure model, perform thermo-mechanical coupling simulation, and obtain the corresponding simulated values ​​of the weld joint strength;

[0010] S3. Establish a relationship model between the simulated strength value of the welded joint and the welding parameters;

[0011] S4. Based on the relationship model and the new welding parameters, predict the strength of the corresponding welded joint.

[0012] According to one embodiment of the present invention, a method for optimizing welding parameters is provided, comprising the following steps:

[0013] Define the range of values ​​for the strength of the welded joint;

[0014] Substitute all the values ​​of each welding parameter into the weld joint strength prediction method described above to obtain the corresponding weld joint prediction value.

[0015] The welding parameters selected when the predicted value of the weld joint meets the specified range are the optimized welding parameters.

[0016] Based on the above description and practical experience, the weld joint strength prediction method of this invention, combined with the establishment of a microstructure model and thermo-mechanical coupling simulation, achieves refined simulation of the welding process. The relationship model between welding parameters and weld joint strength is established based on a statistical high-order multi-scale analysis algorithm, which can accurately predict the weld joint strength corresponding to different welding parameters, providing a theoretical basis for the refined design and optimization of the welding process. In situations where experimental data is insufficient, a method for predicting weld joint strength is provided, and it can be used for the optimized design of welding parameters, improving the quality and efficiency of welding production. Attached Figure Description

[0017] Figure 1 This is a flowchart of a welded joint strength prediction method according to one embodiment of the present invention. Detailed Implementation

[0018] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this disclosure will be more comprehensive and complete, and will fully convey the concept of exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0019] This embodiment discloses a method for predicting the strength of welded joints, such as... Figure 1 As shown, the method for predicting the strength of welded joints mainly includes the following four steps:

[0020] Step S1: Collect actual welding parameters and corresponding weld joint strength data;

[0021] Step S2: Based on the welding parameters, establish a microstructure model, perform thermo-mechanical coupling simulation, and obtain the corresponding simulated value of the weld joint strength.

[0022] Step S3: Establish a relationship model between the simulated strength value of the welded joint and the welding parameters;

[0023] Step S4: Based on the relationship model and the new welding parameters, predict the strength of the corresponding welded joint.

[0024] This method for predicting weld joint strength combines the establishment of a microstructure model with thermo-mechanical coupling simulation, achieving refined simulation of the welding process. The relationship model between welding parameters and weld joint strength is established based on a statistical high-order multi-scale analysis algorithm, which can accurately predict the weld joint strength corresponding to different welding parameters, providing a theoretical basis for the refined design and optimization of the welding process. In situations where experimental data is insufficient, this method for predicting weld joint strength can be used for the optimized design of welding parameters, improving the quality and efficiency of welding production.

[0025] In one embodiment, the actual welding parameters and corresponding weld joint strength data collected in step S1 are further preprocessed, feature-engineered, and normalized to facilitate subsequent calculations such as establishing a microstructure model, performing thermo-mechanical coupling simulation, and establishing a relationship model between the simulated weld joint strength value and the welding parameters.

[0026] Specifically, in this embodiment, the welding parameters include: welding process parameters K, bevel angle θ, welding speed v, welding current I, argon flow rate f, welding tilt angle α, filler material ratio λ, weld width l, gap e, interpass temperature Ti, and heat input Q, etc., which are parameters that need to be set during the welding operation. Correspondingly, preprocessing includes removing outliers and filling missing values ​​in the welding parameters, and converting the data type and unit of the welding parameters to a preset format.

[0027] Outliers in these parameter data refer to data points that deviate significantly from other values ​​in the dataset. These outliers may be caused by measurement errors, data entry errors, or genuine variations. In practical applications, identifying outliers requires combining specific process requirements with historical data, and using statistical analysis methods (such as standard deviation, box plots, etc.) to identify data points that significantly deviate from the normal range. Handling outliers is crucial to ensuring data integrity and accuracy, as they can adversely affect model training and prediction results.

[0028] For example, regarding welding process parameter K, most data points are concentrated within a small range, such as 50-150. Therefore, a sudden high value like 500 or a low value like 0 is considered an outlier. For bevel angle θ, welding bevel angles typically have a reasonable range, such as 30 to 60 degrees. If a data point shows a θ value of 10 or 90 degrees, this exceeds the normal range and can be considered an outlier. For welding speed v, if most welding speed v data points are within a specific range, such as 2 m / min to 5 m / min, then a v value of 0.5 m / min or 10 m / min is considered abnormal. For welding current I, if the normal range for welding current I is 100 amperes to 300 amperes, then a current value such as 50 amperes or 500 amperes is abnormal. For argon flow rate f, if the normal range for argon flow rate f is 5 liters / minute to 20 liters / minute, then a flow rate value such as 2 liters / minute or 30 liters / minute is considered abnormal. For the welding tilt angle α, if the normal range for α is 45 degrees to 85 degrees, then an α value such as 10 degrees or 90 degrees is considered abnormal. For the filler material ratio λ, if the normal range for λ is 0.1 to 0.9, then an λ value such as 0 or 1 is considered abnormal. For the weld width l, if the normal range for l is 3 mm to 10 mm, then an l value such as 1 mm or 15 mm is considered abnormal. For the gap e, if the normal range for e is 0.5 mm to 2 mm, then an e value such as 0.1 mm or 3 mm is considered abnormal. For the interpass temperature Ti, if the normal range for Ti is 100 degrees to 300 degrees, then a Ti value such as 50 degrees or 400 degrees is considered abnormal. For heat input Q, if the normal range of heat input Q is 1 kJ / m to 10 kJ / m, then a Q value such as 0.5 kJ / m or 20 kJ / m is considered abnormal.

[0029] In one embodiment, for a welding parameter x, there exists a reasonable minimum value x. min and a reasonable maximum value x max ,when or When this happens, the actual value of the welding parameter is considered an outlier.

[0030] Converting the data type and units of welding parameters to a preset format ensures that all data is suitable for subsequent analysis and calculations. In the data preprocessing stage, data type conversion and unit standardization are two crucial steps that ensure data consistency and standardization before analysis.

[0031] Data type conversion refers to transforming data from one format or type to another to suit the requirements of analytical tools. This includes numeric type conversion, date type conversion, and categorical data encoding. Numeric type conversion involves converting string-type numbers to numeric types for mathematical operations. Date type conversion involves converting dates from one format (e.g., "YYYY-MM-DD") to another (e.g., "DD / MM / YYYY"), or to a timestamp. Categorical data encoding involves converting text categories into numeric representations, for example, using one-hot encoding or label encoding.

[0032] For example, the welding speed *v* exists in the dataset as a string, such as "5 m / min". Data type conversion will convert this string to a numeric type, i.e., 5, for subsequent calculations and analysis.

[0033] Unit standardization refers to converting data of the same type into the same unit of measurement to ensure comparability between data. For example, if the bevel angle θ appears in both degrees and radians in the dataset, they need to be standardized to a single unit, such as degrees. If the welding speed v appears in both meters per minute and centimeters per second in the dataset, they need to be standardized to a single unit, such as meters per minute. If the interpass temperature Ti appears in both degrees Celsius and Fahrenheit in the dataset, they need to be standardized to a single unit, such as degrees Celsius.

[0034] The aforementioned suitable formats for analysis refer to data that has been processed into a format that can be directly read and processed by analytical tools (such as statistical software, machine learning algorithms, etc.). These typically include: numerical conversion (all data should be numerical for mathematical operations); consistency (all data should follow the same format and units); missing value handling (missing values ​​should be imputed or removed); outlier handling (outliers should be identified and addressed); and feature encoding (categorical variables should be converted to numerical format so that the model can process them).

[0035] Feature engineering is performed on the actual welding parameters and corresponding weld joint strength data collected in step S1, including:

[0036] The Pearson correlation coefficient method was used to perform correlation analysis on welding parameters and weld joint strength data. For strongly correlated welding parameters, only one of them was retained. The calculation formula is as follows:

[0037]

[0038] In the formula, x i and y iThese represent two variables in the welding parameters and weld joint strength data, respectively. and It is the average of the corresponding variables, where n is the sample size and r is the mean. xy The correlation coefficient, |r|, represents the correlation coefficient between two variables. xy |≥0.6 indicates a strong correlation.

[0039] In this step, x i and y i This refers to any two data points from welding parameters and weld joint strength data, with a correlation coefficient r. xy The correlation coefficient r is a statistic that measures the degree of linear correlation between two variables. xy The value of r is between -1 and 1. xy A value close to 1 or -1 indicates a strong correlation, meaning there is a significant linear relationship between the two variables; r xy A value close to 0 indicates no correlation, meaning there is no significant linear relationship between the two variables. In one embodiment, |r xy |≥0.6 indicates a strong correlation between the two variables, |r xy | < 0.6 indicates that there is no correlation between the two variables. For example, if the absolute value of the correlation coefficient between welding speed and welding current is much greater than 0.6, it indicates that there is a strong linear relationship between the two features, and one of the parameters needs to be removed to avoid overfitting in the subsequent relationship model.

