Method and apparatus for predicting formability information of steel material

Abstract

A method and an apparatus for predicting formability information of a steel material are disclosed. A method for predicting formability information of a steel material according to an embodiment of the present invention may comprise the steps of: training a machine learning model by using tensile property information of the steel material as input data and formability information as output data; receiving target tensile property information from a user; and outputting predicted formability information relating to the target tensile property information by using the trained machine learning model, wherein the formability information includes at least one of a forming limit curve, a hole expansion ratio, and a fracture strain.

Description

Method and apparatus for predicting forming property information of steel materials

[0001] The present invention relates to a method and apparatus for predicting forming property information of a steel material. More specifically, the present invention relates to a method and apparatus for predicting forming property information based on basic tensile property information using an artificial intelligence-based machine learning model.

[0002] One method for evaluating the performance of steel materials is to obtain various material property information through uniaxial tensile testing. This material property information includes tensile properties, such as yield point, tensile strength, and uniform elongation, as well as formability information, which allows for an intuitive understanding of the material's formability. In this context, formability information serves not only as an important indicator for material selection in industrial settings where steel is actually utilized, but is also essential data for improving the precision of Computer-Aided Engineering (CAE) analysis conducted by research institutions.

[0003] Since the forming properties of steel materials are closely correlated with tensile properties, research on methods to predict forming properties based on tensile data has been ongoing for a long time. However, existing predictions have primarily relied on empirical formulas, requiring significant experimental costs such as extensive experimental equipment and post-processing techniques for experimental data. With the recent rapid advancement of artificial intelligence, there is a pressing need for AI-based methods and devices to predict forming properties of steel materials that can predict them more accurately and allow users to access them more conveniently through computer programs.

[0004] (Patent Document 1) Republic of Korea Registered Patent Publication No. 10-2696205

[0005] The technical problem of the present invention is to provide a method and apparatus for predicting forming property information of a steel material.

[0006] Another technical objective of the present invention is to provide a method and apparatus for predicting forming property information, which is used as an important indicator for material selection in industrial sites where steel materials are utilized, through basic tensile property information.

[0007] Another technical objective of the present invention is to provide a method and apparatus for predicting forming property information of steel materials based on artificial intelligence, which provides high prediction accuracy and enhances convenience by allowing the user to perform the prediction simply by executing a simple executable file.

[0008] The technical problems to be solved by the present invention are not limited to those mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art to which the present invention belongs from the description below.

[0009] According to one aspect of the present invention, a method for predicting forming property information of a steel material comprises the steps of: training a machine learning model using tensile property information of the steel material as input data and forming property information as output data; receiving target tensile property information from a user; and outputting predicted forming property information for the target tensile property information using the trained machine learning model, wherein the forming property information may include at least one of a forming limit curve, a hole expansion rate, and a fracture strain.

[0010] In the method for predicting forming property information of the steel material, the step of training the machine learning model further comprises: a step of performing preprocessing using normalization or one-hot encoding on the input data; a step of training the machine learning model by inputting the preprocessed input data into the machine learning model and outputting forming property information regarding the tensile property information; and a step of evaluating the performance of the machine learning model by comparing the forming property information output by the machine learning model with forming property information not using the machine learning model, wherein the forming property information not using the machine learning model may be received from the user or extracted from a database.

[0011] In the method for predicting forming property information of the above steel material, the step of evaluating the performance may be a step of evaluating using the MAPE (Mean Absolute Percentage Error) evaluation index.

[0012] In the method for predicting forming property information of the above steel material, the machine learning model may be at least one of multiple linear regression, Random Forest, Gradient Boosting, and XG Boost.

[0013] In the method for predicting forming property information of the above steel material, the tensile property information may include at least one of a yield point, tensile stress, uniform elongation, total elongation, work hardening index (n-value), and anisotropy index (r-value).

[0014] According to another aspect of the present invention, a computer-readable recording medium may be provided that records a program for executing a method for predicting forming property information of the steel material.

