• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to previous page

Development of a combined method of graph embedding and machine learning for optimal design of skeletal structures

Research Project

Project/Area Number 21K20461
Research Category

Grant-in-Aid for Research Activity Start-up

Allocation TypeMulti-year Fund
Review Section 0304:Architecture, building engineering, and related fields
Research InstitutionKyoto University

Principal Investigator

Hayashi Kazuki  京都大学, 工学研究科, 助教 (80908757)

Project Period (FY) 2021-08-30 – 2023-03-31
Project Status Completed (Fiscal Year 2022)
Budget Amount *help
¥2,860,000 (Direct Cost: ¥2,200,000、Indirect Cost: ¥660,000)
Fiscal Year 2022: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
Fiscal Year 2021: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
Keywords構造最適化 / 機械学習 / グラフ埋め込み / 幾何学的深層学習 / 離散構造物 / 強化学習 / 教師あり学習 / 建築構造最適化 / 鋼構造平面骨組 / 断面最適化 / 非線形問題
Outline of Research at the Start

複雑な構造性状を有する大規模空間構造など,人間の直感だけでは設計が難しい構造物には最適化技術が用いられる。しかし,既往の最適化アルゴリズムは膨大な繰り返し計算を要し,現実的な時間で解を得ることが困難なことも多い.建築構造設計プロセスにおいて「設計者とコンピュータが協働してより複雑な構造物をより効率的に設計することは可能か?」という問いに対し,建築構造物の部材接続関係を考慮した特徴量を抽出できるグラフ埋め込みを利用した汎用的な機械学習モデルとそのモデルを用いた構造最適化手法を開発する.

Outline of Final Research Achievements

In order to obtain optimal solutions for complex and large-scale design problems for discrete structures with less computational effort, the formulation of graph embedding and machine learning algorithms have been improved. This improvement contributed to developing more efficient and versatile structural optimization workflows. The improved graph modeling and graph embedding methods are applied to various structures such as trusses, steel structural frames, lattice shells, and kerf-bending structures, and demonstrated the efficacy, efficiency, and versatility of the structural optimization workflow encapsulating the proposed method.

Academic Significance and Societal Importance of the Research Achievements

人工知能技術は、入力・出力するデータのトポロジー(データ同士がどのような接続関係を以て連関しているか)を機械学習モデルの中で明示的に考慮することによって性能を劇的に向上させており、データのトポロジーに着目した機械学習手法は幾何学的深層学習という領域でも活発に研究されている。本研究は、建築構造物の部材の複雑な接続関係を明示的に考慮した幾何学的深層学習モデルを構築することにより、機械学習モデルの性能の大幅な向上を実現した。建築分野から独自の機械学習モデルを構築した学術的意義を有するだけでなく、人工知能技術を用いて安全で安心な構造物を効率的に設計するための萌芽的成果であり、大きな社会的意義を有する。

Report

(3 results)
  • 2022 Annual Research Report   Final Research Report ( PDF )
  • 2021 Research-status Report
  • Research Products

    (26 results)

All 2023 2022 2021 Other

All Int'l Joint Research (1 results) Journal Article (6 results) (of which Int'l Joint Research: 2 results,  Peer Reviewed: 6 results,  Open Access: 6 results) Presentation (18 results) (of which Int'l Joint Research: 9 results,  Invited: 1 results) Remarks (1 results)

  • [Int'l Joint Research] Laboratoire Navier(フランス)

    • Related Report
      2022 Annual Research Report
  • [Journal Article] Deep deterministic policy gradient and graph attention network for geometry optimization of latticed shells2023

    • Author(s)
      C. Kupwiwat, K. Hayashi and M. Ohsaki
    • Journal Title

      Applied Intelligence

      Volume: - Issue: 17 Pages: 19809-19826

    • DOI

      10.1007/s10489-023-04565-w

    • Related Report
      2022 Annual Research Report
    • Peer Reviewed / Open Access / Int'l Joint Research
  • [Journal Article] DEEP DETERMINISTIC POLICY GRADIENT AND GRAPH CONVOLUTIONAL NETWORKS FOR TOPOLOGY OPTIMIZATION OF BRACED STEEL FRAMES2023

    • Author(s)
      Chi-tathon Kupwiwat, Yuichi Iwagoe, Kazuki Hayashi, Makoto Ohsaki
    • Journal Title

      Journal of Structural Engineering B

      Volume: 69B Issue: 0 Pages: 129-139

    • DOI

      10.3130/aijjse.69B.0_129

    • ISSN
      0910-8033, 2436-6285
    • Related Report
      2022 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Deep deterministic policy gradient and graph convolutional network for bracing direction optimization of grid shells2022

    • Author(s)
      Chi-tathon Kupwiwat, Kazuki Hayashi, Makoto Ohsaki
    • Journal Title

      Frontiers in Built Environment

      Volume: 8

    • DOI

      10.3389/fbuil.2022.899072

    • Related Report
      2022 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Assembly sequence optimization of spatial trusses using graph embedding and reinforcement learning2022

