CN101947693B - A Tailored Laser Welded Blank Process Optimization System and Method Based on Performance Prediction - Google Patents

Abstract

The invention relates to an optimization system of a laser tailor-welded blank technology based on property prediction and a method. The system comprises a database, a pre processing module, a mechanical property prediction module, a technology optimization module and a post processing module, wherein the pre processing module is used for reading basic information and technological parameter on welders and base metals needed in the laser welding process from the database so as to provide initial conditions for the follow-up flows; the mechanical property prediction module is used for simulating different technological states to obtain the property prediction value of a target and a corresponding optimal PLS prediction model set, wherein the relative error of the property prediction value is less than or equal to 5%; the technology optimization module is used for calling out all built-in yield-strength ratios of the tailor-welded blanks in the system after users input the related information on the welded base metals based on the built-in optimization property scope in the system database, and then is used for selecting the yield-strength ratio, and finally the optimal PLS prediction model set is used for carrying out reverse solution to obtain the technological scheme for optimalizing the mechanical property of the tailor-welded blanks; and the post processing module is used for finishing display and output of the result.

Description

A Tailored Laser Welded Blank Process Optimization System and Method Based on Performance Prediction

technical field

The invention relates to the technical field of laser welding, in particular to a system and method for formulating and optimizing a laser welding process according to the performance requirements of tailor welded blank products.

Background technique

In recent years, with the growth of the national economy, economical and practical cars have attracted much attention because of their light weight, low fuel consumption and high safety. Since the application of tailor-welded blanks reduces the quality of the body and production costs, it is widely used in automobile manufacturers. However, for a long time, the research and development work on various high-performance tailor-welded blanks has been limited to experimental observation and general theoretical discussion, especially when it comes to the process design and process optimization of various tailor-welded blank structures, it is rarely given Quantitative process simulation and its prediction results are produced, which greatly increases its research and development costs and cycle. If advanced computer simulation and prediction technology can be introduced in the field of laser welding, the mechanical properties of tailor-welded blanks can be predicted quickly and accurately, and the laser welding process can be adjusted and optimized in time, which is very important for the development of high-performance tailor-welded blanks engineering significance. However, the relevant results of such research are rarely reported at home and abroad.

The latest retrieval work shows that Kobe Steel Corporation of Japan applied for the patent "Weld metalexcellent in toughness and SR cracking resistance" (Patent No.: US07597841) to the US Patent Office in 2007. The content of this patent is mainly through the adjustment of the welding metal composition, In order to obtain excellent weld mechanical properties. In addition, Josida Khirosi of Nippon Steel Corporation applied to the European Patent Office in 2009 for a patent Device to forecast disruption of part subjected to pointwelding method to this end computer software and machine-readable data carrier (patent number: RU2006013994820050412). The main content is to invent a device for predicting the rupture of spot weldments based on terminal computer processing. Because during spot welding, as the number of welding points increases, the plastic deformation of the electrode head causes the diameter of the electrode head to increase and other reasons will cause desoldering, which requires manual intervention. However, there is no report on the patent results of the mechanical property prediction and process optimization methods of laser tailor welded blanks.

Contents of the invention

The purpose of the present invention is to take the commonly used laser tailor welded steel plate as the research object, specifically related to the cold-rolled deep-drawing plate series St12 and its galvanized plate; high-strength galvanized steel DOGAL800DP/super-drawn steel BUSD, high-strength low-alloy steel plate series B240 /390DP, etc., provide a system and method for formulating and optimizing the laser welding process according to the performance requirements of tailor-welded blank products.

The technical solution to achieve one of the objectives of the present invention is: a tailor-welded blank process optimization system based on performance prediction, the system is composed of SQL database, pre-processing module, mechanical performance prediction module, process optimization module, and post-processing module;

The SQL database contains the basic information of the tailor-welded blank base metal;

The pre-processing module is used to read the basic information and process parameters of the weldment and base metal required for the laser welding process from the database, and provide initial conditions for the subsequent process;

The mechanical performance prediction module obtains a performance prediction value with a target relative error ≤ 5% and a corresponding optimal PLS prediction model group by simulating different process states;

