Material property prediction program, structure, material property prediction device, and material property prediction method

Abstract

To provide a material characteristic prediction program, a structure, a material characteristic prediction device, and a material characteristic prediction method that output a relation between a chemical composition of solidified metal and material characteristics of the metal without performing an experiment.SOLUTION: In a material characteristic prediction device 10, a material characteristic prediction program 100a comprises a first display control function 101a, a factor data acquisition function 104a, and a solidification brittleness temperature range prediction function 105a. The factor data acquisition function 104a acquires at least one of factor data representing factors affecting a solidification brittleness temperature range of a metal material. The solidification brittleness temperature range prediction function 105a inputs the factor data to a machine learning device 20, and causes the machine learning device 20 to predict a solidification brittleness temperature range of molten metal based upon factors represented with the factor data, and causes the machine learning device to output solidification brittleness temperature range data representing a result of prediction of the solidification brittleness temperature range of the metal material.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] The present invention relates to a material property prediction program, a structure, a material property prediction device, and a material property prediction method. [Background technology]

[0002] Currently, welding is used in a variety of fields, including the manufacturing of industrial products and the construction of buildings. Welding is a technique in which a portion of the base materials to be joined or a filler metal is melted and then solidified to form a weld metal, which metallurgically joins multiple base materials together. When welding multiple base materials, it is necessary to form a weld metal with a chemical composition that provides material properties that enable the welded parts to withstand the conditions under which they will be used. For example, Patent Document 1 discloses the chemical composition of the weld metal of a steel arc-welded joint that has excellent fatigue strength. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2007-290032 Summary of the Invention [Problem to be solved by the invention]

[0004] However, experiments to determine the chemical composition of a weld metal that can withstand the conditions under which a welded part is used must be conducted for each weld metal with a huge variety of chemical compositions, which places a heavy burden on the experimenter. This also applies to metallic materials cast using a continuous casting machine. That is, experiments to determine the chemical composition of a metallic material that can withstand the conditions under which a metallic material cast using a continuous casting machine is used must be conducted for each metallic material with a huge variety of chemical compositions, which places a heavy burden on the experimenter.

[0005] The present invention has been made in consideration of the above-mentioned circumstances, and provides a material property prediction program, a structure, a material property prediction device, and a material property prediction method that can output the relationship between the chemical composition of a solidified metal and the material properties of the metal without conducting experiments. [Means for solving the problem]

[0006] One aspect of the present invention is to describe factors that affect the solidification brittle temperature range of metallic materials. Data expressed as at least one of chromium-nickel equivalent ratio and mass percent content of selected elements including chromium, molybdenum, silicon, nickel, manganese, copper, carbon, nitrogen, phosphorus, sulfur, niobium, titanium, zirconium, and boron. The material property prediction program causes a computer to realize a factor data acquisition function that acquires at least one piece of factor data, and a solidification brittle temperature range prediction function that inputs the factor data into a machine learning device, causes the machine learning device to predict the solidification brittle temperature range of the metallic material based on the factors indicated by the factor data, and causes the machine learning device to output solidification brittle temperature range data that indicates the predicted result of the solidification brittle temperature range of the metallic material.

[0007] A material property prediction program according to one embodiment of the present invention further causes a computer to execute a first display control function that displays on a display a first user interface for inputting values ​​of factors indicated by the factor data, and the factor data acquisition function acquires the factor data indicating the values ​​input to the first user interface.

[0008] A material property prediction program according to one embodiment of the present invention further causes a computer to implement a chemical composition data acquisition function that acquires chemical composition data indicating at least a part of the chemical composition of the metallic material, and a factor data generation function that generates the factor data by calculating factors that affect the solidification brittleness temperature range of the metallic material based on the chemical composition indicated by the chemical composition data, and the factor data acquisition function acquires the factor data generated by the factor data generation function.

[0009] A material property prediction program according to one embodiment of the present invention further causes a computer to execute a second display control function that displays on a display a second user interface for inputting a chemical composition indicated by the chemical composition data, and the chemical composition data acquisition function acquires the chemical composition data indicating the chemical composition input to the second user interface.

[0010] In a material property prediction program according to one embodiment of the present invention, the factor data acquisition function acquires at least one of the factor data indicating factors expressed by the following formulas (1) to (3), the factor data indicating factors expressed by the following formulas (4) and (5), the factor data indicating factors expressed by the following formula (6), and the factor data indicating factors expressed by the following formula (7).

[0011]

number

[0012]

number

[0013]

number

[0014]

number

[0015]

number

[0016]

number

[0017]

number

[0018] One aspect of the present invention is a structure at least a portion of which is formed from a metallic material whose solidification brittle temperature range has been predicted using any one of the material property prediction programs described above.

[0019] One aspect of the present invention is to describe factors that affect the solidification brittle temperature range of metallic materials. data expressed as at least one of chromium-nickel equivalent ratio and mass percent content of selected elements including chromium, molybdenum, silicon, nickel, manganese, copper, carbon, nitrogen, phosphorus, sulfur, niobium, titanium, zirconium, and boron; The material property prediction device includes a factor data acquisition unit that acquires at least one piece of factor data, and a solidification brittle temperature range prediction unit that inputs the factor data into a machine learning device, causes the machine learning device to predict the solidification brittle temperature range of the metallic material based on the factors indicated by the factor data, and causes the machine learning device to output solidification brittle temperature range data that indicates the predicted result of the solidification brittle temperature range of the metallic material.

[0020] One aspect of the present invention is to describe factors that affect the solidification brittle temperature range of metallic materials. data expressed as at least one of chromium-nickel equivalent ratio and mass percent content of selected elements including chromium, molybdenum, silicon, nickel, manganese, copper, carbon, nitrogen, phosphorus, sulfur, niobium, titanium, zirconium, and boron; A material property prediction method comprising: acquiring at least one piece of factor data; inputting the factor data into a machine learning device; causing the machine learning device to predict the solidification brittle temperature range of the metallic material based on the factors indicated by the factor data; and causing the machine learning device to output solidification brittle temperature range data indicating the predicted result of the solidification brittle temperature range of the metallic material. [Effects of the Invention]

[0021] According to the present invention, it is possible to output the relationship between the chemical composition of a solidified metal and the material properties of the metal without conducting an experiment. [Brief explanation of the drawings]

[0022] [Figure 1] 1 is a diagram illustrating an example of a material property prediction device that executes a material property prediction program according to a first embodiment. [Figure 2] FIG. 2 is a diagram illustrating an example of the functional configuration of a material property prediction program according to the first embodiment. [Figure 3]FIG. 2 is a diagram showing an example of a weld metal according to the first embodiment. [Figure 4] FIG. 2 is a diagram showing an example of training data input to a machine learning model implemented in the machine learning device according to the first embodiment. [Figure 5] 5 is a diagram showing an example of the chemical composition of the weld metal that provides the training data shown in FIG. 4 and the minimum, maximum and average values ​​of the solidification brittle temperature range. FIG. [Figure 6] 4 is a flowchart illustrating an example of processing executed by a material property prediction program according to the first embodiment. [Figure 7] FIG. 1 is a diagram showing an example of the results of predicting the relationship between the mass percentage of phosphorus contained in an Fe-25Cr-22Ni alloy and the solidification brittle temperature range using the material property prediction device according to the first embodiment. [Figure 8] FIG. 1 is a diagram showing an example of the results of predicting the relationship between the mass percentage of nickel contained in an Fe-20Cr-Ni alloy and the solidification brittle temperature range using the material property prediction device according to the first embodiment. [Figure 9] FIG. 10 is a diagram illustrating an example of the functional configuration of a material property prediction program according to a second embodiment. [Figure 10] 10 is a flowchart illustrating an example of processing executed by a material property prediction program according to a second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0023] [First embodiment] A material property prediction device and a material property prediction program according to a first embodiment will be described with reference to Figures 1 to 7. Figure 1 is a diagram showing an example of a material property prediction device that executes a material property prediction program according to the first embodiment. As shown in Figure 1, a material property prediction device 10 includes a processor 11, a main memory device 12, a communication interface 13, an auxiliary memory device 14, an input / output device 15, and a bus 16.

[0024] The processor 11 is, for example, a CPU (Central Processing Unit), which reads and executes a material property prediction program 100a (described later) to realize each function of the material property prediction program 100a. The processor 11 may also read and execute a program other than the material property prediction program 100a to realize a function required to realize each function of the material property prediction program 100a.

[0025] The main storage device 12 is, for example, a RAM (Random Access Memory), and stores in advance the material property prediction program 100a and other programs that are read and executed by the processor 11.

[0026] The communication interface 13 is an interface circuit for communicating with the machine learning device 20 and other devices via the communication network NW shown in Fig. 1. The communication network NW is, for example, a local area network (LAN), a wide area network (WAN), an intranet, or the Internet.

[0027] The auxiliary storage device 14 is, for example, a hard disk drive (HDD), a solid state drive (SSD), a flash memory, or a read only memory (ROM).

[0028] The input / output device 15 is, for example, an input / output port. For example, the keyboard 30, mouse 40, and display 50 shown in FIG. 1 are connected to the input / output device 15. The keyboard 30 and mouse 40 are used, for example, to operate the material property prediction apparatus 10 and input data to the material property prediction apparatus 10. The display 50 is used, for example, by the material property prediction apparatus 10 to present information to a user.

[0029] The bus 16 connects the processor 11, the main memory device 12, the communication interface 13, the auxiliary memory device 14, and the input / output device 15 so that data can be transmitted and received among them.

[0030] Fig. 2 is a diagram showing an example of the functional configuration of the material property prediction program according to the first embodiment. As shown in Fig. 2, the material property prediction program 100a includes a first display control function 101a, a factor data acquisition function 104a, and a solidification brittle temperature range prediction function 105a. The first display control function 101a, the factor data acquisition function 104a, and the solidification brittle temperature range prediction function 105a are all realized by the processor 11 reading and executing the material property prediction program 100a stored in the main storage device 12 or the auxiliary storage device 14.

[0031] The first display control function 101a causes the display 50 to display a first user interface for inputting values ​​of factors that affect the solidification brittleness temperature range (BTR) of a metallic material.

[0032] The metallic material referred to here is, for example, a weld metal. FIG. 3 is a diagram showing an example of a weld metal according to the first embodiment. The weld metal W shown in FIG. 3 is formed by applying heat to an end B10 of a base metal B1 and an end B20 of a base metal B2, melting them, and then solidifying them, thereby metallurgically joining the base metals B1 and B2. The chemical composition of the weld metal W depends on the chemical composition of the base metal B1, the amount of the melted base metal B1, the chemical composition of the base metal B2, and the amount of the melted base metal B2. Furthermore, when a filler metal is used in welding, the chemical composition of the weld metal W depends on the chemical composition of the filler metal and the amount of the melted filler metal. Furthermore, the base metals B1 and B2 are, for example, stainless steel, and may have the same chemical composition or different chemical compositions. In the following description, the weld metal W is used as an example of a metallic material.

[0033] The factor affecting the solidification brittle temperature range is, for example, the chromium-nickel equivalent ratio expressed by the following formula (8). eq " and "Ni eq " are expressed by formula (9) and formula (10), respectively. Formula (9) is a formula representing the chromium equivalent. Furthermore, "Cr", "Mo", and "Si" in formula (9) represent the mass percentages of chromium, molybdenum, and silicon, respectively. Formula (10) is a formula representing the nickel equivalent. Furthermore, "Ni", "Mn", "C", and "N" in formula (10) represent the mass percentages of nickel, manganese, carbon, and nitrogen, respectively.

[0034]

number

[0035]

number

[0036]

number

[0037] Alternatively, the factor affecting the solidification brittle temperature range may be a factor expressed by the following formula (11): Calc " is expressed by formula (12). In addition, "Cr," "Mo," "Si," "Ni," "Mn," "Cu," "C," and "N" included in formula (12) represent the mass percent contents of chromium, molybdenum, silicon, nickel, manganese, copper, carbon, and nitrogen, respectively.