[0015] According to another aspect of the present invention, a device for predicting forming property information of a steel material comprises a model learning unit that learns a machine learning model using tensile property information of the steel material as input data and forming property information as output data, an input / output unit that receives target tensile property information from a user, and a control unit that controls the output of predicted forming property information for the target tensile property information by utilizing the machine learning model learned from the model learning unit, wherein the input / output unit outputs the output predicted forming property information to the user, and the forming property information may include at least one of a forming limit curve, a hole expansion rate, and a fracture strain.

[0016] In the device for predicting forming property information of the steel material, the model learning unit, when training the machine learning model, performs preprocessing using normalization or one-hot encoding on the input data, inputs the preprocessed input data into the machine learning model, and trains the model by outputting forming property information regarding the tensile property information, and evaluates the performance of the machine learning model by comparing the forming property information output by the machine learning model with forming property information that does not use the machine learning model, wherein the forming property information that does not use the machine learning model can be received from the user through the input / output unit or extracted from a database.

[0017] In the above-mentioned device for predicting forming properties of steel materials, the model learning unit can evaluate the performance of the machine learning model using the MAPE (Mean Absolute Percentage Error) evaluation index.

[0018] In the device for predicting forming property information of the steel material, the machine learning model may be at least one of multiple linear regression, Random Forest, Gradient Boosting, and XG Boost.

[0019] In the device for predicting forming property information of the above steel material, the tensile property information may include at least one of a yield point, tensile stress, uniform elongation, total elongation, work hardening index (n-value), and anisotropy index (r-value).

[0020] According to another aspect of the present invention, a computer program stored on software or a computer-readable medium having executable instructions for performing a method to predict forming property information of a steel material may be provided.

[0021] The features briefly summarized above regarding the present invention are merely exemplary aspects of the detailed description of the present invention that follows and do not limit the scope of the present invention.

[0022] According to the present invention, a method and apparatus for predicting forming property information of a steel material may be provided.

[0023] In addition, according to the present invention, a method and apparatus can be provided for predicting forming property information, which is used as an important indicator for material selection in industrial sites where steel materials are utilized, through basic tensile property information.

[0024] In addition, according to the present invention, a method and apparatus for predicting forming property information of a steel material can be provided, which predicts forming property information of a steel material based on artificial intelligence to provide high prediction accuracy and improves convenience by allowing the user to perform the prediction simply by executing a simple executable file.

[0025] The effects obtainable from the present invention are not limited to those mentioned above, and other unmentioned effects will be clearly understood by those skilled in the art from the description below.

[0026] FIG. 1 is a flowchart illustrating a method for predicting forming property information of a steel material according to one embodiment of the present invention.

[0027] Figure 2 is a diagram illustrating the forming limit curve.

[0028] FIGS. 3, FIGS. 4a, and FIGS. 4b are drawings for explaining the prediction accuracy when predicting a forming limit curve according to one embodiment of the present invention.

[0029] FIGS. 5, FIGS. 6a, and FIGS. 6b are drawings for explaining the prediction accuracy when predicting the hole expansion rate according to one embodiment of the present invention.

[0030] FIGS. 7a and 7b are drawings illustrating the prediction accuracy when predicting the fracture strain in one embodiment of the present invention.

[0031] FIG. 8 is a drawing for explaining a device for predicting forming property information of a steel material according to one embodiment of the present invention.

[0032] Hereinafter, embodiments of the present invention are described in detail with reference to the attached drawings so that those skilled in the art can easily implement the present invention. However, the present invention may be embodied in various different forms and is not limited to the embodiments described herein.

[0033] In describing the embodiments of the present invention, if it is determined that a detailed description of known configurations or functions may obscure the essence of the invention, such detailed description is omitted. Additionally, parts of the drawings unrelated to the description of the invention have been omitted, and similar parts are denoted by similar reference numerals.

