Construction method of mechanical property prediction model of marine high-strength steel welding material, model and prediction method

Abstract

The invention provides a construction method of a mechanical property prediction model of a marine high-strength steel welding material, the model and a prediction method. The construction method comprises the following steps: S1, obtaining uniform mechanical property data samples of the marine high-strength steel welding material; s2, verifying and determining an input characteristic variable; s3, data standardization; s4, establishing a model, randomly selecting 80%-90% of the sample data as a training set, taking the rest 10%-20% of the sample data as a test set, modeling by adopting an XGBoost algorithm, and adjusting parameters; and calculating a determination coefficient R2, an average error and a root-mean-square error of the constructed model by using the test set, and when the determination coefficient R2 is greater than 0.92 and the average error percentage is less than or equal to 12%, representing that modeling succeeds. The mechanical property model of the marine high-strength steel welding material is wide in application range and high in prediction precision, the experimental workload of research and development of the marine high-strength steel welding material is greatly reduced, the experimental cost is saved, and the research and development period is shortened.

Description

Technical Field

[0001] The present invention relates to the technical field of welding material performance prediction, and in particular to a method for constructing a mechanical property prediction model for shipbuilding high-strength steel welding materials, a model and a prediction method. Background Art

[0002] With the proposal of the strategy of building a strong maritime nation, more and more new types of ships are emerging to meet more and higher performance requirements; the strength-to-weight ratio, yield strength and tensile strength of high-strength hull structural steel have been improved, which can significantly improve the ship's carrying capacity and play an important role in the design and construction of modern ship structures.

[0003] Since ships are made by welding together steel of various specifications, welded joints are often the weakest parts of ship structures, prone to fracture, fatigue cracks and other material failures. For ship-made high-strength steel weld metals, mechanical properties such as strength, plasticity and toughness are the most important performance indicators, which directly determine the overall performance and service safety of ship structures.

[0004] As an important material for ships and offshore structures, the performance level of welding materials will affect the safety of service. However, the current design method is still mainly based on trial and error, which leads to a large amount of experiments and a certain degree of randomness. To this end, the Chinese patent application number 200810236873.7 discloses a method for predicting the strength of 550-850MPa gas shielded welding welds, including (1) selecting welding materials and obtaining the chemical composition of the welds welded on the specified steel plate; (2) designing the welding process; (3) calculating. However, this model is only for gas shielded welding wires with ferrite weld structure and the calculation process is relatively complicated. Since the weld metal of high-strength steel for ships contains more bainite or martensite structures in addition to acicular ferrite structure, and is also required to have high low-temperature toughness and high strength, it is obvious that this model cannot be used to predict the strength of high-strength steel weld metal for ships.

[0005] In view of this, the present invention is proposed. Summary of the invention

[0006] The problem solved by the present invention is that the existing technology cannot achieve rapid and accurate prediction of the mechanical properties of high-strength steel weld metal, resulting in a large experimental workload and a long cycle, which seriously restricts the research and development progress of marine ship equipment.

[0007] In order to solve the above problems, the present invention provides a method for constructing a mechanical property prediction model of a marine high-strength steel welding material, wherein the properties include yield strength, tensile strength, elongation and impact energy at different temperatures;

[0008] The construction method comprises:

[0009] S1. Obtain uniform mechanical properties data samples of high-strength marine steel welding materials;

[0010] On the basis of historical test data, orthogonal test was adopted to design and prepare a series of welding materials with different compositions. Welding operation was carried out under different line energy. The alloy elements (C, Si, Mn, Ni, Cr, Mo, V, Ti, Cu), yield strength, tensile strength, elongation and impact energy at different temperatures of weld metal were measured. The above data were taken as data samples, and the maximum value, minimum value, average value and standard deviation of the sample data were calculated.

[0011] S2. The input characteristic variables of the model are determined by correlation analysis and significance t-test, among which the input characteristic variables of yield strength and elongation are C, Si, Mn, Ni, Cr, Mo, V and Ti elements, the input characteristic variables of tensile strength are line energy E and C, Si, Mn, Ni, Cr, Mo, V and Ti elements, and the input characteristic variables of impact energy are line energy E, temperature T, welding material type and C, Mn, Ni, Cr, Mo, V and Ti elements;

[0012] S3, data standardization;

[0013] S4. Model building

[0014] Randomly select 80%-90% of the sample data as the training set, and the remaining 10%-20% of the sample data as the test set, use the XGBoost algorithm to build the model and adjust the parameters; use the test set to calculate the determination coefficient R of the constructed model 2 , average error and root mean square error, when the determination coefficient R 2 When >0.92 and the average error percentage ≤12%, it means that the modeling is successful.

[0015] Preferably, step S2 includes analyzing the correlation coefficient and t-test value of alloy elements, line energy E and temperature, welding material type and welding material mechanical properties, and determining the input characteristic variables of the model according to the significance level of the correlation between the two. Preferably, when the significance level α=0.05, when the t-test statistic of a factor is greater than 1.96, it indicates that there is a significant correlation between the factor and the mechanical properties.

[0016] Preferably, in the correlation analysis of welding material types, the welding material types are assigned values ​​and then data analysis is performed.

[0017] Preferably, in step S3, the sample data is standardized according to the formula z=(x-μ) / σ, wherein μ and σ represent the mean value and standard deviation of the characteristic parameter respectively, and z represents the processed value.

[0018] Preferably, in step S4, the XGBoost model is adjusted by random grid search and five-fold cross validation method, the accuracy value is calculated for each division of cross validation, the average validation accuracy is calculated for each hyperparameter setting, and the hyperparameter with the highest average validation accuracy is selected as the model hyperparameter.

[0019] Preferably, the value range of the input characteristic variables in step S2 is: w(C): 0.015%~0.177%, w(Si): 0.011%~1.53%, w(Mn): 0.316%~2.6%, w(Ni): 0%~9.05%, w(Cr): 0%~1.23%, w(Mo): 0%~1.6%, w(V): 0%~0.213%, w(Ti): 0%~0.807%, w(Cu): 0%~0.879%, E: 6.6~22.5kJ / cm, T: -100~25℃.

[0020] Preferably, in step S1, a total of 453 sets of data samples related to yield strength, tensile strength and elongation, and 681 sets of data samples related to impact energy are obtained. Preferably, there are 128 sets of data samples for orthogonal test, accounting for no less than 10% of the total data samples, preferably 30% or 50%.

[0021] Compared with the prior art, the performance prediction method of shipbuilding high-strength steel welding materials described in the present invention has the following beneficial effects: 1) a wide range of applications, a large span of strength levels (300MPa to 1000MPa), and a wide temperature range (room temperature to -100°C); 2) all test data are derived from the measured data of shipbuilding structural steel welding materials, with high prediction accuracy, overcoming the shortcoming that empirical formulas cannot reflect interactions, and is suitable for predicting the mechanical properties of shipbuilding steel welding materials, greatly reducing the experimental workload of developing shipbuilding high-strength steel welding materials.

[0022] The present invention also provides a mechanical property prediction model for ship high-strength steel welding materials, which is obtained by using the construction method described in the claims.

[0023] The present invention also provides a performance prediction method for shipbuilding high-strength steel welding materials, which inputs preset "input characteristic variables" into the above-mentioned prediction model to output yield strength, tensile strength, elongation and impact energy at different temperatures. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 The influence of temperature on the impact energy of weld metal in the prior art;

[0025] Figure 2 Correlation analysis between the mechanical properties and input characteristic variables described in the embodiment of the present invention;

[0026] Figure 3This is the prediction result of the mechanical properties based on the constructed model according to the embodiment of the present invention. DETAILED DESCRIPTION

[0027] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings. The technical features of the embodiments of the present invention can be combined with each other without conflict.

[0028] In order to meet the needs of different fields, the types of new ships are also constantly expanding; since welded joints are the weakest parts of ship structures, simple physical tests cannot achieve rapid evaluation of the mechanical properties of weld metals, which seriously restricts the research and development of marine ship equipment. With the rapid development of computer technology, computing power has been greatly improved, and computer models and algorithms have gradually been used in the field of material performance prediction. Accurate prediction methods for the mechanical properties of marine high-strength steel weld metals can quickly evaluate the performance of marine steel welding materials, which is conducive to the design of auxiliary welding material alloy composition, thereby effectively reducing the amount of testing and shortening the development cycle of welding materials. To this end, the applicant proposes the following technical solutions:

[0029] For shipbuilding high-strength steel weld metal, strength, plasticity and toughness are the main mechanical performance indicators, and the main factors affecting the mechanical properties of weld metal are the chemical composition of the weld and thermal cycle parameters, temperature and welding material type. Among them, strength directly determines the bearing capacity of the hull structure, plasticity affects the difficulty of structural construction, and toughness directly affects the safety of the hull structure. With the large-scale use of high-strength hull structural steel in ship construction, weld metal is also required to have corresponding high strength, good plasticity, and a certain toughness.

[0030] The alloying elements contained in the weld metal of high-strength marine steel are mainly carbon (C), silicon (Si), manganese (Mn), chromium (Cr), nickel (Ni), molybdenum (Mo), titanium (Ti), vanadium (V), niobium (Nb), copper (Cu), aluminum (Al), boron (B), zirconium (Zr), etc. The alloying elements exist in the weld of high-strength marine steel mainly in the form of solid solution or compound. The formed compounds are mainly carbides, nitrides, oxides, sulfides or intermetallic compounds, such as Fe3O, Mn3C, VC, NbC, ZrC, TiC, VN, TiN, AlN, NbN, MnS, FeS, FeO, SiO2, etc. Among them, Ti, V, Nb, Mo are easy to form carbides, while Ti, Zr, V, Nb, Al, B are easy to form nitrides. Traditional high-strength steel welding materials are generally C-Mn-Ni-Cr-Mo system, which matches the alloy composition system of the parent material. Among them, carbon is the most important strengthening element. Due to its strong interstitial solid solution effect and influence on the structural transformation, it has a great influence on the strength, plasticity, toughness and crack resistance of the weld metal; the addition of other alloying elements such as Mn, Cr, Ni, Mo, etc., all affect the structural transformation of the weld metal to varying degrees, and at the same time play a certain role in solid solution strengthening. In order to better control the structural composition of the weld metal and reduce the segregation degree of the alloying elements, the weld metal generally adopts a diversified design of alloying elements. As the strength increases, the content of alloying elements also increases accordingly.

[0031] In addition to alloying elements affecting the mechanical properties of weld metal, temperature is also an important factor. The effect of temperature on the impact toughness of weld metal follows the following rule: as the service temperature gradually decreases, the impact energy of weld metal changes in three stages (e.g. Figure 1 As shown in the figure): When the service temperature is high, the fracture surfaces of the weld metal impact specimens are all ductile tearing fracture surfaces, which belong to ductile fracture. At this time, the impact energy of the weld metal is only affected by the inherent characteristics of the material and is little affected by temperature changes. Therefore, the impact energy remains on the upper platform with a very small variation range when the temperature is high; when the service temperature gradually drops to the ductile-brittle transition temperature range, brittle fractures begin to appear on the fracture surfaces of the impact specimens, and as the temperature drops, the area of ​​the brittle fractures increases rapidly, indicating that the impact specimens quickly change from ductile fractures to brittle fractures within this temperature range, and thus the impact energy decreases sharply as the temperature drops; when the service temperature is low, the fracture surfaces of the weld metal specimens are all transformed into brittle fractures, which belong to brittle fractures. Only a small amount of energy is needed to break the weld. At this time, the impact energy is at the lower platform position and does not change with temperature.

[0032] The welding materials for marine high-strength steel include welding rods, gas shielded welding wires, submerged arc welding wires and flux-cored welding wires. Different types of welding materials have significant differences in manufacturing processes and welding processes, which in turn affect the mechanical properties of the corresponding deposited metal, especially the impact toughness of the weld deposited metal.

[0033] In other words: the main factors affecting the mechanical properties of shipbuilding high-strength steel weld metal include the type of welding materials, alloy elements (C, Si, Mn, Ni, Cr, Mo, V, Ti, Cu), welding line energy, and temperature. The mechanical properties include the yield strength, tensile strength, elongation, and impact energy of the weld metal.

[0034] The present invention provides a method for constructing a mechanical property prediction model of a marine high-strength steel welding material, comprising:

[0035] 1) Obtain uniform mechanical properties data samples of high-strength marine steel welding materials

[0036] Since the historical test data of welding materials matching marine structural steel are mainly concentrated on several fixed grades of alloy composition systems, the data volume is small and too concentrated to be conducive to subsequent modeling; 128 groups of orthogonal experiments were designed, and a series of welding materials with different compositions were prepared. Welding operations were carried out under different line energies to measure the alloy elements (C, Si, Mn, Ni, Cr, Mo, V, Ti, Cu), yield strength, tensile strength, elongation and impact energy of weld metal at different temperatures. Combined with the above historical data, a total of 453 groups of data samples related to yield strength, tensile strength and elongation, and 681 groups of data samples related to impact energy were obtained. The maximum, minimum, average and standard deviation of the sample data were calculated. The results are shown in Table 1.

[0037] Table 1 Minimum, maximum, average and standard deviation of sample data

[0038]

[0039] 2) Verify and determine input feature variables

[0040] The correlation coefficients between alloy elements, line energy E and temperature and mechanical properties of welding materials were analyzed, and the input characteristic variables of the model were determined according to the significance level of the correlation between the two. The results are shown in Figure 2 , Table 2; The analysis of correlation coefficients was carried out using SPSS and other software. Since the types of welding materials were text-descriptive values, the types of welding materials were assigned values, where the values ​​of welding rods, gas shielded welding wires, submerged arc welding wires, and flux-cored welding wires were assigned as 1, 2, 3, and 4, respectively, and then data processing was performed.

[0041] Table 2 Correlation analysis between input characteristic variables and mechanical properties

[0042]

[0043] According to the significance level α = 0.05 and the degree of freedom n-2, we can check the t distribution table and get: t α / 2=1.96; if the t-test value of an input characteristic variable is greater than 1.96, it indicates that there is a significant correlation with the mechanical properties. As shown in Table 2, the input characteristic variables that pass the t-test for yield strength and elongation are C, Si, Mn, Ni, Cr, Mo, V and Ti elements, the input characteristic variables that pass the t-test for tensile strength are line energy E and C, Si, Mn, Ni, Cr, Mo, V and Ti elements, and the input characteristic variables that pass the t-test for impact energy are line energy E, temperature T, welding material type and C, Mn, Ni, Cr, Mo, V and Ti elements. In summary, the input characteristic variables of each model are determined according to the correlation coefficient r and t-test results as shown in Table 3:

[0044] Table 3 Input characteristic variables and output variables of the model

[0045]

[0046] 3) Sample data preprocessing

[0047] The sample data is standardized according to the formula z = (x-μ) / σ, where μ and σ represent the mean and standard deviation of the characteristic parameters, respectively, and z represents the value after processing. Since data such as line energy, alloy element content, and test temperature have different physical meanings and different dimensions, the sample data should also be standardized before modeling. By transforming the input features into a standard normal distribution (i.e., the mean is 0 and the standard deviation is 1), it can ensure that all features have the same importance in subsequent modeling, and avoid the numerical values ​​of some characteristic variables being too large or too small to dominate the model learning process.

[0048] 4) Construct a mechanical property prediction model for marine high-strength steel welding materials, including:

[0049] S41. Randomly select 80%-90% of the sample data as the training set, and the remaining 10%-20% of the sample data as the test set. Adjust the parameters of the XGBoost model through random grid search and five-fold cross validation. The specific range and step size of the hyperparameters are shown in Table 4. The accuracy value is calculated for each division of the cross validation. The average validation accuracy is calculated for each hyperparameter setting. The hyperparameter with the highest average validation accuracy is selected as the model hyperparameter to obtain the optimized hyperparameter combination, as shown in Table 5. The mechanical properties prediction model of high-strength steel welding materials is established using the optimized hyperparameter combination.

[0050] Table 4 Hyperparameter search range and step size

[0051]

[0052] Table 5. Optimized XGBOOST model hyperparameters

[0053]

[0054] Since the prediction of mechanical properties of welding materials is a complex problem with the interaction of multiple factors, traditional regression fitting methods are difficult to solve this complex problem, while machine learning algorithms such as BP neural network, multi-layer perceptron (MLP), support vector machine (SVM), random forest (BF), gradient boosting machine (GBDT) and extreme gradient boosting machine (XGBOOST) can be used to deal with such problems. After research, it was found that XGBOOST is more effective than other algorithms in modeling.

[0055] S42. Use the test set to calculate the coefficient of determination R of the constructed model 2 , average error and root mean square error, determine whether the coefficient R 2 >0.92 and the average error percentage is ≤12%. If so, it means that the model is successfully constructed; if not, the hyperparameter search range is expanded, including increasing the hyperparameter range and refining the step size, for example, the hyperparameter range is expanded to 120% of the original range and the step size is refined to 1 / 2-2 / 3 of the original step size, and then the process returns to step S41.

[0056] The coefficient of determination R 2 , average error and root mean square error and other indicators are used to analyze the prediction accuracy of the constructed model, and the prediction performance of the model is verified using the test set data. 2 >0.92 and the average error percentage ≤12%, indicating that the prediction accuracy of the constructed model is good; when the determination coefficient R 2 When <0.92 or the average error percentage is >12%, it means that the prediction accuracy of the model is poor. Then the hyperparameter search range is expanded (including increasing the hyperparameter range and refining the step size), and the five-fold cross validation and grid search methods are continued to obtain the best hyperparameters until a prediction model with ideal prediction accuracy is obtained. The results are shown in Table 6. Figure 3 .

[0057] Table 6 Model prediction accuracy indicators

[0058]

[0059] The present invention also provides a mechanical property prediction model for high-strength steel welding materials for ships, which is obtained by using the construction method described in the claims. Figure 3It can be seen that the mechanical property model of ship high-strength steel welding materials established in the present invention has a wide range of application, a large span of strength levels (300MPa~1000MPa), and a wide temperature range (room temperature~-100℃); all test data are derived from the measured data of ship structural steel welding materials, with high prediction accuracy, overcoming the shortcoming that empirical formulas cannot reflect interactions, and is suitable for the prediction of mechanical properties of ship steel welding materials, greatly reducing the experimental workload of the development of ship high-strength steel welding materials, saving test costs and shortening the development cycle, greatly improving the research and development speed of marine ship equipment.

[0060] The present invention also provides a performance prediction method for shipbuilding high-strength steel welding materials, which inputs preset "input characteristic variables" into the above-mentioned prediction model to output yield strength, tensile strength, elongation and impact energy at different temperatures.

[0061] Although the present invention is disclosed as above, the present invention is not limited thereto. Any person skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the scope defined by the claims.

Claims

1. A method for constructing a mechanical property prediction model for high-strength steel welding materials for ships, characterized in that: The properties include yield strength, tensile strength, elongation, and impact energy at different temperatures; The construction method comprises: S1. Obtain uniform mechanical properties data samples of high-strength marine steel welding materials; On the basis of historical test data, orthogonal test was adopted to design and prepare a series of welding materials with different compositions. Welding operation was carried out under different line energy. The alloy elements (C, Si, Mn, Ni, Cr, Mo, V, Ti, Cu), yield strength, tensile strength, elongation and impact energy at different temperatures of weld metal were measured. The above data were taken as data samples, and the maximum value, minimum value, average value and standard deviation of the sample data were calculated. S2. The input characteristic variables of the model are determined by correlation analysis and significance t-test, among which the input characteristic variables of yield strength and elongation are C, Si, Mn, Ni, Cr, Mo, V and Ti elements, the input characteristic variables of tensile strength are line energy E and C, Si, Mn, Ni, Cr, Mo, V and Ti elements, and the input characteristic variables of impact energy are line energy E, temperature T, welding material type and C, Mn, Ni, Cr, Mo, V and Ti elements; S3, data standardization; S4. Model building 80%-90% of the sample data are randomly selected as the training set, and the remaining 10%-20% of the sample data are used as the test set. The XGBoost algorithm is used to build the model and adjust the parameters; the test set is used to calculate the determination coefficient R of the constructed model. 2 , average error and root mean square error, when the determination coefficient R 2 When >0.92 and the average error percentage ≤12%, it means that the modeling is successful.

2. The method for constructing a mechanical property prediction model for marine high-strength steel welding materials according to claim 1, characterized in that: Step S2 includes analyzing the correlation coefficients among alloy elements, line energy E and temperature, welding material type and welding material mechanical properties, and determining the input characteristic variables of the model according to the significance level of the correlation between the two.

3. The method for constructing a mechanical property prediction model for marine high-strength steel welding materials according to claim 2, characterized in that: In the correlation analysis of welding material types, the welding material types are assigned values ​​and then data analysis is performed.

4. The method for constructing a mechanical property prediction model for marine high-strength steel welding materials according to claim 1, characterized in that: In step S3, the sample data is standardized according to the formula z=(x-μ) / σ, where μ and σ represent the mean value and standard deviation of the characteristic parameter respectively, and z represents the processed value.

5. The method for constructing a mechanical property prediction model for marine high-strength steel welding materials according to claim 1, characterized in that: In step S4, the XGBoost model is adjusted through random grid search and five-fold cross validation. The accuracy value is calculated for each division of cross validation, the average validation accuracy is calculated for each hyperparameter setting, and the hyperparameter with the highest average validation accuracy is selected as the model hyperparameter.

6. The method for constructing a mechanical property prediction model for marine high-strength steel welding materials according to claim 1, characterized in that: The value range of the input characteristic variables in step S2 is: w(C): 0.015%~0.177%, w(Si): 0.011%~1.53%, w(Mn): 0.316%~2.6%, w(Ni): 0%~9.05%, w(Cr): 0%~1.23%, w(Mo): 0%~1.6%, w(V): 0%~0.213%, w(Ti): 0%~0.807%, w(Cu): 0%~0.879%, E: 6.6~22.5kJ / cm, T: -100~25℃.

7. A mechanical properties prediction model for high-strength marine steel welding materials, characterized in that: Obtained by the construction method described in any one of claims 1 to 6.

8. A method for predicting the performance of high-strength steel welding materials for ships, characterized in that: The preset "input characteristic variables" are input into the prediction model described in claim 7, and the yield strength, tensile strength, elongation and impact energy at different temperatures are output.

Images