Granted Patent B2

US 12,248,742 · App. 17/424,164 · Granted Mar 11, 2025

Inventors: Kazuhiro Nakatsuji (Tokyo, JP); Osamu Yamaguchi (Tokyo, JP); Hiroyuki Takagi (Tokyo, JP)

Assignee: JFE STEEL CORPORATION

G06F30/27B22D11/16C22F1/00G06F18/214G06V10/443

Abstract

Claims (31)

1. A design support method carried out by a design support apparatus comprising a processor, an acquisition interface, and a display for supporting design, of a metal material that has desired characteristics, the design support method comprising:

acquiring, by the processor via the acquisition interface, constraint conditions as input information, the constraint conditions including a range of a chemical composition of the metal material and a range of production conditions of the metal material;

searching, by the processor, for design conditions that satisfy the constraint conditions and yield the desired characteristics using a prediction model for predicting a characteristic value of the metal material from the design conditions, the prediction model being constructed based on past performance data associating the design conditions, which include the chemical composition and production conditions of the metal material, with the characteristic value; and

displaying, by the processor via the display, at least the chemical composition and the production conditions among the design conditions that are searched for by the processor and correspond to the desired characteristics, wherein

the design conditions are searched for so that deviation among a plurality of predicted values based on a plurality of different training data sets is reduced, the deviation among the plurality of predicted values indicating a degree of difference of the predicted value from an average value of the plurality of predicted values, and

the metal material is produced based on at least the chemical composition and the production conditions.

2. The design support method of claim 1 , wherein

the design conditions that are searched for and correspond to the desired characteristics satisfy the constraint conditions.

3. The design support method of claim 1 , wherein the design conditions are also searched for in a new area differing from the past performance data so that a difference between the design conditions that are searched for and the design conditions in the past performance data increases.

4. The design support method of claim 1 , wherein

the design conditions include a feature vector based on image data of a metallic structure of the metal material, and

the design conditions that are searched for include the feature vector.

5. A prediction model generation method for generating the prediction model used in the design support method of claim 1 , the prediction model generation method comprising:

acquiring the past performance data associating the design conditions with the characteristic value; and

constructing the prediction model, for predicting the characteristic value from the design conditions, based on the acquired past performance data.

6. A design support apparatus for supporting design of a metal material that has desired characteristics, the design support apparatus comprising:

an acquisition interface configured to acquire constraint conditions as input information, the constraint conditions including a range of a chemical composition of the metal material and a range of production conditions of the metal material;

a processor configured to search for design conditions that satisfy the constraint conditions and yield the desired characteristics using a prediction model for predicting a characteristic value of the metal material from the design conditions, the prediction model being constructed based on past performance data associating the design conditions, which include the chemical composition and production conditions of the metal material, with the characteristic value; and

a display configured to display at least the chemical composition and the production conditions among the design conditions, searched for by the processor, that correspond to the desired characteristics, wherein

the processor searches for the design conditions so that deviation among a plurality of predicted values based on a plurality of different training data sets is reduced, the deviation among the plurality of predicted values indicating a degree of difference of the predicted value from an average value of the plurality of predicted values, and

the metal material is produced based on at least the chemical composition and the production conditions.

7. The design support method of claim 2 , wherein the design conditions are also searched for in a new area differing from the past performance data so that a difference between the design conditions that are searched for and the design conditions in the past performance data increases.

8. The design support method of claim 2 , wherein

the design conditions include a feature vector based on image data of a metallic structure of the metal material, and

the design conditions that are searched for include the feature vector.

9. The design support method of claim 3 , wherein

the design conditions include a feature vector based on image data of a metallic structure of the metal material, and

the design conditions that are searched for include the feature vector.

10. A prediction model generation method for generating the prediction model used in the design support method of claim 4 , the prediction model generation method comprising:

acquiring the past performance data associating the design conditions with the characteristic value; and

constructing the prediction model, for predicting the characteristic value from the design conditions, based on the acquired past performance data.

Assignments (1)

Priority Claims (1)

Continuity (1)

References Cited (62)