[0038]

number

[0039]

number

[0040] Alternatively, the factor affecting the solidification brittle temperature range may be a factor represented by the following formula (13): In addition, "P" and "S" included in formula (13) represent the mass percentages of phosphorus and sulfur, respectively.

[0041]

number

[0042] Alternatively, the factor affecting the solidification brittle temperature range may be a factor expressed by the following formula (14): Formula (14) represents the mass percent of silicon content.

[0043]

number

[0044] The factor data acquisition function 104a acquires at least one factor data indicating a factor affecting the solidification brittle temperature range of the weld metal W. For example, the factor data acquisition function 104a acquires at least one of the factors expressed by Equation (8), Equation (11), Equation (13), and Equation (14) input to the first user interface. Alternatively, the factor data acquisition function 104a may acquire a factor obtained by taking the logarithm of these factors. Note that the factor data acquisition function 104a may acquire factors other than the four factors described above. Examples of such factors include the mass percentages C (mass%), Nb (mass%), Ti (mass%), Zr (mass%), and B (mass%) of carbon, niobium, titanium, zirconium, boron, and the like. Alternatively, a factor such as REM-6×O (mass%), which is expressed using the mass percentage REM (mass%) of rare earth elements and the mass percentage O (mass%) of oxygen, may be used.

[0045] The solidification brittle temperature range prediction function 105a inputs the factor data into the machine learning device 20 shown in Figure 2 and causes the machine learning device 20 to predict the solidification brittle temperature range of the weld metal W based on the factors indicated by the factor data.

[0046] The machine learning device 20 is equipped with a machine learning model 20M that receives factor data acquired by the factor data acquisition function 104a as input and outputs solidification brittle temperature range data indicating the solidification brittle temperature range of the weld metal W. The machine learning model 20M is a machine learning model that performs supervised learning. The machine learning model 20M is realized, for example, by support vector regression (SVR). Support vector regression is a complex model and has low model interpretability, but it is capable of both linear regression and nonlinear regression and has the advantage of being able to achieve high prediction accuracy with less training data than deep learning.

[0047] Before the factor data is input, the machine learning model 20M learns using training data in which the problem is factor data indicating factors identical to or similar to the above-mentioned factors, and the answer is solidification brittle temperature range data indicating the solidification brittle temperature range.

[0048] FIG. 4 is a diagram showing an example of training data input to a machine learning model implemented in a machine learning device according to the first embodiment. The horizontal axis of FIG. 4 represents the chromium-nickel equivalent ratio, which is an example of a training data problem. The vertical axis of FIG. 4 represents the solidification brittle temperature range, which is an example of a training data answer. The open circles in FIG. 4 represent the chromium-nickel equivalent ratio and solidification brittle temperature range indicated by one training data. The training data shown in FIG. 4 was created by collecting experimental data on 81 stainless steels, including 3040 series, 316 series, etc., from academic papers.

[0049] Fig. 5 is a diagram showing an example of the minimum, maximum, and average values ​​of the chemical composition and solidification brittle temperature range of the weld metal that provides the training data shown in Fig. 4. The chemical composition and solidification brittle temperature range of the weld metal that provides the training data shown in Fig. 4 are distributed, for example, within the range shown in Fig. 5.

[0050] In addition, it is preferable that the training data for training the machine learning model 20M is data created using experimental results on a metal material having a chemical composition relatively close to the chemical composition of the metal material for which the solidification brittle temperature range is predicted based on the factors indicated by the factor data.

[0051] Then, the solidification brittle temperature range prediction function 105a outputs solidification brittle temperature range data indicating the result of predicting the solidification brittle temperature range of the weld metal W to the machine learning device 20. The first display control function 101a, for example, displays the solidification brittle temperature range indicated by the solidification brittle temperature range data on the display 50. Furthermore, the weld metal W whose solidification brittle temperature range has been predicted by the solidification brittle temperature range prediction function 105a is used for the purpose of forming structures such as industrial products and buildings.

[0052] Next, an example of processing executed by the material property prediction program 100a according to the first embodiment will be described with reference to Fig. 6. Fig. 6 is a flowchart showing an example of processing executed by the material property prediction program according to the first embodiment.

[0053] In step S11, the first display control function 101a causes the display 50 to display a first user interface for inputting the chemical composition indicated by the chemical composition data.

[0054] In step S14, the factor data acquisition function 104a acquires at least one piece of factor data indicating a factor that affects the solidification brittle temperature range of the metallic material.

[0055] In step S15, the solidification brittle temperature range prediction function 105a inputs the factor data into the machine learning device 20, causes the machine learning device 20 to predict the solidification brittle temperature range of the metallic material based on the factors indicated by the factor data, and causes the machine learning device 20 to output solidification brittle temperature range data indicating the results of the prediction of the solidification brittle temperature range of the metallic material.

[0056] In step S16, the first display control function 101a causes the display 50 to display the solidification brittle temperature range indicated by the solidification brittle temperature range data.

[0057] The above describes the material property prediction program 100a according to the first embodiment. The material property prediction program 100a includes a first display control function 101a, a factor data acquisition function 104a, and a solidification brittleness temperature range prediction function 105a.

[0058] The first display control function 101a causes the display 50 to display a first user interface for inputting values ​​of factors indicated by the factor data. The factor data acquisition function 104a acquires at least one piece of factor data indicating factors that affect the solidification brittle temperature range of the metallic material. The solidification brittle temperature range prediction function 105a causes the machine learning device 20 to predict the solidification brittle temperature range of the metallic material based on the factors indicated by the factor data, and causes the machine learning device 20 to output solidification brittle temperature range data indicating the predicted result of the solidification brittle temperature range of the metallic material.

[0059] In other words, the material property prediction program 100a does not require experiments to actually form metallic materials that can have various chemical compositions and verify the material properties of the metallic materials. Therefore, the material property prediction program 100a can predict and output the relationship between the chemical composition of a metallic material and the material properties of the metallic material without conducting such experiments. Furthermore, an example of such an experiment is an experiment to evaluate the solidification cracking susceptibility of a metallic material.

[0060] Furthermore, unlike techniques that predict the material properties of metallic materials using numerical analysis, the material property prediction program 100a does not use physical property values. Therefore, even if physical property values ​​do not exist or are unreliable, the relationship between the chemical composition of a metallic material and the material properties of the metallic material can be predicted with high accuracy. Note that these effects can be achieved even if the material property prediction program 100a does not have the first display control function 101a.

[0061] Next, specific examples of the effects achieved by the material property prediction program according to the first embodiment will be described with reference to FIGS.

[0062] 7 shows an example of the results of predicting the relationship between the mass percentage of phosphorus contained in an Fe-25Cr-22Ni alloy and the solidification brittle temperature range using the material property prediction device according to the first embodiment. The horizontal axis of FIG. 7 represents the mass percentage of phosphorus contained in the Fe-25Cr-22Ni alloy. The vertical axis of FIG. 7 represents the solidification brittle temperature range of the Fe-25Cr-22Ni alloy.

[0063] Using the material property prediction program 100a, the material property prediction device 10 predicted the relationship between the mass percent of phosphorus contained in the Fe-25Cr-22Ni alloy and the solidification brittle temperature range, as shown by the dotted line in Figure 7. The expansion factor of the solidification brittle temperature range calculated from this prediction result was 2067.9 [°C / mass%P], which accurately matches the experimentally confirmed expansion factor of the solidification brittle temperature range of 1984 [°C / mass%P] (Saida et al., Journal of the Japan Welding Society, 31 (2013), 56-65.). Therefore, the material property prediction program 100a is considered to be a program capable of accurately predicting the solidification brittle temperature range for various chemical compositions of Fe-25Cr-22Ni alloys.

[0064] 8 is a graph showing an example of the results of predicting the relationship between the mass percentage of nickel contained in an Fe-20Cr-Ni alloy and the solidification brittle temperature range using the material property prediction device according to the first embodiment. The horizontal axis of FIG. 8 represents the mass percentage of nickel contained in the Fe-20Cr-Ni alloy. The vertical axis of FIG. 8 represents the solidification brittle temperature range of the Fe-20Cr-Ni alloy.

[0065] Furthermore, "F," "FA," "AF," and "A" shown in Figure 8 all represent the solidification modes described below. F mode is an α single-phase solidification mode in which solidification is completed in a single ferrite phase. FA is a primary α + (α + γ) two-phase solidification mode in which ferrite is generated by solidification, followed by ferrite and austenite. AF is a primary γ + (α + γ) two-phase solidification mode in which austenite is generated by solidification, followed by ferrite and austenite. A is a γ single-phase solidification mode in which solidification is completed in a single austenite phase.

[0066] Using the material property prediction program 100a, the material property prediction device 10 predicted the relationship between the mass percentage of nickel contained in the Fe-20Cr-Ni alloy and the solidification brittle temperature range, as shown by the dotted line in Figure 8. This prediction result is consistent with experimentally confirmed findings that as the mass percentage of nickel contained in the Fe-20Cr-Ni alloy increases, the solidification mode changes in the order of F mode, FA mode, AF mode, and A mode, and that the solidification brittle temperature range is small in the FA mode region and wide in the AF mode or A mode region. Therefore, the material property prediction program 100a is considered to be a program capable of accurately predicting the solidification brittle temperature range for various chemical compositions of Fe-20Cr-Ni alloys.

[0067] [Second embodiment] A material property prediction device and a material property prediction program according to a second embodiment will be described with reference to FIG. 9. FIG. 9 is a diagram showing an example of the functional configuration of the material property prediction program according to the second embodiment. Unlike the material property prediction device 10 and the material property prediction program 100a according to the first embodiment, the material property prediction device and the material property prediction program according to the second embodiment start from chemical composition data (described later) rather than factor data. Therefore, in the description of the second embodiment, differences from the first embodiment will be mainly described, and descriptions of matters overlapping with the first embodiment will be omitted as appropriate.

[0068] As shown in FIG. 9, the material property prediction program 100b includes a second display control function 101b, a chemical composition data acquisition function 102b, a factor data generation function 103b, a factor data acquisition function 104b, and a solidification brittle temperature range prediction function 105b.

[0069] The second display control function 101b causes the display 50 to display a second user interface for inputting the chemical composition of the weld metal W for predicting the solidification brittle temperature range.

[0070] The chemical composition data acquisition function 102b acquires chemical composition data indicating at least a portion of the chemical composition of the weld metal W. For example, the chemical composition data acquisition function 102b acquires chemical composition data indicating the chemical composition input to the second user interface. The chemical composition data indicates, for example, the mass percentage of elements included in the above-mentioned formula (9), formula (10), formula (12), formula (13), or formula (14).

[0071] The factor data generation function 103b generates factor data by calculating factors that affect the solidification brittle temperature range of the weld metal W based on the chemical composition indicated by the chemical composition data. For example, the factor data generation function 103b generates factor data by calculating factors using at least one of the above-mentioned formulas (8) to (14).

[0072] The factor data acquisition function 104b acquires the factor data generated by the factor data generation function 103b.

[0073] 2 and causes the machine learning device 20 to predict the solidification brittle temperature range of the weld metal W based on the factors indicated by the factor data. Then, the solidification brittle temperature range prediction function 105a causes the machine learning device 20 to output solidification brittle temperature range data indicating the results of predicting the solidification brittle temperature range of the weld metal W. The second display control function 101b causes the display 50 to display, for example, the solidification brittle temperature range indicated by the solidification brittle temperature range data.

[0074] Next, an example of processing executed by the material property prediction program 100a according to the second embodiment will be described with reference to Fig. 10. Fig. 10 is a flowchart showing an example of processing executed by the material property prediction program according to the second embodiment.

[0075] In step S21, the second display control function 101b causes the display 50 to display a second user interface for inputting the chemical composition indicated by the chemical composition data.

[0076] In step S22, the chemical composition data acquisition function 102b acquires chemical composition data that indicates at least a part of the chemical composition of the metallic material.

[0077] In step S23, the factor data generating function 103b generates factor data by calculating factors that affect the solidification brittle temperature range of the metallic material based on the chemical composition indicated by the chemical composition data.

[0078] In step S24, the factor data acquisition function 104b acquires at least one piece of factor data.

[0079] In step S25, the solidification brittle temperature range prediction function 105b inputs the factor data into the machine learning device 20, causes the machine learning device 20 to predict the solidification brittle temperature range of the metallic material based on the factors indicated by the factor data, and causes the machine learning device 20 to output solidification brittle temperature range data indicating the results of the prediction of the solidification brittle temperature range of the metallic material.

[0080] In step S26, the second display control function 101b causes the display 50 to display the solidification brittle temperature range indicated by the solidification brittle temperature range data.

[0081] The material property prediction program 100b according to the second embodiment has been described above. The material property prediction program 100b includes a second display control function 101b, a chemical composition data acquisition function 102b, a factor data generation function 103b, a factor data acquisition function 104b, and a solidification brittleness temperature range prediction function 105b.

[0082] The second display control function 101b causes the display 50 to display a second user interface for inputting the chemical composition of the metallic material whose solidification brittle temperature range is to be predicted. The chemical composition data acquisition function 102b acquires chemical composition data indicating at least a portion of the chemical composition of the metallic material. The factor data generation function 103b generates factor data by calculating factors that affect the solidification brittle temperature range of the metallic material based on the chemical composition indicated by the chemical composition data. The factor data acquisition function 104b acquires the factor data generated by the factor data generation function 103b. The solidification brittle temperature range prediction function 105b causes the machine learning device 20 to predict the solidification brittle temperature range of the metallic material based on the factors indicated by the factor data, and outputs solidification brittle temperature range data indicating the predicted result of the solidification brittle temperature range of the metallic material to the machine learning device 20.

[0083] In other words, the material property prediction program 100b does not require experiments to actually form metallic materials that can have various chemical compositions and verify the material properties of the metallic materials. Therefore, the material property prediction program 100a can predict and output the relationship between the chemical composition of a metallic material and the material properties of the metallic material without conducting such experiments.

[0084] Furthermore, unlike techniques that predict the material properties of metallic materials using numerical analysis, the material property prediction program 100b does not use physical property values. Therefore, even if physical property values ​​do not exist or are unreliable, the relationship between the chemical composition of a metallic material and the material properties of the metallic material can be accurately predicted. Note that these effects can be achieved even if the material property prediction program 100b does not have the second display control function 101b.

[0085] At least a part of the functions of the material property prediction program 100a or the material property prediction program 100b may be realized by hardware including circuitry such as an LSI (Large Scale Integration), an ASIC (Application Specific Integrated Circuit), or an FPGA (Field-Programmable Gate Array).

[0086] In the above-described embodiment, the material property prediction program 100a and the material property prediction program 100b predict the solidification brittle temperature range of the weld metal W. However, the present invention is not limited to this. For example, the material property prediction device 10 may predict the solidification brittle temperature range of a metallic material cast using a continuous casting machine.

[0087] In the above-described embodiment, the machine learning model 20M is implemented by support vector regression, but is not limited thereto. The machine learning model 20M may be implemented by, for example, random forest or gradient boosting.

[0088] Although the embodiments of the present invention have been described above in detail with reference to the drawings, the specific configurations of the embodiments of the present invention are not limited to the above-described embodiments, and various combinations, modifications, substitutions, and / or design changes may be added to the above-described embodiments without departing from the spirit and scope of the present invention.

[0089] Furthermore, the effects of the present invention described in the above-described embodiments are merely examples, and therefore the present invention may also achieve other effects that a person skilled in the art can recognize from the description of the above-described embodiments. [Explanation of symbols]

[0090] 10...Material property prediction device, 11...Processor, 12...Main memory device, 13...Communication interface, 14...Auxiliary memory device, 15...Input / output device, 16...Bus, 20...Machine learning device, 20M...Machine learning model, 30...Keyboard, 40...Mouse, 50...Display, 100a, 100b...Material property prediction program, 101a...First display control function, 101b...Second display control function, 102b...Chemical composition data acquisition function, 103b...Factor data generation function, 104a, 104b...Factor data acquisition function, 105a, 105b...Solidification brittleness temperature range prediction function, B1, B2...Base material, B10, B20...End, W...Weld metal

Claims

1. a factor data acquisition function for acquiring at least one piece of factor data representing a factor affecting the solidification brittle temperature range of a metallic material, the factor data being expressed as at least one of a chromium-nickel equivalent ratio and a mass percentage of a predetermined element including chromium, molybdenum, silicon, nickel, manganese, copper, carbon, nitrogen, phosphorus, sulfur, niobium, titanium, zirconium, and boron; a solidification brittle temperature range prediction function that inputs the factor data into a machine learning device, causes the machine learning device to predict the solidification brittle temperature range of the metallic material based on the factors indicated by the factor data, and outputs solidification brittle temperature range data indicating the predicted result of the solidification brittle temperature range of the metallic material to the machine learning device; A material property prediction program that enables computers to achieve this.

2. further executing a first display control function in the computer to display on a display a first user interface for inputting a value of a factor indicated by the factor data; the factor data acquisition function acquires the factor data indicating a value input to the first user interface; The material property prediction program according to claim 1 .

3. a chemical composition data acquisition function for acquiring chemical composition data indicating at least a part of the chemical composition of the metallic material; a factor data generation function that generates the factor data by calculating factors that affect the solidification brittle temperature range of the metallic material based on the chemical composition indicated by the chemical composition data; We will further realize this on computers, the factor data acquisition function acquires the factor data generated by the factor data generation function; The material property prediction program according to claim 1 .

4. further executing a second display control function in the computer to display a second user interface for inputting the chemical composition indicated by the chemical composition data on a display; the chemical composition data acquisition function acquires the chemical composition data indicating the chemical composition input to the second user interface; The material property prediction program according to claim 3 .

5. The factor data acquisition function acquires at least one of the factor data indicating factors expressed by the following formulas (1) to (3), the factor data indicating factors expressed by the following formulas (4) and (5), the factor data indicating factors expressed by the following formula (6), and the factor data indicating factors expressed by the following formula (7):

5. A material property prediction program according to claim 1. [Equation 1] [Equation 2] [Equation 3] [Equation 4] [Equation 5] [Equation 6] [Equation 7]

6. a factor data acquiring unit that acquires at least one piece of factor data that indicates a factor that affects the solidification brittle temperature range of the metallic material, the factor data being data expressed as at least one of a chromium-nickel equivalent ratio and a mass percentage of a predetermined element including chromium, molybdenum, silicon, nickel, manganese, copper, carbon, nitrogen, phosphorus, sulfur, niobium, titanium, zirconium, and boron; a solidification brittle temperature range prediction unit that inputs the factor data into a machine learning device, causes the machine learning device to predict the solidification brittle temperature range of the metallic material based on the factors indicated by the factor data, and causes the machine learning device to output solidification brittle temperature range data indicating the predicted result of the solidification brittle temperature range of the metallic material; A material property prediction device comprising:

7. Obtain at least one piece of factor data that indicates a factor that affects the solidification brittle temperature range of the metallic material, the factor data being expressed as at least one of a chromium-nickel equivalent ratio and a mass percentage of a predetermined element including chromium, molybdenum, silicon, nickel, manganese, copper, carbon, nitrogen, phosphorus, sulfur, niobium, titanium, zirconium, and boron; inputting the factor data into a machine learning device, causing the machine learning device to predict the solidification brittle temperature range of the metallic material based on the factors indicated by the factor data, and causing the machine learning device to output solidification brittle temperature range data indicating the predicted result of the solidification brittle temperature range of the metallic material; A material property prediction method comprising:

Images