[0040] The actual welding parameters and corresponding weld joint strength data collected in step S1 are normalized, and the calculation formula that maps the data to the [0,1] interval is as follows:

[0041]

[0042] In the formula, x i The data can be any of the raw data from welding parameters and weld joint strength data, where min(x) and max(x) are the minimum and maximum values ​​in the corresponding datasets, respectively. i ′ represents the normalized data. For each type of data in the welding parameters and corresponding weld joint strength data, a sample dataset can be obtained after normalization.

[0043] In step S2, the normalized sample dataset is used as input parameters, and a weld microstructure model is established using numerical simulation methods, such as phase diagram methods or linear growth algorithms. A Kinetic Monte Carlo model is used to achieve real-time updates of the phase interface, simulating the microstructure evolution during the welding process. Periodic boundaries are established for the weld microstructure, the cooling process is simulated, and the coefficient of thermal expansion, thermal conductivity, specific heat capacity, and Young's modulus are calculated to complete the thermo-mechanical coupling solution process for the weld microstructure.

[0044] In one embodiment, step S3 involves establishing a relationship model between the simulated strength values ​​of the welded joint and the welding parameters, including:

[0045] Using the simulated strength values ​​of the welded joint and welding parameters as sample inputs, and employing a Gaussian multivariate kernel function as the basis kernel function, a high-dimensional mapping between the feature space and the samples is obtained. The calculation formula is as follows:

[0046]

[0047] Using the simulated strength values ​​of the welded joint and welding parameters as sample inputs, and employing the RBF kernel function as the basis kernel function, a nonlinear mapping is performed between the samples and the feature space. The calculation formula is as follows:

[0048] K RBF (x,y)=exp(-γ||xy|| 2 (4)

[0049] Combine formulas (3) and (4) in a predetermined manner to obtain the combined kernel function value K. combined (x,y), where x and y are input samples, σ is the parameter of the Gaussian multivariate kernel function, and γ is the parameter of the RBF kernel function;

[0050] Calculate the kernel function value K for all samples. combined (x,y) are used to form a kernel matrix, and the kernel vector for each sample is calculated.

[0051] In this step, a complex model capable of handling nonlinear relationships was constructed using Gaussian multivariate kernel functions and RBF kernel functions, thereby improving the accuracy of weld joint strength prediction. The Gaussian multivariate kernel function maps the data to a high-dimensional space, helping the model capture the complex nonlinear relationships between input features; the RBF kernel function, as a composite kernel function, further enhances the model's ability to handle nonlinear data. Through the combined use of these two kernel functions, the model can more effectively learn and predict the relationship between welding parameters and weld joint strength.

[0052] Both of these functions are kernel functions used to map data to a high-dimensional feature space. The purpose of a kernel function is to calculate the similarity between two samples in a high-dimensional space without explicitly calculating their coordinates. Combining these two functions means that they can capture similarities in different aspects of the data. For example, the Gaussian multivariate kernel function and the RBF kernel function can be combined to form a new kernel function, which can better capture the nonlinear relationships in the data. The specific combination method needs to be determined through experiments and validation with specific data to achieve the best predictive performance.

[0053] For example, in some embodiments, K is obtained by combination. combined The ways to represent (x,y) include the following:

[0054] K combined (x,y)=K Gaussian (x,y);

[0055] K combined (x,y)=K RBF (x,y);

[0056] K combined (x,y)=K Gaussian (x,y)+K RBF (x,y);

[0057] K combined (x,y)=K Gaussian (x,y)×K RBF (x,y).

[0058] In other embodiments, the Gaussian multivariate kernel function and the RBF kernel function can each be multiplied by a coefficient and then added together to obtain K. combined (x,y). For example, in one embodiment, K combined (x,y)=pK Gaussian (x,y)+qK RBF (x,y), the values ​​of coefficients p and q can be set according to actual needs.

[0059] In this step, the kernel function value K for all samples is calculated. combined (x, y) are used to form a kernel matrix, and the kernel vector for each sample is calculated. The kernel matrix is ​​formed by the kernel function values ​​K corresponding to all samples. combined It consists of (x, y). Specifically, if there is a dataset containing n samples {x1, x2, ..., x...} n}, then the elements K of the kernel matrix K ij It is determined by the kernel function K combined (x,y) is calculated. Therefore, the kernel matrix is ​​an n×n matrix that represents the similarity between all sample pairs.

