Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
TRAINING METHOD FOR LEARNING DEVICE, METHOD FOR DESIGNING DESIGN PATTERN, METHOD FOR MANUFACTURING LAMINATE, AND DESIGN DEVICE FOR DESIGN PATTERN
Document Type and Number:
WIPO Patent Application WO/2022/158446
Kind Code:
A1
Abstract:
Provided is a training method for a learning device, in which the learning device is trained on a search model for searching for a lamination pattern of a laminate formed by using fiber sheets as unidirectional materials in which the alignment direction of fibers is one direction and by laminating the fiber sheets, wherein the search model serves as a training model in which a value function and a policy function including constraint conditions pertaining to lamination of the fiber sheets are used, and the learning device executes a step for acquiring an initial lamination pattern, which is the lamination pattern of an initial state, and a step for using the initial lamination pattern as an input and training the search model such that the lamination pattern satisfies the constraint conditions.

Inventors:
FUKUDA AKIO (JP)
Application Number:
PCT/JP2022/001599
Publication Date:
July 28, 2022
Filing Date:
January 18, 2022
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
MITSUBISHI HEAVY IND LTD (JP)
International Classes:
B32B37/00; B32B5/12; B32B5/26; G06N99/00
Foreign References:
US20190315076A12019-10-17
JP6625306B12019-12-25
JP2020144484A2020-09-10
JP2019219741A2019-12-26
US20200279152A12020-09-03
JP2018156689A2018-10-04
US20150170022A12015-06-18
Other References:
HIKIDA, SATOSHI: "Experiments on Combinatorial Optimization with Reinforcement Learning Using Deep Learning and Monte Carlo Tree Search and a Consideration of Frame Problem", JSAI TECHNICAL REPORT, no. SIG-AGI-009-007, 1 January 2018 (2018-01-01), pages 1 - 7, XP009538426, DOI: 10.11517/jsaisigtwo.2018.AGI-009_07
NARITA, MINORU (UNIV. TOKYO), KIMURA, DAIKI (IBM RESEARCH AI): "Deep reinforcement method that introduces the degree of failure", THT 37TH ANNUAL CONFERENCE OF THE ROBOTICS SOCIETY OF JAPAN, 3 September 2019 (2019-09-03)
MARUYAMA, MOTOKI; ENDO, SATOSHI; YAMADA, KOJI; TOMA, NARUAKI; AKAMINE, YUHEI: "F-003 A Study of Model-Based Deep Reinforcement Learning Using GAN", PROCEEDINGS OF 18TH FORUM ON INFORMATION TECHNOLOGY (FIT2019); SEPTEMBER 3-5, 2019, vol. 18, no. 2, 1 January 2019 (2019-01-01) - 5 September 2019 (2019-09-05), pages 259 - 260, XP009538343
Attorney, Agent or Firm:
SAKAI INTERNATIONAL PATENT OFFICE (JP)
Download PDF: