Method and apparatus for predicting mechanical properties

Abstract

To accurately predict mechanical properties of a casting by considering various influences occurring in the casting, and to easily obtain favorable prediction results with a small number of explanatory variables.SOLUTION: A method for predicting mechanical properties of a casting C molded by pouring molten metal into a cavity formed in a pair of molds that can be opened and closed, and solidifying the molten metal includes: an explanatory variable-obtaining step (step S2) of performing a predetermined analysis by using a mold model for CAE analysis obtained by dividing the cavity into a plurality of elements to obtain explanatory variables, which are information related to casting conditions or the molten metal of the casting C; and a mechanical property-predicting step (step S4) of inputting the explanatory variables into a prediction model obtained by calculating a correlation between the explanatory variables and objective variables, which are information related to mechanical properties of the casting C, thereby predicting the objective variables. As the explanatory variables, factors related to a solid solution amount of strengthening elements contained in the casting C after solidification are used.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to a method and an apparatus for predicting mechanical properties. [Background technology]

[0002] In the past, in the design stage of a casting, in order to determine whether the casting has the desired mechanical properties, the designer has repeatedly produced prototypes while adjusting the casting conditions, material composition, etc., based on his or her experience. However, such a determination method requires a lot of time and incurs a great deal of cost. Therefore, in recent years, an objective method that does not depend solely on the designer's experience has been used to predict the mechanical properties of a casting.

[0003] For example, Patent Document 1 listed below discloses a technology for predicting the mechanical properties of a cast product by creating a mold model for CAE (Computer Aided Engineering) analysis, using this mold model to perform melt flow analysis and solidification analysis under specified casting conditions, calculating specified factors from the analysis results, using the factors as explanatory variables, and then performing multiple regression analysis with desired mechanical properties as target variables. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2019-105592 Summary of the Invention [Problem to be solved by the invention]

[0005] However, this technique does not take into account the effects on mechanical properties of the amount of strengthening elements dissolved in the cast product after solidification, or the changes in the wall thickness of the cast product due to changes in mold temperature, leaving room for improvement in the selection of explanatory variables. Furthermore, because this technique requires at least three factors as explanatory variables, the calculation costs for obtaining the explanatory variables are high, and a simpler prediction method was needed.

[0006] The present invention has been made in consideration of the above-mentioned problems existing in the prior art, and its purpose is to provide a method for predicting the mechanical properties of a casting with high accuracy by taking into account various influences that occur in the casting, and also to provide a method for easily obtaining good prediction results with a small number of explanatory variables. [Means for solving the problem]

[0007] The present invention will be described below. In order to facilitate understanding of the present invention, reference numbers in the accompanying drawings are added in parentheses, but the present invention is not limited to the illustrated forms.

[0008] The mechanical property prediction method according to the present invention is a mechanical property prediction method for predicting the mechanical properties of a casting (C) formed by injecting molten metal into a cavity formed in a pair of openable and closable molds and solidifying the molten metal, and includes an explanatory variable acquisition step (step S2) for obtaining explanatory variables, which are information about the casting conditions of the casting (C) or the molten metal, by performing a predetermined analysis using a mold model for CAE analysis in which the cavity is divided into a plurality of elements; and a mechanical property prediction step (step S4) for inputting the explanatory variables into a prediction model that calculates a correspondence between the explanatory variables and objective variables, which are information about the mechanical properties of the casting (C), and predicting the objective variables, wherein the explanatory variables are used as factors related to the amount of solid solution of strengthening elements contained in the casting (C) after solidification.

[0009] Furthermore, in the mechanical property prediction method according to the present invention, factors relating to the growth of the solidification structure of the casting (C) can be added to the explanatory variables.

[0010] Furthermore, in the mechanical property prediction method according to the present invention, the amount of change in wall thickness of the casting (C) can be added to the explanatory variables.

[0011] Furthermore, in the mechanical property prediction method according to the present invention, a factor related to the cleanliness of the molten metal can be added to the explanatory variables.

[0012] Furthermore, in the mechanical property prediction method according to the present invention, the solid fraction of the molten metal during flow can be added to the explanatory variables.

[0013] Another mechanical property prediction method according to the present invention is a mechanical property prediction method for predicting the mechanical properties of a casting (C) formed by injecting molten metal into a cavity formed in a pair of openable and closable molds and solidifying the molten metal, and includes an explanatory variable acquisition step (step S2) for performing a predetermined analysis using a mold model for CAE analysis in which the cavity is divided into a plurality of elements to obtain explanatory variables that are information about the casting conditions of the casting (C) or the molten metal, and a mechanical property prediction step (step S4) for inputting the explanatory variables into a prediction model that calculates a correspondence between the explanatory variables and objective variables that are information about the mechanical properties of the casting (C), and predicting the objective variables, wherein the solid fraction of the molten metal during flow is used as the explanatory variable.

[0014] Furthermore, in the mechanical property prediction method of the present invention and other mechanical property prediction methods of the present invention, the prediction model can be obtained from linear regression or ensemble learning, which combines multiple learning models to generate a single learning model.