[0060] The kernel matrix represents the relationship between samples in the feature space, and the kernel function K... combined The (x, y) matrix calculates the similarity between two samples in the feature space, while the kernel matrix is ​​the set of these similarities. The kernel vector is calculated by taking the kernel function values ​​of a sample and all other samples in the dataset, forming a vector. Specifically, for a sample {x1, x2, ..., x...}... n}, whose kernel vector k i It is K combined (x i ,x j The result is calculated for all j, where K is the sum of the values ​​of j and j. combined (x i ,x j The kernel function mentioned above is x. j It is the j-th sample in this sample set. In other words, the kernel vector k i The calculation is done by taking a sample x i Compared with all other samples x in the dataset j The kernel function values ​​(where j≠i) are calculated and arranged into a vector.

[0061] The purpose of kernel vector calculation is to represent samples in a high-dimensional feature space. In machine learning, kernel vectors can be used to calculate the similarity of samples in the feature space. Kernel vectors can map raw data to a high-dimensional space without explicitly calculating coordinates in that space. In this embodiment, kernel vector calculation is used to establish a model relating welding parameters to weld joint strength. Kernel vectors can map simulated values ​​of welding parameters and joint strength to a high-dimensional feature space, and these mappings are then used to train the model to predict weld joint strength. Kernel vector calculation, as a key step in establishing this relational model, enables effective similarity calculation in a high-dimensional space, thereby improving the model's predictive ability.

[0062] After establishing the relationship model between the simulated strength value of the welded joint and the welding parameters in step S3, the model is trained to improve its accuracy. The training method includes the following steps:

[0063] Step S301: Divide the welding parameters and corresponding weld joint strength data from step S1 into a training set and a test set;

[0064] Step S302: Substitute the training set into the relation model and optimize the relation model using the generalized expectation-maximization algorithm;

[0065] Step S303: Substitute the test set into the optimized relational model, and calculate the mean absolute percentage error (MAPE) between the predicted weld joint strength and the simulated weld joint strength obtained from the relational model. The calculation formula is as follows:

[0066]

[0067] In the formula, n is the sample size. Let σ be the predicted strength of the i-th weld joint. i Let be the simulated value of the strength of the i-th welded joint;

[0068] Step S304: Adjust the hyperparameters in the relation model. The hyperparameters include parameters σ and γ. Select different σ and γ to train the relation model. Repeat steps S302 and S303, and select the hyperparameter corresponding to the minimum mean absolute error (MAPE) to configure the relation model.

[0069] Regarding the allocation of welding parameters and corresponding weld joint strength data, the dataset used for training the model typically accounts for 70% to 80% of the total data; the dataset used for evaluating model performance typically accounts for 20% to 30% of the total data. That is, the training set accounts for 70% to 80% of the total data, and the test set accounts for 20% to 30% of the total data, which together equals 100%.

[0070] In specific data allocation, random sampling or specific data partitioning techniques, such as k-fold cross-validation, can be used. In k-fold cross-validation, the data is divided into k equal parts, the model is trained on k-1 parts, and tested on the remaining parts. This process is repeated k times, each time selecting a different part as the test set.

[0071] Optimizing the model using the generalized expectation-maximization algorithm is a conventional method for model optimization training, mainly including the following steps: Initialization: Selecting initial parameter estimates, including σ and γ; E-step: Calculating the expected value of each sample based on the current parameter estimates, for example, calculating the expected value of the actual joint strength under the given welding parameters and simulated joint strength; M-step: Maximizing the likelihood function or log-likelihood function based on the expected values ​​calculated in the E-step to update the estimates of σ and γ; Iteration: Repeating the E-step and M-step until the parameter estimates converge. As this is existing technology, it will not be described in detail here.

[0072] Training this relational model allows for the selection of a more accurate combination of the two kernel functions mentioned above, while also optimizing the parameters σ and γ to form the most accurate relational model between various welding parameters and the strength of the welded joint.

[0073] Based on the above relationship model, users only need to input new welding parameters to predict the strength of the corresponding welded joint.

[0074] Therefore, in this embodiment, a welding parameter optimization method is also disclosed, including the following steps:

[0075] First, set the range of values ​​for the strength of the welded joint and clarify the upper and lower limits of the required welding loudness.

[0076] Next, all values ​​of each welding parameter are substituted into the weld joint strength prediction method described above to obtain the corresponding predicted weld joint value. In this step, "all values ​​of each welding parameter" refers to all values ​​of any welding parameter within the standard or reasonable range, which is equivalent to traversing all possible values ​​within these ranges to obtain the corresponding predicted weld joint value.

[0077] Finally, the welding parameters corresponding to the predicted values ​​of the welded joints that meet the specified range are selected as the optimized welding parameters.