[0034] In the present invention, when it is stated that a component is "connected," "combined," or "joined" with another component, this may include not only a direct connection but also an indirect connection in which another component exists in between. Furthermore, when it is stated that a component "includes" or "has" another component, this means that, unless specifically stated otherwise, it does not exclude the other component but may include an additional component.

[0035] In the present invention, terms such as first, second, etc. are used solely for the purpose of distinguishing one component from another component and do not limit the order or importance of the components unless specifically stated otherwise. Accordingly, within the scope of the present invention, a first component in one embodiment may be referred to as a second component in another embodiment, and likewise, a second component in one embodiment may be referred to as a first component in another embodiment.

[0036] In the present invention, distinct components are intended to clearly explain their respective features and do not necessarily imply that the components are separated. That is, multiple components may be integrated to form a single hardware or software unit, or a single component may be distributed to form multiple hardware or software units. Therefore, such integrated or distributed embodiments are included within the scope of the present invention, even without separate mention.

[0037] In the present invention, the components described in various embodiments do not necessarily mean essential components, and some may be optional components. Accordingly, embodiments composed of a subset of the components described in one embodiment are also included within the scope of the present invention. Furthermore, embodiments including additional components in addition to the components described in various embodiments are also included within the scope of the present invention.

[0038] In the method and apparatus for predicting forming property information of a steel material according to one embodiment of the present invention, experimental data of automotive steel sheets ranging from mild steel with the lowest strength to martensitic steel with ultra-high strength and a material thickness in the range of 1.0 to 2.0t were used when constructing the prediction model, and steel materials actually included in the corresponding range may be suitable for application to this embodiment. In addition, all tensile property information used in the method and apparatus for predicting forming property information of a steel material according to one embodiment of the present invention is based on the data of the Rolling Direction (RD) of ASTM E8 standard specimens.

[0039] In addition, the method for predicting forming property information of a steel material according to the present invention may have the effect that, according to one embodiment, an artificial intelligence machine learning model is also provided in the form of Python-based source code, allowing the user to conveniently access it as an .exe executable file.

[0040] Hereinafter, a method and apparatus for predicting forming property information of a steel material according to an embodiment of the present invention will be described with reference to each drawing.

[0041] FIG. 1 is a flowchart illustrating a method for predicting forming property information of a steel material according to one embodiment of the present invention.

[0042] Referring to FIG. 1, each step of a method for predicting forming property information of another steel material is illustrated in one embodiment of the present invention, and each step of the method illustrated in FIG. 1 may be performed by a device for predicting forming property information of a steel material to be described later. Alternatively, each step of the method may be performed by a computer program (e.g., Python-based source code) or a recording medium storing said program. Hereinafter, it will be written assuming that each step of the method illustrated in FIG. 1 is performed by a device for predicting forming property information of a steel material, but as described above, it may be performed by a computer program or a recording medium containing said program according to other embodiments.

[0043] Referring to FIG. 1, a molding property information prediction device according to one embodiment can train a machine learning model by using tensile property information as input data and molding property information therefrom as output data (S101).

[0044] Tensile material property information according to one embodiment may include, but is not limited to, yield point, tensile stress, uniform elongation, total elongation, work hardening index (n-value), anisotropy index (r-value), and material thickness.

[0045] In addition, the forming property information related to the above tensile property information may include, but is not limited to, a forming limit curve (FLC), hole expansion ratio (HER), and fracture strain.

[0046] The above step S101 may include a step of performing preprocessing on the tensile physical property information of the material, which is the input data. Specifically, since the tensile physical property information of the material, which is the input data, may be composed of different scales, normalization may be performed to convert it to the same scale, or one-hot encoding may be performed to convert the microstructure phase information of the steel material into a numerical form. The normalization and one-hot encoding described above are merely examples, and the preprocessing process according to one embodiment of the present invention is not limited to those described above.

[0047] Furthermore, the above step S101 can train the machine learning model by inputting the input data on which the preprocessing has been performed and outputting forming property information corresponding to the tensile property information of the input data. That is, when the input data is input into the machine learning model, training can be performed so that the machine learning model outputs appropriate forming property information according to the input data.

[0048] At this time, the machine learning model according to one embodiment of the present invention may be at least one of multiple linear regression, Random Forest, Gradient Boosting, and XG Boost, but is not necessarily limited thereto.

[0049] That is, model training can be performed for each of the one or more machine learning models described above.

[0050] In addition, the molding material property information prediction device according to one embodiment may evaluate the performance of the machine learning model to be trained when performing step S101. That is, after training one or more machine learning models to output molding material property information according to one embodiment, the performance of the models may be evaluated using predetermined evaluation indicators to select a suitable machine learning model with the smallest error range for each molding material property information.

[0051] In this case, as an example, the MAPE (Mean Absolute Percentage Error) evaluation metric may be utilized to evaluate the machine learning model, but it is not necessarily limited to this. For example, when the MAPE evaluation metric is utilized, the forming property information output when the input data, which is tensile property information, is input into the machine learning model, and the forming property information based on tensile property information that was not utilized by the AI ​​model, can be used to confirm that the AI ​​model with the lowest MAPE value is the AI ​​model suitable for predicting the corresponding forming property information. The forming property information based on tensile property information that was not utilized by the AI ​​model may be information input by a user or extracted from a separate database.

[0052] Hereinafter, an example of training for each type of molding property information will be explained in which a molding property information prediction device according to one embodiment of the present invention trains the machine learning model of step S101.

[0053] First, the forming limit curve can generally have a curve shape as shown in FIG. 2. In the graph shown in FIG. 2, the y-axis represents the major strain (e1 or ε1), and the x-axis represents the minor strain (e2 or ε2). Additionally, the y-intercept of the graph may represent the major strain in the plane strain mode, where the minor strain is zero. In this case, the y-intercept may be referred to as FLC0 in the forming limit curve (FLC). Generally, safe forming is possible only if the major strain and minor strain do not exceed the forming limit curve (FLC) during material deformation. As shown in FIG. 2, the forming limit curve in the plane strain mode is located at the very bottom, which may be an unfavorable condition for material forming, and most forming defects may occur in that deformation mode. Therefore, since the most important value to be considered in the forming limit curve is the y-intercept FLC0, the forming property information prediction device according to one embodiment of the present invention can output FLC0 as the predicted result value.

[0054] In addition, the molding property information prediction device according to one embodiment of the present invention may output all curve shapes by utilizing not only the FLC0 value described above but also the formula according to the example shown in FIG. 2. In the formula shown in FIG. 2, X, K, L, and M may be predetermined rational numbers.

[0055] For example, when predicting the FLC0 value from the forming limit curve (FLC), material thickness, tensile strength, total elongation, and work hardening index, which are material properties highly correlated with FLC0 among the tensile material properties (yield point, tensile strength, uniform elongation, total elongation, work hardening index, anisotropy index, material thickness, etc.), can be used as model variables for model training.

[0056] In addition, since the value ranges of each variable differ, data normalization can be performed during the preprocessing stage. In this case, if Python is used, the Python MinMaxScaler function can be used to convert all variable data into values ​​between 0 and 1.

[0057] In addition, some of the collected data can be used as training data and the remainder as test data to proceed with model training.

[0058] In the following Figures 3, 4a, and 4b, 178 tensile properties of a material and corresponding FLC0 data were collected as initial data, 75% of which were classified as training data and the remaining 25% as test data to perform prediction model training, and the model with the highest performance among multiple linear regression, random forest, gradient boosting, and XG boost was selected as the optimal model using the MAPE value of the test data as a model performance evaluation metric, and the prediction result of the optimal model is shown.

[0059] Referring to Figure 3, it is shown that when a prediction is performed according to the example described above, the prediction error rate for the test data (Test set) is 7.83%, and the prediction error rate for the training data (Train Set) is 2.34%.