[0015] A mechanical property prediction device (10) according to the present invention is a mechanical property prediction device (10) that predicts the mechanical properties of a casting (C) formed by pouring molten metal into a cavity formed in a pair of openable and closable molds and solidifying the molten metal, and includes: an explanatory variable acquisition unit (22) that acquires explanatory variables, which are information about the casting conditions of the casting (C) or the molten metal, by performing a predetermined analysis using a mold model for CAE analysis in which the cavity is divided into a plurality of elements; and a mechanical property prediction unit (32) that inputs the explanatory variables into a prediction model that calculates a correspondence between the explanatory variables and objective variables, which are information about the mechanical properties of the casting (C), and predicts the objective variables, and is characterized in that factors related to the amount of solid solution of strengthening elements contained in the casting (C) after solidification are used as the explanatory variables.

[0016] Another mechanical property prediction device (10) according to the present invention is a mechanical property prediction device (10) that predicts the mechanical properties of a casting (C) formed by pouring molten metal into a cavity formed in a pair of openable and closable molds and solidifying the molten metal, and includes: an explanatory variable acquisition unit (22) that acquires explanatory variables, which are information about the casting conditions of the casting (C) or the molten metal, by performing a predetermined analysis using a mold model for CAE analysis in which the cavity is divided into a plurality of elements; and a mechanical property prediction unit (32) that inputs the explanatory variables into a prediction model that calculates a correspondence between the explanatory variables and objective variables, which are information about the mechanical properties of the casting (C), and predicts the objective variables, and is characterized in that the solid fraction of the molten metal during flow is used as the explanatory variables. [Effects of the Invention]

[0017] The present invention provides a method for predicting the mechanical properties of a casting with high accuracy by taking into account various influences that occur on the casting, and also provides a method for easily obtaining good prediction results with a small number of explanatory variables. [Brief explanation of the drawings]

[0018] [Figure 1]FIG. 1 is a functional block diagram of a mechanical property prediction device according to an embodiment of the present invention. [Figure 2] FIG. 1 is a hardware configuration diagram of a mechanical property prediction device according to an embodiment of the present invention. [Figure 3] 1 is a flowchart of a mechanical property prediction method according to an embodiment of the present invention. [Figure 4] FIG. 1 is a schematic diagram of a casting used to obtain predicted values ​​and measured values ​​in a mechanical property prediction method according to an embodiment. [Figure 5] 1 is a graph showing the prediction accuracy of the dependent variable for the conventional example and the mechanical property prediction method according to the present embodiment, where sub-graph (a) in the figure is a graph showing the results for the conventional example, and sub-graph (b) in the figure is a graph showing the results for the mechanical property prediction method according to the present embodiment. [Figure 6] 10 is a graph comparing the predicted values ​​and actual measured values ​​of the objective variable in Example 2, where subgraph (a) in the figure is a graph comparing the predicted values ​​obtained by linear regression with the actual measured values, and subgraph (b) in the figure is a graph comparing the predicted values ​​obtained by ensemble learning with the actual measured values. [Figure 7] 10 is a graph comparing the predicted values ​​and actual measured values ​​of the objective variable in Example 3, where subgraph (a) in the figure is a graph comparing the predicted values ​​obtained by linear regression with the actual measured values, and subgraph (b) in the figure is a graph comparing the predicted values ​​obtained by ensemble learning with the actual measured values. [Figure 8] 10 is a graph comparing the predicted values ​​and actual measured values ​​of the objective variable in Example 4, where subgraph (a) in the figure is a graph comparing the predicted values ​​obtained by linear regression with the actual measured values, and subgraph (b) in the figure is a graph comparing the predicted values ​​obtained by ensemble learning with the actual measured values. [Figure 9] 10 is a graph comparing the predicted values ​​and actual measured values ​​of the objective variable in Example 5, where subgraph (a) in the figure is a graph comparing the predicted values ​​obtained by linear regression with the actual measured values, and subgraph (b) in the figure is a graph comparing the predicted values ​​obtained by ensemble learning with the actual measured values. [Figure 10]10 is a graph comparing the predicted values ​​and actual measured values ​​of the objective variable in Example 6, where subgraph (a) in the figure is a graph comparing the predicted values ​​obtained by linear regression with the actual measured values, and subgraph (b) in the figure is a graph comparing the predicted values ​​obtained by ensemble learning with the actual measured values. [Figure 11] 13 is a graph comparing predicted values ​​obtained by ensemble learning with actual measured values ​​for the dependent variable in Example 7. [Figure 12] 10 is a graph comparing the predicted values ​​and actual measured values ​​of the objective variable in Example 8, where subgraph (a) in the figure is a graph comparing the predicted values ​​obtained by linear regression with the actual measured values, and subgraph (b) in the figure is a graph comparing the predicted values ​​obtained by ensemble learning with the actual measured values. [Figure 13] 10 is a graph comparing the predicted values ​​and actual measured values ​​of the objective variable in Example 9, where subgraph (a) in the figure is a graph comparing the predicted values ​​obtained by linear regression with the actual measured values, and subgraph (b) in the figure is a graph comparing the predicted values ​​obtained by ensemble learning with the actual measured values. [Figure 14] 1 is a graph showing the correlation between the Mg and Si contents and the 0.2% yield strength. [Figure 15] 10 is a graph comparing predicted values ​​obtained by linear regression with actual measured values ​​for the dependent variable in Example 10. DETAILED DESCRIPTION OF THE INVENTION

[0019] Preferred embodiments for carrying out the present invention will be described below with reference to the drawings. Note that the following embodiments do not limit the inventions according to the claims, and not all of the combinations of features described in the embodiments are necessarily essential to the solution of the invention.

[0020] [Mechanical property prediction device] First, the configuration of a mechanical property prediction device 10 according to this embodiment will be described with reference to Figures 1 and 2. Here, Figure 1 is a functional block diagram of the mechanical property prediction device 10 according to this embodiment, and Figure 2 is a hardware configuration diagram of the mechanical property prediction device 10 according to this embodiment.

[0021] 1, a mechanical property prediction device 10 according to this embodiment includes a CAE analysis unit 20 and a model prediction unit 30. The CAE analysis unit 20 includes a mold model creation unit 21 and an explanatory variable acquisition unit 22, and the model prediction unit 30 includes a prediction model acquisition unit 31 and a mechanical property prediction unit 32.

[0022] The mold model creation unit 21 creates a mold model, which is a mesh model for CAE analysis, by dividing a cavity formed in a pair of openable and closable molds into multiple elements, based on information on the model creation conditions and casting conditions. Here, the model creation conditions include design data for the casting, such as three-dimensional CAD (Computer Aided Design) data. Furthermore, the casting conditions include information on the temperature and composition of the molten metal to be poured into the cavity. The division size of the mold model and the shape of the elements can be determined arbitrarily.

[0023] The explanatory variable acquisition unit 22 acquires explanatory variables including information about the casting conditions of the cast product or the molten metal by performing a predetermined analysis using the mold model created by the mold model creation unit 21. In the mechanical property prediction device 10 according to this embodiment, predetermined explanatory variables are obtained by performing either solidification analysis or flow analysis of the molten metal, or both. Note that, in addition to solidification analysis and flow analysis of the molten metal, an analysis method can be freely selected depending on the explanatory variables to be acquired.

[0024] It is known that the mechanical properties of cast products, including die-cast products, are significantly affected by their wall thickness. Therefore, the inventors previously constructed a simple prediction model for the mechanical properties of cast products by using the wall thickness of the cast product as an explanatory variable and fitting a straight line obtained by the least squares method to a scatter plot of the mechanical properties against the wall thickness of the cast product. However, the prediction model had significant plot variance relative to the straight line, leaving room for improvement in prediction accuracy. Therefore, the inventors conducted extensive research to develop new explanatory variables and discovered that accuracy could be improved by incorporating factors related to the amount of strengthening elements dissolved in the cast product after solidification. Specifically, they came up with the idea of ​​using the cooling rate of the cast product after solidification as an explanatory variable, and substituting the temperature of the cast product at the time of removal from the mold as an indicator.

[0025] The inventors also predicted that the accuracy of predicting the mechanical properties of a cast product would be improved by adding additional explanatory variables that are correlated with the mechanical properties of the cast product in addition to the explanatory variables described above. Based on this prediction, the present embodiment employs one or more of the following explanatory variables: "factors related to the amount of dissolved strengthening elements in the cast product after solidification," "factors related to the growth of the solidification structure of the cast product," "change in wall thickness of the cast product," "factors related to the cleanliness of the molten metal," and "solid fraction of the molten metal during flow." The "amount of dissolved strengthening elements" includes information about the amount of the strengthening elements dissolved in the base material of the cast product. The "factors related to the growth of the solidification structure of the cast product" are factors that represent changes in the size of the structure formed by the base material of the cast product. A faster solidification rate of the molten metal correlates with a smaller size of the structure and higher strength.

[0026] Specific combinations of the above factors include using either or both of the "temperature of the casting when it is removed from the mold" and the "temperature of the molten metal at the completion of filling" as the "factor related to the amount of dissolved strengthening elements contained in the casting after solidification," using the "solidification rate" as the "factor related to the growth of the solidified structure after casting," and using the "contact time of the molten metal with air" as the "factor related to the cleanliness of the molten metal."

[0027] The prediction model acquisition unit 31 acquires a prediction model that calculates the correspondence between the explanatory variables acquired by the explanatory variable acquisition unit 22 and the objective variables, which are information related to the mechanical properties of the cast product. Here, the inventors have conducted extensive research to further improve the prediction accuracy, and have come up with the idea of ​​using a machine learning technique, particularly ensemble learning, as the prediction model. Therefore, in the mechanical property prediction device 10 according to this embodiment, a machine learning model using linear regression or ensemble learning is adopted as the prediction model. Here, linear regression is a model in which the correspondence between the explanatory variables and the objective variables is linearly modeled as shown in the following mathematical formula (Mathematical Formula 1). Note that X k is the kth explanatory variable, Y is the target variable, a k represents the coefficient for the kth explanatory variable, and a0 represents a constant.