The process optimization module is based on the built-in optimization performance range in the system database, which is the minimum value to the maximum value of the yield ratio of tailor-welded blanks, and is adjusted by the user after inputting relevant welding base material information. All the yield strength ratios of the tailor-welded blanks built in the system are obtained and selected; for a specific yield ratio selected, the mechanical properties of the tailor-welded blanks can be obtained by inverse calculation by the optimal PLS prediction model group Optimal process plan;

The post-processing module completes the display and output of the results, outputs the results of the process optimization module in the form of tables or graphs, and outputs ordinary text analysis reports at the same time.

details as follows:

SQL database function: This database contains the basic information of the tailor-welded blank base metal, including material grade, chemical composition, base metal thickness, weldment size and other physical performance parameters, welding process parameters and other data required by the system.

The specific process of the pre-processing module is: first read the welding process parameters from the SQL database, including: welding base metal grade, composition, base metal thickness, weldment size, laser power, welding speed, spot diameter, line energy, defocus Welding process parameters such as weight, focal length, heat absorption coefficient, etc. If the reading is correct, it will be transferred to the mechanical performance prediction module; if there is an error, it can return to re-read the process parameters.

The mechanical performance prediction module uses steel composition, tailor-welded blank base metal thickness, and welding process parameters to establish a partial least squares prediction (Partial Least-Squares Regression, referred to as "PLS") formula, and to test and control the accuracy of the model, in order to The prediction formula of PLS mechanical properties with high precision is obtained.

In the process optimization prediction module, after the user enters the relevant welding base metal information, all the yield strength ratios of the tailored welded blank built in the system can be called out and selected. For a selected specific yield strength ratio, the process scheme for optimizing the mechanical properties of tailor-welded blanks can be obtained from the optimal PLS prediction model group. When the user selects the default option, the process optimization module defaults the middle limit of the extreme value range of the yield strength ratio of the tailored welded blank as the optimal performance value, and thus obtains the corresponding optimal process plan factor level combination.

The post-processing module is used to display and output calculation results, output the results of the process optimization module by means of tables, charts, etc., and also includes ordinary text analysis reports.

The technical solution to achieve another object of the invention is: a method for optimizing the laser tailor welded blank process based on performance prediction, the method includes the following steps:

(1) Read the base metal information and process parameters required for the laser welding process from the database, and provide initial conditions for the subsequent process;

(2) Combining the basic information of the weldment to be tested, including composition, thickness and process parameters, a PLS mechanical performance prediction model group is established, and the accuracy of the PLS formula is checked and controlled to obtain the most accurate PLS mechanical performance prediction model group steps;

(3) Based on the optimized performance range already built in the system database, this range is the minimum value to maximum value of the yield ratio of tailor-welded blanks. All the yield strength ratios of the tailored welded blank are selected and selected, and the process plan for optimizing the mechanical properties of the tailor welded blank is obtained according to a specific selected yield strength ratio.

(4) Output optimized process parameters.

The step (1) further includes, when the welding process parameters are read correctly, then go to the next step; if there is an error, then return to re-read the process parameters; the welding process parameters are well known to those skilled in the art It includes: the steel type and its composition of the welded plate, the thickness specification of the plate to be welded, the laser power, the welding speed and the spot diameter, etc.

The step (2) specifically includes: using the cross-validity to determine the number of principal components to obtain the prediction equation for the minimization of PRESSh, and using the precision inspection and control of the target relative error of 5% to obtain the highest precision PLS mechanical performance prediction The model group and the optimal laser welding scheme obtained by inverse mapping.

The step (3) further includes, for the selected specific yield strength ratio, the optimal PLS prediction model group is reversely obtained to obtain the optimal process plan for the mechanical properties of the tailored welded blank; when the user selection item is By default, the process optimization module defaults the middle limit of the extreme value range of the yield strength ratio of tailor-welded blanks as the optimal performance value, and thus obtains the corresponding optimal process plan.

The way of outputting in the step (4) may be displaying and outputting each prediction result in a form, a curve or an intelligent report.

Advantage of the present invention is:

1. According to the target mechanical properties of tailor-welded blanks proposed by users, the corresponding optimal laser welding process scheme can be provided.

2. According to the optimal laser welding process plan, the corresponding process simulation can be carried out and the mechanical properties of tailor-welded blanks can be predicted. By using the system of the invention, the prediction accuracy of the yield strength, tensile strength and elongation of the tailored welded blank can all reach more than 95%, thereby ensuring the precision of material control of the tailored welded blank.