[0060] In addition, FIG. 4a shows that the prediction error rate for all 178 data collected in the example described above is 3.73%, and FIG. 4b shows that when the Keeler-Brazier model, which was previously used to predict forming limit curves, is utilized, the prediction error rate is 25.84%. That is, according to the example described above, when the method for predicting forming physical property information according to one example of the present invention is used, there may be a prediction performance effect that is significantly improved compared to the existing method.

[0061] In addition, as an example for predicting hole expansion rate (HER), in addition to the six types of tensile material properties (yield point, tensile strength, uniform elongation, total elongation, work hardening index, and anisotropy index), the phase information of the material can be considered as single-phase or multi-phase. In this case, as an example, one-hot encoding can be used in the preprocessing process for predicting hole expansion rate to convert the phase information into numeric data.

[0062] In the following Figures 5, 6a, and 6b, 119 data points regarding the tensile properties of the material and the corresponding hole expansion rate (HER) were collected as initial data. A total of five types of model variables were adopted, including four types of tensile properties—tensile strength, uniform elongation, total elongation, and anisotropy index—and phase information. Of these, 75% were classified as training data and the remaining 25% as test data to perform prediction model training. When the model with the highest performance among multiple linear regression, random forest, gradient boosting, and XG boosting was selected as the optimal model using the MAPE value of the test data as a model performance evaluation metric, the prediction results of the optimal model are shown.

[0063] Referring to Fig. 5, it is shown that when a prediction is performed according to the example described above, the prediction error rate for the test data (Test set) is 12.69%, and the prediction error rate for the training data (Train Set) is 4.88%.

[0064] In addition, Fig. a shows that the prediction error rate for all 119 data collected in the example described above is 6.85%, and Fig. 6b shows that the prediction error rate is 24.04% when the empirical formula previously used to predict hole expansion rates is utilized. That is, according to the example described above, when the molding property information prediction method according to one example of the present invention is used, there may be a prediction performance effect that is significantly improved compared to the existing method.

[0065] According to another example, in the case of the fracture strain of a material, a fracture limit curve is constructed for each deformation mode, and whether a part will fracture can be predicted. That is, according to one example of the present invention, a fracture strain prediction model based on tensile material property information can be constructed for uniaxial tension and plane strain as deformation modes.

[0066] In the following Figures 7a and 7b, 190 data points regarding the tensile properties of the material and the corresponding fracture strain were collected as initial data. Four variables—material thickness, tensile strength, uniform elongation, and total elongation—were adopted as model variables. Of these, 75% were classified as training data and the remaining 25% as test data to perform prediction model training. When the model with the highest performance among multiple linear regression, random forest, gradient boosting, and XG boosting was selected as the optimal model using the MAPE value of the test data as the model performance evaluation metric, the prediction results of the optimal model are shown.

[0067] When a prediction is performed according to the example described above, FIG. 7a shows that the prediction error rate for the test data (Test set) in tension mode is 3.47%, and the prediction error rate for the training data (Train Set) is 1.33%. Additionally, FIG. 7b shows that the prediction error rate for the test data (Test set) in plane deformation mode is 4.05%, and the prediction error rate for the training data (Train Set) is 3.22%.

[0068] The examples of the forming limit curve, hole expansion rate, and fracture strain described above are merely examples of a method for predicting forming property information based on tensile property information, and the types of model variables used to predict each forming property information and the total number of data points used are not limited to the examples described above.

[0069] Referring again to FIG. 1, a molding property information prediction device according to one embodiment can receive target tensile property information from a user of the device (S102). In order to predict molding property information using an artificial intelligence model that has completed training as described above, the user can input target tensile property information.

[0070] In addition, the molding property information prediction device according to one embodiment can output predicted molding property information for the target tensile property information received from the user in step S102 by utilizing the machine learning model learned in step S101 (S103).