[0028]

number

[0029] Ensemble learning is a machine learning model that generates a single learning model by combining multiple learning models that have learned the correspondence between explanatory variables and target variables. Here, the correspondence between explanatory variables and target variables in linear regression and ensemble learning is calculated using previously accumulated experimental data. It should be noted that the correspondence can also be calculated using data other than past experimental data, such as publicly available external databases, or by combining these.

[0030] In this embodiment, the ensemble learning uses a random forest, which combines multiple decision trees in parallel as a learning model. Note that the present invention can also use other ensemble learning models, or a method of combining multiple learning models in series.

[0031] The mechanical property prediction unit 32 predicts the dependent variable by inputting the explanatory variables acquired by the explanatory variable acquisition unit 22 into the prediction model acquired by the prediction model acquisition unit 31. In this embodiment, any one of "0.2% proof stress," "tensile strength," and "fracture elongation" is adopted as the dependent variable.

[0032] 2 shows a hardware configuration diagram of the mechanical property prediction device 10 according to this embodiment. The mechanical property prediction device 10 is physically implemented in a computer 40, and is composed of an arithmetic unit 41, an input unit 42, a main memory unit 43, an auxiliary memory unit 44, and an output unit 45.

[0033] The operations of each unit in the mechanical property prediction apparatus 10 are stored in the form of a program in an auxiliary storage device 44, which is composed of a hard disk drive (HDD) and a solid state drive (SSD). The arithmetic unit 41, which is composed of a central processing unit (CPU) and a graphics processing unit (GPU), reads the program from the auxiliary storage device 44, stores it in a main storage device 43, which is composed of a random access memory (RAM) and a read-only memory (ROM), and executes the processing of each unit in the mechanical property prediction apparatus 10 according to the program. The input device 42, which is composed of a mouse, keyboard, etc., can input data related to model creation conditions and casting conditions to the mold model creation unit 21 in the mechanical property prediction apparatus 10. The output device 45, which is composed of a display, etc., can display the prediction results calculated by the mechanical property prediction unit 32. Note that although the mechanical property prediction apparatus 10 according to this embodiment is configured on a single computer 40, it may also be configured on multiple computers.

[0034] The configuration of the mechanical property prediction device 10 according to this embodiment has been described above using Figures 1 and 2. Next, the mechanical property prediction method according to this embodiment will be described using Figure 3. Here, Figure 3 is a flowchart of the mechanical property prediction method according to this embodiment.

[0035] [Mechanical property prediction method] As shown in FIG. 3, the mechanical property prediction method according to this embodiment comprises a mold model creation step (step S1), an explanatory variable acquisition step (step S2), a prediction model acquisition step (step S3), and a mechanical property prediction step (step S4), and each step is executed by a mechanical property prediction device 10.

[0036] In the die model creation step (step S1), various conditions such as model creation conditions and casting conditions are input to the die model creation unit 21. As a result, the die model creation unit 21 creates a mesh model for CAE analysis, which is a die model, based on the various input conditions.

[0037] In the explanatory variable acquisition step (step S2), solidification analysis and flow analysis of molten metal are performed based on the mold model and casting conditions created in the mold model creation step (step S1), thereby obtaining the desired explanatory variables.

[0038] In the prediction model acquisition step (step S3), a prediction model is acquired by the prediction model acquisition unit 31, and then the prediction model is stored from the prediction model acquisition unit 31 in the mechanical property prediction unit 32. Note that the prediction model acquisition step (step S3) only needs to be executed before the mechanical property prediction step (step S4), and can also be executed before the die model creation step (step S1) or the explanatory variable acquisition step (step S2).

[0039] In the mechanical property prediction step (step S4), the explanatory variables acquired in the explanatory variable acquisition step (step S2) are input to the mechanical property prediction unit 32, and a predicted value of the objective variable is output. After the above steps, the mechanical property prediction method according to this embodiment is completed.

[0040] The mechanical property prediction method according to this embodiment has been described above with reference to Fig. 3. Next, in order to verify the prediction accuracy of the predicted values ​​of mechanical properties obtained by the mechanical property prediction method according to this embodiment, the predicted values ​​were compared with the actual measured values ​​obtained by actually measuring the mechanical properties of a cast product. An example for verifying such prediction accuracy will be described below.

[0041] [Verification of prediction accuracy] First, the configuration of the casting C used in the examples will be described with reference to Fig. 4. Here, Fig. 4 is a schematic diagram of the casting C used to obtain predicted values ​​and measured values ​​in the mechanical property prediction method according to the examples.

[0042] The casting C is a die-cast product made of an aluminum alloy and formed by injecting molten metal into a cavity formed in a pair of openable and closable dies and solidifying the molten metal. The casting C may have any shape and may be made of any material.

[0043] The configuration of the casting C used to obtain the predicted values ​​and measured values ​​in the mechanical property prediction method according to this embodiment has been described above with reference to Fig. 4. Below, we will describe the results of verifying the prediction accuracy of the mechanical property prediction method for the casting C while changing the explanatory variables and the target variable.