3. The database provided by the present invention has a large amount of basic information of tailor-welded blank base metal and laser welding production process parameters, with friendly interface, input and output consistent with the production process, and easy to operate. Each module of the system runs on a computer with standard configuration, which realizes the separation of calculation and result output, and facilitates program debugging, upgrading, maintenance and transplantation.

4. The present invention has excellent universality, and can be popularized and applied to the laser welding application field of various new and super toughened materials. It provides a reliable basis for the final performance of a new type of super-toughened tailor-welded blank.

Description of drawings

Fig. 1 is a system overall block diagram of the present invention;

Figure 2 is a block diagram of the establishment of the mechanical performance prediction module;

Figure 3 is a block diagram of the establishment process of the process optimization module.

Detailed ways

The present invention will be described in further detail below in conjunction with the accompanying drawings.

As shown in Fig. 1, the present invention takes commonly used tailor-welded blanks as the research object, and establishes a method and system for process optimization of tailor-welded blanks based on performance prediction. The overall model consists of a database, a pre-processing module, a mechanical performance prediction module, a process optimization module, and a post-processing module. Among them, the pre-processing module and the post-processing module are auxiliary modules to realize the establishment of the mechanical property prediction module and the display output of the optimization results of the process optimization module.

When establishing the mechanical performance prediction module of tailor welded blanks as shown in Figure 2, in order to facilitate the control of the prediction accuracy of the mechanical performance of tailor welded blanks, it is necessary to establish a PLS mechanical performance prediction model group in combination with the thickness of the weldment to be measured and the process parameters, and PLS The accuracy is tested and controlled by the formula, and the PLS mechanical performance prediction model group with the highest accuracy is obtained.

In order to make the best fitting degree between the output value of the PLS mechanical property prediction module of tailor-welded blanks and the actual value, the error square sum PRESS _h of the output value should be minimized; at the same time, in order to eliminate the influence of multiple correlations of each input variable, it is necessary to The cross validity is used to determine the number of principal components to obtain the prediction equation for the minimization of PRESS _h , and the calculation tool uses MATLAB software.

In addition, in order to make the built module have the ability to explain the input and output variable space similar to the typical multiple linear correlation analysis and make it reach a higher level, the principal component t _h should be made to have the input variable X and the output variable Y. Cumulative explanatory power - RdXtt and RdYtt maximized.

The control of the prediction accuracy of the built module is to control the target relative error between the predicted value of the module and the real value (target relative error ≤ 5%). If the accuracy requirement is met, the result will be output; if not, it will return The pre-processing module repeats the modeling work in Figure 2 until the target relative error is reached.

The process optimization module shown in Figure 3 is based on the optimized performance range built in the system database. This range is the minimum value to maximum value of the yield ratio of tailor-welded blanks. The user inputs the relevant welding After the parent metal information is obtained, all the yield strength ratios of the tailor-welded blank built in the system can be recalled and selected. For a specific selected yield strength ratio (user target yield strength ratio), the optimal PLS prediction model group can be inversely calculated to obtain the corresponding optimized process factor level combination, which means that the optimal mechanical properties of tailor welded blanks can be obtained process plan. When the user selects the default option, the system defaults the middle limit of the extreme value range of the yield strength ratio of tailor-welded blanks as the optimal target performance value, and thus obtains the corresponding optimal process plan factor level combination. This default value is based on the fact that tailor-welded blanks in engineering applications require subsequent forming processes, and the subsequent forming processes require that tailor-welded blanks have high formability, that is, the yield strength ratio should not be too high. If the yield ratio is too high, the formability will be significantly reduced, and various forming defects will easily occur.

Below in conjunction with the accompanying drawings, the implementation process of the present invention will be further described step by step through four embodiments.

Example 1

Taking 1.5mm St12 sheet/galvanized sheet tailor-welded blank as an example, the welding process adopts single-sided welding and double-sided forming. The welding process parameters are: power P=1525~1850W, welding speed 1.6~2.0m/min, spot diameter Φ0. 3mm to 1mm, the absorption rate is 0.7, and the focal length of the welding lens is 127mm.

j＝0，1，…k，i＝1，2)进行PLS计算，其算法步骤如下。Using MATLAB software to map the existing welding process parameters and mechanical properties

j=0, 1,...k, i=1, 2) to perform PLS calculation, the algorithm steps are as follows.

Step 1: Relevant process data of weldment samples (≥9), as shown in Table 1.

Table 1 Process data related to weldment samples

Step 2: Investigate the issue of multiple correlations between input variables. It can be seen from Table 2 that the multiple correlations between input variables are obvious.

Table 2 Correlation coefficient matrix between input variables and output variables in weld area

Step 3: After being processed by the principle of cross validity, the optimal number of principal components of the yield strength is obtained, as shown in Table 3, when h=3, PRESS _hmin =0.324679, h=1, 2, 3; the same reason Obtain the optimal number of principal components of tensile strength, as shown in Table 4, when h=3, PRESS _hmin =0.376854; similarly, obtain the optimal number of principal components of elongation, as shown in Table 5, When h=3, PRESS _hmin =0.489121; therefore, the present invention finally takes three main components to establish yield strength, tensile strength and elongation prediction modules, respectively as follows.

Yield strength _y1 is

y ₁ =-0.0289x ₁ +36.9053x ₂ -20.9374x ₃ +175.3353

The tensile strength _y2 is:

y ₂ =-0.0121x ₁ +17.6536x ₂ -38.8001x ₃ +358.1600

The elongation _y3 is:

y ₃ =0.0028x ₁ -3.4113x ₂ -6.6175x ₃ +28.6674

In the above two formulas, x ₁ - laser power (W), x ₂ - welding speed (m/min), x ₃ - spot diameter (mm).

Step 4: Principal Component Analysis of the PLS Equation

After performing principal component analysis on the above y ₁ , y ₂ , and y ₃ respectively, Table 3, Table 4, and Table 5 are obtained.

In Table 3, the symbol RdXt represents the explanatory ability of the principal component t _h to the input variable X; the symbol RdXtt represents the cumulative explanatory ability of the principal component t _h to the input variable X; the symbol RdYt represents the explanatory ability of the principal component t _h to the output variable Y; Among them, the symbol RdYtt represents the cumulative explanatory ability of the principal component t _h to the output variable Y; PRESS _h is the sum of squared prediction errors of Y, which is the basis for judging the number of principal components selected by the cross-validation method.

Table 3 shows that three principal components should be selected. At this time, PRESS _hmin = 0.324679, which explains 99.9384% of the variation information in the original input variable system and 97.6633% of the variation information in the output variable system. The cumulative explanatory power is high.

Table 3 Principal component analysis data of yield strength

Table 4 shows that three principal components should be selected. At this time, PRESS _hmin = 0.376854, which explains 93.4899% of the variation information in the original input variable system and 96.4832% of the variation information in the output variable system. The cumulative explanatory power is high.

Table 4 Principal component analysis data of tensile strength

Similarly, Table 5 shows that three principal components should be selected. At this time, PRESS _hmin = 0.489121, which explains 94.3812% of the variation information in the original input variable system and 97.2396% of the variation information in the output variable system. The accumulative explanatory power of both the output variable and the output variable is high.

Table 5 Principal component analysis data of elongation

Step 5: Prediction accuracy inspection and control

Import the process data of the pre-processing module into the mechanical performance module to predict the mechanical performance. The prediction results are shown in Table 6. It can be seen from Table 6 that the relative error of yield strength is 0.1813-4.6043%, the relative error of tensile strength is 0.0274-3.7452%, and the relative error of elongation prediction is 0.0658-4.9447%.

Table 6 Predicted and actual values of tensile strength and elongation

Step 6: User enters desired performance

The extreme value range of the yield ratio of the tailored welded blank is 0.48 to 0.55, and the user determines the yield ratio to be 0.50.

Step 7: Output optimized process parameters

x ₁ =1735W, x ₂ -1.6m/min, x ₃ =1mm, Rel=163, Rm=327MPa, A=21%

Example 2

Taking 0.8mm St12 plate/1.5mm galvanized sheet tailor-welded blank as an example, the welding process is the same as that of Example 1, and its modeling and optimization process and steps are the same as those of Example 1

Step 1: Relevant process data of weldment samples (≥9), as shown in Table 7.