[0071] According to another embodiment, if the above-described method for predicting molding property information is provided in the form of Python source code, it may be cumbersome for the user to install Python-based source code and related packages in order to utilize the above-described machine learning model. Therefore, according to one embodiment, the Joblib library is utilized so that the trained model is permanently saved in a file, allowing the user to immediately utilize the already trained optimal prediction model by loading the saved file without needing to retrain the prediction model. Furthermore, in addition to the above, according to one embodiment, the pyInstaller library is utilized so that the user can easily use it in any PC (Personal Computer) environment without using Python, and the Python source code file may be converted into an .exe executable file and provided.

[0072] Below, the configuration of a molding property information prediction device, which is an example capable of performing the molding property information prediction method described above, is explained.

[0073] FIG. 8 is a drawing for explaining a device (100) for predicting forming properties information of a steel material according to an embodiment of the present invention. Since the device (100) for predicting forming properties information of a steel material shown in FIG. 8 can perform the method for predicting forming properties information of a steel material described above, all of the above descriptions can be equally applied to the device (100) for predicting forming properties information of a steel material.

[0074] Referring to FIG. 8, a device (100) for predicting forming properties of a steel material according to one embodiment may include an input / output unit (110), a model learning unit (120), and a control unit (130), but is not necessarily limited to what is shown and may further include other unillustrated components for predicting forming properties.

[0075] The input / output unit (110) can receive target tensile physical property information from the user.

[0076] The model learning unit (120) can train a machine learning model by using tensile material property information of a steel material as input data and forming material property information as output data. At this time, the machine learning model may be at least one of multiple linear regression, Random Forest, Gradient Boosting, and XG Boost.

[0077] Additionally, the tensile property information may include at least one of a yield point, tensile stress, uniform elongation, total elongation, work hardening index (n-value), and anisotropy index (r-value), and the forming property information may include at least one of a forming limit curve, hole expansion rate, and fracture strain.

[0078] Specifically, the model learning unit (120) can train the machine learning model by performing preprocessing using normalization or one-hot encoding on the input data, inputting the preprocessed input data into the machine learning model, and outputting the molding property information for the tensile property information.

[0079] Additionally, the model learning unit (120) can evaluate the performance of the machine learning model by comparing the molding material property information output by the machine learning model with the molding material property information that does not use the machine learning model. At this time, the molding material property information that does not use the machine learning model may be received from the user through the input / output unit (110) or extracted from a database. Also, as an example, the model learning unit (120) can evaluate the performance of the machine learning model using the MAPE (Mean Absolute Percentage Error) evaluation metric.

[0080] The control unit (130) can control the output of predicted forming property information for the target tensile property information by utilizing a machine learning model learned from the model learning unit (120).

[0081] That is, in the device (100) for predicting forming properties of a steel material according to one embodiment of the present invention, when a user inputs target tensile properties information through the input / output unit (110), a machine learning model that has completed learning by the model learning unit (120) by the control unit (130) is used, and predicted forming properties information can be output and provided to the user.

[0082] The exemplary methods of the present invention are described as a series of operations for clarity of description, but this is not intended to limit the order in which the steps are performed, and if necessary, each step may be performed simultaneously or in a different order. To implement the method according to the present invention, additional steps may be included in addition to the steps exemplified, steps excluding some steps and including the remaining steps, or steps excluding some steps and including additional steps.

[0083] The various embodiments of the present invention are not intended to list all possible combinations but are intended to explain representative aspects of the invention, and the matters described in the various embodiments may be applied independently or in combination of two or more.

[0084] In addition, various embodiments of the present invention may be implemented by hardware, firmware, software, or a combination thereof. In the case of implementation by hardware, it may be implemented by one or more ASICs (Application Specific Integrated Circuits), DSPs (Digital Signal Processors), DSPDs (Digital Signal Processing Devices), PLDs (Programmable Logic Devices), FPGAs (Field Programmable Gate Arrays), general processors, controllers, microcontrollers, microprocessors, etc.