[0044] [Example 1] First, as Example 1, a comparison was made with a conventional example. Specifically, the prediction accuracy was compared for the "wall thickness of the cast product," a conventional explanatory variable used in simplified prediction of mechanical properties, and the "temperature of the cast product at the time of removal from the mold," an explanatory variable discovered by the inventors. In this comparison, the "0.2% proof stress" of the mechanical properties of cast product C was used as the dependent variable, and predictions were made using linear regression as a prediction model. The results are shown in FIG. 5. FIG. 5 is a graph comparing the prediction accuracy of the dependent variable between the conventional example and the mechanical property prediction method according to this embodiment. Subdivision (a) in the figure is a graph showing the results for the conventional example, and subdivision (b) in the figure is a graph showing the results for the mechanical property prediction method according to this embodiment.

[0045] The dashed lines in Figure 5 show the linear function obtained by linear regression for each plot. The constants and coefficients of the linear function were calculated using the least squares method, and the smaller the plot's variance relative to the dashed line, the higher the prediction accuracy. Here, to examine the prediction accuracy of "0.2% proof stress" for each explanatory variable in Figure 5, the coefficient of determination (R 2 ) was calculated, the coefficient of determination when "wall thickness of casting" was used as an explanatory variable was R 2 = 0.3236, the coefficient of determination when "the temperature of the casting when it is removed from the mold" is used as an explanatory variable is R 2 =0.5893. Here, the coefficient of determination (R 2 ) represents the goodness of fit of the line to the data; the closer this value is to 1, the better the approximation to the data.

[0046] From the above, it was confirmed that by adopting "factors related to the amount of strengthening elements dissolved in the casting after solidification," i.e., "the temperature of the casting at the time of removal from the mold," as an explanatory variable, it is possible to simply predict the mechanical properties of casting C with better accuracy than before.

[0047] [Example 2] In Example 2, as in Example 1, the "0.2% proof stress" was used as the objective variable, and the "factor related to the amount of strengthening elements dissolved in the cast product after solidification," i.e., the "temperature of the cast product at the time of removal from the mold," was used as the explanatory variable. Meanwhile, ensemble learning was used as the prediction model.

[0048] The results of comparing the actual and predicted values ​​of "0.2% proof stress" under these conditions are shown in Figure 6. Here, Figure 6 is a graph comparing the predicted and measured values ​​of the dependent variable in Example 2, where sub-graph (a) in the figure is a graph comparing the predicted values ​​obtained by linear regression with the measured values, and sub-graph (b) in the figure is a graph comparing the predicted values ​​obtained by ensemble learning with the measured values.

[0049] In Figure 6, the actual measured values ​​are plotted against the predicted values. The straight line represents the correspondence when the predicted values ​​and the actual measured values ​​match, and the closer the plot is to the straight line, the higher the prediction accuracy. Here, to examine the variability of the plots relative to the straight line in Figure 6, i.e., the prediction accuracy relative to the actual measured values, the coefficient of determination (R 2 ) was calculated, and the coefficient of determination in linear regression was R 2 =0.5893, and the coefficient of determination in ensemble learning is R 2 =0.8910.

[0050] From the above, it was confirmed that when the "temperature of the casting at the time of removal from the mold" was used as an explanatory variable, the prediction accuracy could be further improved by adopting a prediction model based on ensemble learning rather than linear regression.

[0051] [Example 3] In Example 3, the same objective variables and prediction model as in Example 2 were used, but the explanatory variables were "factors related to the amount of strengthening elements dissolved in the cast product after solidification" and "factors related to the growth of the solidification structure of the cast product." Specifically, "solidification rate" was used as the "factor related to the growth of the solidification structure of the cast product." In addition, "temperature of the cast product at the time of removal from the mold" was used as the "factor related to the amount of strengthening elements dissolved in the cast product after solidification."

[0052] The results of comparing the actual and predicted values ​​of "0.2% proof stress" under these conditions are shown in Figure 7. Here, Figure 7 is a graph comparing the predicted and measured values ​​of the dependent variable in Example 3, where sub-graph (a) in the figure is a graph comparing the predicted values ​​obtained by linear regression with the measured values, and sub-graph (b) in the figure is a graph comparing the predicted values ​​obtained by ensemble learning with the measured values.

[0053] To examine the scatter of the plots relative to the straight lines in Figure 7, i.e., the prediction accuracy relative to the actual measured values, the coefficient of determination (R 2) was calculated, the coefficient of determination in linear regression was R 2 =0.7445, the coefficient of determination in ensemble learning is R 2 = 0.9118. From the above, it was confirmed that the prediction accuracy was improved in all prediction models by adding "factors related to the growth of the solidification structure of the casting," specifically "solidification rate," to the explanatory variables in Example 2.

[0054] [Example 4] Next, in Example 4, the same objective variables and prediction model as in Example 3 were employed, but the explanatory variables were "factors related to the amount of dissolved strengthening elements contained in the casting after solidification," "factors related to the growth of the solidification structure of the casting," and "change in wall thickness of the casting." Specifically, the "temperature of the casting when it is removed from the mold" and "temperature of the molten metal at the completion of filling" were employed as "factors related to the amount of dissolved strengthening elements contained in the casting after solidification," and the "solidification rate" was employed as "factors related to the growth of the solidification structure of the casting."