Table 7 Process data related to weldment samples

After steps 2 to 4 described in Example 2, a PLS prediction model group of mechanical properties is obtained, as shown below.

The yield strength _y1 is:

y ₁ =-0.0471x ₁ +39.1667x ₂ -21.4414x ₃ +202.0455

The tensile strength _y2 is:

y ₂ =-0.0472x ₁ +38.3333x ₂ -23.8288x ₃ +359.7018

The elongation _y3 is:

y ₃ =0.0103x ₁ -6.6667x ₂ +4.2793x ₃ +16.6625

Step 5: Prediction accuracy inspection and control

Import the process data of the pre-processing module into the mechanical property prediction module to predict the mechanical properties. The prediction results are shown in Table 8. It can be seen from Table 8 that the relative error of yield strength is 0.6948-4.8694%, the relative error of tensile strength is 0.1224-4.5038%, and the relative error of elongation prediction is 1.3132-4.4867%.

Table 8 Predicted and actual values of tensile strength and elongation

Step 6: User enters desired performance

The extreme value range of the yield strength ratio of the tailored welded blank is: 0.50～0.58, and the user determines the yield strength ratio to be 0.54.

Step 7: Output optimized process parameters

x ₁ =1735W, x ₂ -1.6m/min, x ₃ =1mm, Rel=163, Rm=327MPa, A=21%

Example 3

Taking 1.5mm high-strength galvanized steel DOGAL800DP/super-drawn steel BUSD tailor-welded blank as an example, the welding process adopts single-sided welding and double-sided forming. The welding process parameters are: laser power 900-1400W, welding speed 1-2m/ min, the spot diameter is 0.3-1.0mm, the absorption rate is 0.7, and the focal length of the welding lens is 127mm. Its modeling and optimization process and steps are the same as those in Example 1.

Step 1: Call a group (≥9) of relevant process data randomly, as shown in Table 9.

Table 9 Process data related to weldment samples

After steps 2 to 4 described in Example 1, a PLS prediction model group of mechanical properties is obtained, as shown below.

Yield strength _y1 of tailor welded blank is:

y ₁ =-0.0187x ₁ +9.0000x ₂ -13.8739x ₃ +250.9757

The tensile strength _y2 of tailor welded blank is:

y ₂ =-0.0193x ₁ +10.3333x ₂ -13.8739x ₃ +339.7423

Tailored welded blank elongation y ₃ is:

y ₃ =0.0047x ₁ -2.6667x ₂ +3.7838x ₃ +29.9036

In the above two formulas, x ₁ - laser power (W), x ₂ - welding speed (m/min), x ₃ - spot diameter (mm).

Step 5: Prediction accuracy inspection and control

Import the process data of the pre-processing module into the mechanical performance module to predict the mechanical performance. The prediction results are shown in Table 10. It can be seen from Table 10 that the relative error of tensile strength is 0.0549-3.3265%, the relative error of tensile strength is 0.0900-3.8579%, and the relative error of elongation prediction is 0.5213-3.9148%.

Table 10 Predicted and actual values of tensile strength and elongation

Step 6: User enters desired performance

The extreme value range of the yield ratio of the tailored welded blank is: 0.68～0.75, and the default value of the yield ratio is determined by the user.

Step 7: Output optimized process parameters

x ₁ =1400W, x ₂ -1.0m/min, x ₃ =0.7mm, Rel=224, Rm=313MPa, A=37%

Example 4

Taking 1.5mm high-strength galvanized steel DOGAL800DP/2.0mm super-drawn steel BUSD tailor-welded blank as an example, the welding process is the same as Example 3, and the modeling and optimization process and steps are the same as Example 1.

Step 1: Relevant process data of weldment samples (≥9), as shown in Table 11.

Table 11 Process data related to weldment samples

After steps 2 to 4 described in Example 2, a PLS prediction model group of mechanical properties is obtained, as shown below.

The tensile strength _y1 is:

y ₁ =-0.0271x ₁ +13.2568x ₂ -19.4452x ₃ +267.8816

The tensile strength _y2 is:

y ₂ =-0.0132x ₁ +7.8176x ₂ -22.8733x ₃ +347.6128

The elongation _y3 is:

y ₃ =0.0033x ₁ -2.1053x ₂ +5.8724x ₃ +22.4861

Step 5: Prediction accuracy inspection and control

Import the process data of the pre-processing module into the mechanical performance module to predict the mechanical performance. The prediction results are shown in Table 12. It can be seen from Table 12 that the relative error of yield strength is 0.3931-3.9920%, the relative error of tensile strength is 0.3931-3.9920%, and the relative error of elongation prediction is 1.6324-3.1255%.