[0085] The scope of the present invention includes software or machine-executable instructions (e.g., an operating system, an application, firmware, a program, etc.) that enable an operation according to a method of various embodiments to be executed on a device or computer, and a non-transitory computer-readable medium on which such software or instructions, etc. are stored and which is executable on a device or computer.

[0086] (Explanation of symbols)

[0087] 100: Device for predicting forming property information of steel materials

[0088] 110: Input / Output Section

[0089] 120: Model learning section

[0090] 130: Control unit

Claims

1. In a method for predicting forming property information of steel materials, A step of training a machine learning model using tensile property information of the above steel material as input data and forming property information as output data; A step of receiving target tensile property information from a user; and The method includes the step of outputting predicted forming property information for the target tensile property information by utilizing the above-mentioned learned machine learning model, and The above molding material property information is, Characterized by including at least one of a forming limit curve, a hole expansion rate, and a fracture strain. Method for predicting forming property information of steel materials.

2. In Paragraph 1, The step of training the above machine learning model is, A step of performing preprocessing on the above input data using normalization or one-hot encoding; A step of inputting the input data on which the above-mentioned preprocessing has been performed into the machine learning model and training it in a manner that outputs the forming property information for the above-mentioned tensile property information; and The method further includes a step of evaluating the performance of the machine learning model by comparing the molding material property information output by the machine learning model with the molding material property information not using the machine learning model. The molding property information that does not use the machine learning model is characterized by being received from the user or extracted from a database. Method for predicting forming property information of steel materials.

3. In Paragraph 2, The step of evaluating the above performance is, Characterized by being a step of evaluating using the MAPE (Mean Absolute Percentage Error) evaluation metric, Method for predicting forming property information of steel materials.

4. In Paragraph 1, The above machine learning model is, Characterized by being at least one of multiple linear regression, Random Forest, Gradient Boosting, and XG Boost, Method for predicting forming property information of steel materials.

5. In Paragraph 1, The above tensile property information is, Characterized by including at least one of yield point, tensile stress, uniform elongation, total elongation, work hardening index (n-value), and anisotropy index (r-value). Method for predicting forming property information of steel materials.

6. A computer-readable recording medium having a program that executes a method for predicting forming property information of a steel material according to any one of paragraphs 1 to 5.

7. In a device for predicting forming property information of steel materials, A model learning unit that trains a machine learning model using tensile property information of the above steel material as input data and forming property information as output data; An input / output unit that receives target tensile physical property information from a user; and It includes a control unit that controls the output of predicted forming property information for the target tensile property information by utilizing a machine learning model learned from the above model learning unit, and The above input / output unit is, Output the above-mentioned predicted molding material property information to the user, and The above molding material property information is, Characterized by including at least one of a forming limit curve, a hole expansion rate, and a fracture strain. Device for predicting forming property information of steel materials.

8. In Paragraph 7, The above model learning unit is, In training the above machine learning model, Preprocessing is performed on the above input data using normalization or one-hot encoding, and The input data on which the above preprocessing has been performed is input into the machine learning model, and the model is trained by outputting the forming property information for the tensile property information. The performance of the machine learning model is evaluated by comparing the molding property information output by the machine learning model with the molding property information not using the machine learning model. The molding material property information that does not use the machine learning model is characterized by being received from the user through the input / output unit or extracted from a database. Device for predicting forming property information of steel materials.

9. In Paragraph 8, The above model learning unit is, Characterized by evaluating the performance of the machine learning model using the MAPE (Mean Absolute Percentage Error) evaluation metric. Device for predicting forming property information of steel materials.

10. In Paragraph 7, The above machine learning model is, Characterized by being at least one of multiple linear regression, Random Forest, Gradient Boosting, and XG Boost, Device for predicting forming property information of steel materials.

11. In Paragraph 7, The above tensile property information is, Characterized by including at least one of yield point, tensile stress, uniform elongation, total elongation, work hardening index (n-value), and anisotropy index (r-value). Device for predicting forming property information of steel materials.

Images