[0055] The results of comparing the actual and predicted values ​​of "0.2% proof stress" under these conditions are shown in Figure 8. Here, Figure 8 is a graph comparing the predicted and measured values ​​of the dependent variable in Example 4, where sub-graph (a) in the figure is a graph comparing the predicted values ​​obtained by linear regression with the measured values, and sub-graph (b) in the figure is a graph comparing the predicted values ​​obtained by ensemble learning with the measured values.

[0056] To examine the variability of the plots relative to the straight lines in Figure 8, i.e., the prediction accuracy relative to the actual measured values, the coefficient of determination (R 2 ) was calculated, the coefficient of determination in linear regression was R 2 =0.9147, the coefficient of determination in ensemble learning is R 2 = 0.952117. From the above, it was confirmed that the prediction accuracy was improved in all prediction models by adding, to the explanatory variables in Example 3, "the temperature of the molten metal at the end of filling" and "the amount of change in wall thickness of the cast product" from among "factors related to the amount of solid solution of strengthening elements contained in the cast product after solidification."

[0057] [Example 5] Next, in Example 5, the same explanatory variables and prediction model as in Example 4 were employed, while "tensile strength" was employed as the objective variable.

[0058] The results of comparing the actual and predicted values ​​of "tensile strength" under these conditions are shown in Figure 9. Here, Figure 9 is a graph comparing the predicted and measured values ​​of the dependent variable in Example 5, with sub-graph (a) in the figure being a graph comparing the predicted values ​​obtained by linear regression with the measured values, and sub-graph (b) in the figure being a graph comparing the predicted values ​​obtained by ensemble learning with the measured values.

[0059] To examine the variability of the line in Figure 9, i.e., the prediction accuracy relative to the actual measured values, we calculated the coefficient of determination (R 2 ) was calculated, the coefficient of determination in linear regression was R 2 =0.32131, the coefficient of determination in ensemble learning is R 2 = 0.78045. From the above, it was confirmed that when "tensile strength" was used as the objective variable, the prediction accuracy was not good in linear regression, but the prediction accuracy was good in ensemble learning.

[0060] [Example 6] Next, in Example 6, the same explanatory variables and prediction model as in Example 5 were employed, while "elongation at break" was employed as the objective variable.

[0061] The results of comparing the actual and predicted values ​​of "breaking elongation" under these conditions are shown in Figure 10. Here, Figure 10 is a graph comparing the predicted values ​​and actual measured values ​​of the dependent variable in Example 6, where sub-graph (a) in the figure is a graph comparing the predicted values ​​obtained by linear regression with the actual measured values, and sub-graph (b) in the figure is a graph comparing the predicted values ​​obtained by ensemble learning with the actual measured values.

[0062] To examine the scatter of the plots relative to the straight lines in Figure 10, i.e., the prediction accuracy relative to the actual measured values, the coefficient of determination (R 2 ) was calculated, the coefficient of determination in linear regression was R 2 =0.23164, the coefficient of determination in ensemble learning is R 2 = 0.72687. From the above, it was confirmed that when "breaking elongation" was used as the objective variable, the prediction accuracy was not good in linear regression, but the objective variable could be predicted well in ensemble learning.

[0063] [Example 7] Next, in Example 7, the same objective variables and prediction model as in Example 6 were employed, but the explanatory variables were "factors related to the amount of dissolved strengthening elements in the casting after solidification," "factors related to the growth of the solidification structure of the casting," "change in wall thickness of the casting," and "factors related to the cleanliness of the molten metal." Specifically, the "temperature of the casting when it is removed from the mold" and "temperature of the molten metal at the completion of filling" were used as "factors related to the amount of dissolved strengthening elements in the casting after solidification," "solidification rate" as "factors related to the growth of the solidification structure of the casting," and "contact time of the molten metal with air" as "factors related to the cleanliness of the molten metal."

[0064] The results of comparing the actual and predicted values ​​of "breaking elongation" under these conditions are shown in Figure 11. Here, Figure 11 is a graph comparing the predicted values ​​obtained by ensemble learning with the actual measured values ​​for the dependent variable in Example 7.

[0065] To examine the variability of the plots relative to the straight lines in Figure 11, i.e., the prediction accuracy relative to the actual measured values, the coefficient of determination (R 2 ) was calculated, and the coefficient of determination was R 2 = 0.8143. From the above, it was confirmed that adding "factors related to the cleanliness of the molten metal," specifically "contact time of the molten metal with air," to the explanatory variables in Example 6 improved the prediction accuracy in the prediction model using ensemble learning.

[0066] [Example 8] Next, in Example 8, the same objective variables and prediction model as in Example 6 were employed, but the explanatory variables were "factors related to the amount of dissolved strengthening elements contained in the casting after solidification," "factors related to the growth of the solidification structure of the casting," "change in wall thickness of the casting," and "solid fraction during molten metal flow." Specifically, the "temperature of the casting when it is removed from the mold" and "temperature of the molten metal at the completion of filling" were employed as "factors related to the amount of dissolved strengthening elements contained in the casting after solidification," and the "solidification rate" was employed as "factors related to the growth of the solidification structure of the casting."