Table 12 Predicted and actual values of tensile strength and elongation

Step 6: User enters desired performance

The extreme value range of the yield ratio of the tailored welded blank is: 0.70~0.78, and the default value of the yield ratio is determined by the user.

Step 7: Output optimized process parameters

x ₁ =1016W, x ₂ -1.3m/min, x ₃ =1mm, Rel=238, Rm=321MPa, A=29%.

Claims (6)

1. the laser assembly solder plate technique optimizing system based on performance prediction is characterized in that, this system is comprised of SQL database, pre-processing module, mechanical properties prediction module, process optimization module, post-processing module;

Described SQL database has comprised tailor welded mother metal essential information;

Described pre-processing module is for the mother metal essential information and the welding condition that read the required weldment of laser beam welding from SQL database, for follow-up flow process provides primary condition;

Described mechanical properties prediction module by the simulation to the different process state, draws performance prediction value and the corresponding optimum PLS forecast model group of target relative error≤5%;

Described process optimization module, Optimal performance scope built-in in the SQL database is as foundation, this scope is the minimum ~ maximum of tailor welded yield tensile ratio, by the user after the mother metal essential information of the relevant weldment of input, all yield tensile ratio values of this tailor welded that the system that accesses is built-in, and it is selected; For selected a certain concrete yield tensile ratio value, inquire into the process program that obtains making tailor welded mechanical property optimum by the PLS forecast model group of optimum is counter;

Described post-processing module is used for finishing result's demonstration output, and employing form or graph mode are exported the result of process optimization module, export simultaneously the plain text analysis report.

2. the laser assembly solder plate technique optimization method based on performance prediction is characterized in that, the method may further comprise the steps:

(1) from SQL database, reads laser beam welding required mother metal essential information and welding condition, the step of primary condition is provided for follow-up flow process;

(2) in conjunction with mother metal essential information and the welding condition of weldment, the mother metal essential information comprises steel grade, composition and the sheet metal thickness specification to be welded of tailor welded, set up PLS mechanical properties prediction model group, and the PLS formula is carried out check and the control of precision, obtain the step of the highest PLS mechanical properties prediction model group of precision;

(3) in the SQL database built-in Optimal performance scope as foundation, this scope is the minimum～maximum of tailor welded yield tensile ratio, by the user after the mother metal essential information of the relevant weldment of input, all yield tensile ratio values of this tailor welded that the system that accesses is built-in, and it is selected; And according to selected a certain concrete yield tensile ratio value, ask the process program that obtains making tailor welded mechanical property optimum;

(4) welding condition is optimized in output.

3. the laser assembly solder plate technique optimization method based on performance prediction according to claim 2 is characterized in that, described step further comprises in (1), when welding condition reads errorlessly, then changes next step over to; Errors excepted, then return and again read welding condition; Described welding condition comprises: laser power, speed of welding and spot diameter.

4. the laser assembly solder plate technique optimization method based on performance prediction according to claim 2, it is characterized in that: described step specifically comprises in (2): adopt and intersect validity and determine the number of principal component, to obtain PRESS _h Minimized predictive equation utilizes target relative error 5 %Accuracy test and control, obtain the highest PLS mechanical properties prediction model group and the anti-optimum laser weld scheme that obtains of asking of inverse mapping of precision.

5. the laser assembly solder plate technique optimization method based on performance prediction according to claim 2, it is characterized in that, described step further comprises in (3), for selected a certain concrete yield tensile ratio value, inquire into the process program that obtains making tailor welded mechanical property optimum by the PLS forecast model group of optimum is counter; When user selection item when being default, the process optimization module then is defaulted as the Optimal performance value with the middle limit value of tailor welded yield tensile ratio extreme value scope, and draws thus corresponding optimum process scheme.

6. the laser assembly solder plate technique optimization method based on performance prediction according to claim 2 is characterized in that, adopts form, chart or plain text analysis report mode that the result of process optimization is shown output in the described step (4).

Images