[0078] The aforementioned methods for predicting welded joint strength and optimizing weld joint performance establish a model relating welding parameters to welded joint strength, achieving a quantitative relationship between the two and providing technical support for improving the mechanical properties of welded joints. Compared with traditional methods, this invention combines simulation calculations with statistical analysis, improving prediction accuracy and providing a foundation for welding parameter optimization and research on welded mechanical properties.

[0079] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method of predicting strength of a welded joint, characterized by, Comprising the following steps: S1, collecting real welding parameters and corresponding welding joint strength data; preprocessing, feature engineering and normalization processing of the welding parameters and welding joint strength data; the preprocessing includes: removing outliers in the welding parameters and filling missing values, converting the data type and unit of the welding parameters to a preset format; The feature engineering of the welding parameters and welding joint strength data comprises: The welding parameters and welding joint strength data are analyzed by Pearson correlation coefficient method, and for the strongly correlated welding parameters, only one of them is retained, and the calculation formula is: (1) wherein, and denote two variables in the welding parameter and the weld joint strength data, respectively, and are the average values of the respective variables, n is the number of samples, denotes the correlation coefficient between the two variables, denotes a strong correlation; The welding parameters and welding joint strength data are normalized to map the data to the interval [0, 1], and the calculation formula is: (2) wherein, is any of the raw data of the welding parameters and the welding joint strength data, min( x ) and max( x ) are the minimum and maximum values in the respective data set, respectively, is the normalized data; S2, according to the welding parameters, a microstructure model is established, and a thermal-mechanical coupling simulation is carried out to obtain the corresponding welding joint strength simulation value; including: using the normalized data as input parameters, establishing a weld microstructure model by using a numerical simulation method, using a Kinetic Monte Carlo model to realize real-time updating of the phase interface, simulating the microstructure evolution in the welding process, establishing a periodic boundary for the weld microstructure, simulating the cooling process, calculating the thermal expansion coefficient, thermal conductivity, specific heat capacity and Young's modulus, and completing the thermal-mechanical coupling solution process of the weld microstructure; S3, a relationship model between the welding joint strength simulation value and the welding parameters is established; including: The welding joint strength simulation value and the welding parameters are input as samples, a Gaussian multivariate kernel function is used as a base kernel function, a high-dimensional mapping of the feature space and the samples is obtained, and the calculation formula is: (3) The welding joint strength simulation value and the welding parameters are input as samples, and an RBF kernel function is used as a base kernel function to perform nonlinear mapping of the samples and the feature space, and the calculation formula is: (4) Combining formula (3) and formula (4) in a preset manner to obtain a combined kernel function value , wherein, x and y is an input sample, is a parameter of a Gaussian multi-element kernel function, is a parameter of an RBF kernel function; computing the kernel function values for all samples forming a kernel matrix and computing kernel vectors for each sample; S4, based on the relationship model and new welding parameters, the corresponding welding joint strength is predicted.

2. The welding joint strength prediction method of claim 1, wherein the welding parameters include welding process parameters K, groove angle θ, welding speed v, welding current I, argon flow rate f, welding inclination angle α, filler material ratio λ, fusion width l, gap e, interlayer temperature Ti and heat input Q.

3. The welding joint strength prediction method of claim 1, wherein after the relationship model between the welding joint strength simulation value and the welding parameters is established, the relationship model is further trained, and the training method comprises: The welding parameters and corresponding welding joint strength data in step S1 are divided into a training set and a test set; The training set is substituted into the relationship model, and the relationship model is optimized by using a generalized expectation maximization algorithm; The test set is substituted into the optimized relationship model, the average absolute percentage error MAPE of the welding joint strength prediction value obtained from the relationship model and the welding joint strength simulation value is calculated, and the calculation formula is:

4. The welding joint strength prediction method of claim 1, comprising the following steps: Setting the value range of the welding joint strength; (5) In the formula, n is the number of samples, is the predicted value of the strength of the i th welded joint, is the simulated value of the strength of the i th welded joint; adjusting hyperparameters in the relational model, the hyperparameters including parameters and , selecting different and training the relational model, and selecting a configuration of hyperparameters corresponding to a minimum mean absolute percentage error (MAPE) for the relational model. ​ The ; or The ; or The ; or The .

5. A method of optimizing welding parameters, characterized by, ​ ​ all the values of the welding parameters are substituted into the welding joint strength prediction method as claimed in any one of claims 1 to 4 to obtain corresponding welding joint prediction values; the welding parameters corresponding to the welding joint prediction values meeting the value range are selected as the optimized welding parameters.

Images