[0067] The results of comparing the actual and predicted values ​​of "breaking elongation" under these conditions are shown in Figure 12. Here, Figure 12 is a graph comparing the predicted values ​​and actual measured values ​​of the dependent variable in Example 8, with sub-graph (a) in the figure being a graph comparing the predicted values ​​obtained by linear regression with the actual measured values, and sub-graph (b) in the figure being a graph comparing the predicted values ​​obtained by ensemble learning with the actual measured values.

[0068] To examine the variability of the plots relative to the straight lines in Figure 12, i.e., the prediction accuracy relative to the actual measured values, the coefficient of determination (R 2 ) was calculated, the coefficient of determination in linear regression was R 2 =0.7720, the coefficient of determination in ensemble learning is R 2 = 0.87732. From the above, it was confirmed that adding "solid fraction of molten metal when it is flowing" to the explanatory variables in Example 6 improved the prediction accuracy in all prediction models. In particular, as shown in Example 6, when the prediction model was linear regression, the coefficient of determination before adding "solid fraction of molten metal when it is flowing" to the explanatory variables was R 2 =0.23164, it was found that adding this explanatory variable significantly improved the prediction accuracy.

[0069] [Example 9] From the results of Example 8, the inventors came up with the idea that when the objective variable is "fracture elongation," the "solid fraction of the molten metal during flow" is a particularly important explanatory variable. To confirm this, in Example 9, the same objective variable and prediction model as in Example 8 were employed, but only the "solid fraction of the molten metal during flow" was used as the explanatory variable. Figure 13 shows the results of comparing the actual and predicted values ​​of "fracture elongation" under these conditions. Here, Figure 13 is a graph comparing the predicted and measured values ​​of the objective variable in Example 9. Subdivision (a) in the figure is a graph comparing the predicted values ​​obtained by linear regression with the measured values, and subdivision (b) in the figure is a graph comparing the predicted values ​​obtained by ensemble learning with the measured values.

[0070] To examine the variability of the line in Figure 13, i.e., the prediction accuracy relative to the actual measured values, the coefficient of determination (R 2 ) was calculated, the coefficient of determination in linear regression was R 2 =0.6753, the coefficient of determination in ensemble learning is R 2 = 0.9081. From the above, it was confirmed that when the objective variable is "fracture elongation," good prediction accuracy can be achieved in all prediction models by using only "solid fraction of molten metal at the time of flow" as an explanatory variable. In particular, in ensemble learning, it was confirmed that prediction accuracy was improved when only "solid fraction of molten metal at the time of flow" was used as an explanatory variable compared to when the five explanatory variables "temperature of the casting when it is removed from the mold," "temperature of the molten metal at the end of filling," "solidification rate," "change in casting wall thickness," and "solid fraction of molten metal at the time of flow" were used.

[0071] [Example 10] The inventors also came up with the idea that the "amount of the material components contained in the molten metal" is an important explanatory variable that governs the amount of strengthening elements dissolved in the solid solution. To confirm this, the correlation between the "amount of the material components contained in the molten metal" and the "0.2% yield strength" was investigated when the "amount of the material components contained in the molten metal" was used as the explanatory variable and the "0.2% yield strength" was used as the dependent variable. The results are shown in Figures 14 and 15. Specifically, the "amount of the material components contained in the molten metal" was the "contents of Mg and Si." Figure 14 is a graph showing the correlation between the contents of Mg and Si and the 0.2% yield strength. Figure 15 is a graph comparing the predicted values ​​obtained by linear regression with the measured values ​​for the dependent variable in Example 10.

[0072] Referring to Figure 14, when the weight percent concentration of Mg contained in the molten metal is constant, the higher the weight percent concentration of Si contained in the molten metal, the higher the 0.2% proof stress value. Furthermore, when the weight percent concentration of Si is constant, the higher the weight percent concentration of Mg, the higher the 0.2% proof stress value. Therefore, it is clear that there is a correlation between the weight percent concentrations of Mg and Si and the 0.2% proof stress value. In other words, it is clear that there is a correlation between the amounts of the components of the materials contained in the molten metal and the mechanical properties of Casting C. This is thought to be because, as the Mg and Si contents increase, the amount of solid solutions such as Mg2Si in the aluminum matrix that constitutes the molten metal increases, improving the mechanical properties of Casting C.

[0073] To examine the variability of the plots relative to the straight lines in Figure 15, i.e., the prediction accuracy relative to the actual measured values, the coefficient of determination (R 2 ) was calculated, and R 2= 0.9284. From the above, it was confirmed that when the objective variable was "0.2% proof stress," good prediction accuracy could be achieved by using only "the amount of the constituent material contained in the molten metal" as an explanatory variable. In Example 10, the mechanical properties of casting C were predicted using only "the amount of the constituent material contained in the molten metal" as an explanatory variable. However, the inventors have confirmed that prediction accuracy can be further improved by combining this with other explanatory variables. For example, prediction accuracy can be further improved by using, as explanatory variables, one or all of "the temperature of the casting at the time of removal from the mold," "the temperature of the molten metal at the completion of filling," and "the solidification rate" in addition to "the amount of the constituent material contained in the molten metal."

[0074] Although the preferred embodiments of the present invention have been described above, the technical scope of the present invention is not limited to the scope described in the above embodiments. Various modifications and improvements can be made to the above embodiments and examples.

[0075] For example, in the above example, the mechanical properties of the casting C were predicted using up to five explanatory variables, but the inventors have confirmed that good prediction results can be obtained even when any combination of explanatory variables other than those shown in the above example is used.

[0076] In the above examples, "0.2% yield strength," "tensile strength," and "elongation at break" were used as dependent variables, but other mechanical properties may also be used as dependent variables. For example, various mechanical properties such as "shear stress," "hardness," and "density" can be predicted as long as they are correlated with the explanatory variables.

[0077] Although linear regression and ensemble learning are used as the prediction model in this embodiment, other methods may also be used in the present invention, such as neural networks, support vector machines, decision trees, and any other known prediction methods.

[0078] It is clear from the claims that such modifications and improvements may also be included within the technical scope of the present invention. [Explanation of symbols]

[0079] 10 Mechanical property prediction device, 20 CAE analysis unit, 21 Mold model creation unit, 22 Explanatory variable acquisition unit, 30 Model prediction unit, 31 Prediction model acquisition unit, 32 Mechanical property prediction unit, 40 Computer, 41 Arithmetic unit, 42 Input unit, 43 Main memory unit, 44 Auxiliary memory unit, 45 Output unit, C Casting product.

Claims

1. A mechanical property prediction method for predicting mechanical properties of a casting formed by injecting molten metal into a cavity formed in a pair of openable and closable molds and solidifying the molten metal, comprising: an explanatory variable acquisition step of performing a predetermined analysis using a mold model for CAE analysis, which is obtained by dividing the cavity into a plurality of elements, to obtain explanatory variables that are information about the casting conditions of the cast product or the molten metal; a mechanical property prediction step of inputting the explanatory variables into a prediction model that calculates a correspondence relationship between the explanatory variables and a target variable that is information on the mechanical properties of the casting, and predicting the target variable; Including, A method for predicting mechanical properties, characterized in that factors related to the amount of strengthening elements dissolved in the cast product after solidification are used as the explanatory variables.

2. 2. The mechanical property prediction method according to claim 1, A method for predicting mechanical properties, characterized in that factors related to the growth of the solidification structure of the casting are added to the explanatory variables.

3. 3. The mechanical property prediction method according to claim 1 or 2, A method for predicting mechanical properties, characterized in that a change in wall thickness of the casting is added to the explanatory variables.

4. 4. The mechanical property prediction method according to claim 3, A method for predicting mechanical properties, characterized in that a factor related to the cleanliness of the molten metal is added to the explanatory variables.

5. 4. The mechanical property prediction method according to claim 3, A mechanical property prediction method, characterized in that the solid phase fraction of the molten metal during flow is added to the explanatory variables.

6. A mechanical property prediction method for predicting mechanical properties of a casting formed by injecting molten metal into a cavity formed in a pair of openable and closable molds and solidifying the molten metal, comprising: an explanatory variable acquisition step of performing a predetermined analysis using a mold model for CAE analysis, which is obtained by dividing the cavity into a plurality of elements, to obtain explanatory variables that are information about the casting conditions of the cast product or the molten metal; a mechanical property prediction step of inputting the explanatory variables into a prediction model that calculates a correspondence relationship between the explanatory variables and a target variable that is information on the mechanical properties of the casting, and predicting the target variable; Including, A mechanical property prediction method, characterized in that the solid phase fraction of the molten metal during flow is used as the explanatory variable.

7. 10. The mechanical property prediction method according to claim 1 or 6, A mechanical property prediction method characterized in that the prediction model is obtained from linear regression or ensemble learning that combines multiple learning models to generate a single learning model.

8. A mechanical property prediction device that predicts the mechanical properties of a casting formed by injecting molten metal into a cavity formed in a pair of openable and closable molds and solidifying the molten metal, an explanatory variable acquisition unit that acquires explanatory variables that are information about the casting conditions of the casting product or the molten metal by performing a predetermined analysis using a mold model for CAE analysis that is obtained by dividing the cavity into a plurality of elements; a mechanical property prediction unit that inputs the explanatory variables into a prediction model that calculates a correspondence relationship between the explanatory variables and a target variable that is information about the mechanical properties of the casting, and predicts the target variable; Equipped with A mechanical property prediction device characterized in that factors related to the amount of strengthening elements dissolved in the cast product after solidification are used as the explanatory variables.

9. A mechanical property prediction device that predicts the mechanical properties of a casting formed by injecting molten metal into a cavity formed in a pair of openable and closable molds and solidifying the molten metal, an explanatory variable acquisition unit that acquires explanatory variables that are information about the casting conditions of the casting product or the molten metal by performing a predetermined analysis using a mold model for CAE analysis that is obtained by dividing the cavity into a plurality of elements; a mechanical property prediction unit that inputs the explanatory variables into a prediction model that calculates a correspondence relationship between the explanatory variables and a target variable that is information about the mechanical properties of the casting, and predicts the target variable; Equipped with A mechanical property prediction device characterized in that the solid phase fraction of the molten metal during flow is used as the explanatory variable.

Images