    • Author(s)
      K. Hayashi, M. Ohsaki and M. Kotera
    • Journal Title

      J. Int. Assoc. Shell Spatial. Struct.

      Volume: 63-4 Issue: 4 Pages: 232-240

    • DOI

      10.20898/j.iass.2022.016

    • Related Report
      2022 Annual Research Report
    • Peer Reviewed / Open Access
  • [Journal Article] Deep reinforcement learning-based critical element identification and demolition planning of frame structures2022

    • Author(s)
      Shaojun Zhu, Makoto Ohsaki, Kazuki Hayashi, Shaohan Zong, Xiaonong Guo
    • Journal Title

      Frontiers of Structural and Civil Engineering

      Volume: 16 Issue: 11 Pages: 1397-1414

    • DOI

      10.1007/s11709-022-0860-y

    • Related Report
      2022 Annual Research Report
    • Peer Reviewed / Open Access / Int'l Joint Research
  • [Journal Article] Graph-based reinforcement learning for discrete cross-section optimization of planar steel frames2022

    • Author(s)
      Hayashi Kazuki、Ohsaki Makoto
    • Journal Title

      Advanced Engineering Informatics

      Volume: 51 Pages: 101512-101512

    • DOI

      10.1016/j.aei.2021.101512

    • Related Report
      2021 Research-status Report
    • Peer Reviewed / Open Access
  • [Presentation] Reinforcement learning and graph representations for optimization of plane steel building frames2023

    • Author(s)
      Makoto Ohsaki, Kazuki Hayashi, Chi-tathon Kupwiwat
    • Organizer
      the 15th World Congress on Structural and Multidisciplinary Optimisation (WCSMO2023)
    • Related Report
      2022 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Sizing optimization of free-form lattice shells using deep deterministic policy gradient and graph convolutional networks2023

    • Author(s)
      Chi-tathon Kupwiwat, Kazuki Hayashi, Makoto Ohsaki
    • Organizer
      IASS Symposium 2023
    • Related Report
      2022 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Multi-objective optimization of 10-bar truss using multi-agent reinforcement learning2023

    • Author(s)
      Chi-tathon Kupwiwat, Kazuki Hayashi, Makoto Ohsaki
    • Organizer
      日本建築学会大会[近畿]
    • Related Report
      2022 Annual Research Report
  • [Presentation] Isogeometric deep learning framework to predict the structural performance of free-form surfaces2023

    • Author(s)
      Kazuki Hayashi, Makoto Ohsaki
    • Organizer
      IASS Symposium 2023
    • Related Report
      2022 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Knowledge extraction of discrete cross-section optimization of planar steel frames using graph-based reinforcement learning2022

    • Author(s)
      Kazuki Hayashi, Makoto Ohsaki
    • Organizer
      Asian Congress of Structural and Multidisciplinary Optimization (ACSMO)
    • Related Report
      2022 Annual Research Report
    • Int'l Joint Research
  • [Presentation] Topology optimization of braced latticed shells using deep deterministic policy gradient and graph convolutional network2022

    • Author(s)
      Chi-tathon Kupwiwat, Kazuki Hayashi, Makoto Ohsaki
    • Organizer
      Asian Congress of Structural and Multidisciplinary Optimization (ACSMO)
    • Related Report
      2022 Annual Research Report
    • Int'l Joint Research
  • [Presentation] 離散構造部材の特徴量を抽出するためのグラフ埋め込みを用いた機械学習モデル2022

    • Author(s)
      林 和希, 大崎 純
    • Organizer
      第66回理論応用力学講演会
    • Related Report
      2022 Annual Research Report 2021 Research-status Report
  • [Presentation] グラフ埋め込みと機械学習による単層ラチスシェルの弾性座屈荷重低減係数の予測2022

    • Author(s)
      林 和希, 大崎 純
    • Organizer
      日本建築学会近畿支部研究発表会
    • Related Report
      2022 Annual Research Report 2021 Research-status Report
  • [Presentation] Geometry Optimization of Lattice Shells using GAT-DDPG with Bezier Surface2022

    • Author(s)
      Chi-tathon Kupwiwat, Kazuki Hayashi, Makoto Ohsaki
    • Organizer
      日本建築学会近畿支部研究発表会
    • Related Report
      2022 Annual Research Report
  • [Presentation] Graph and machine learning-based approach to prediction of ultimate load of latticed shells considering geometric nonlinearity2022

    • Author(s)
      Kazuki Hayashi, Makoto Ohsaki
    • Organizer
      15th World Congress on Computational Mechanics (WCCM-XV) 8th Asian Pacific Congress on Computational Mechanics (APCOM-VIII)
    • Related Report
      2022 Annual Research Report
    • Int'l Joint Research
  • [Presentation] グラフ埋め込みと機械学習を用いたトラスの外力仕事予測モデル2022

    • Author(s)
      中里 桂也, 林 和希, 大崎 純
    • Organizer
      日本建築学会大会[北海道]
    • Related Report
      2022 Annual Research Report
  • [Presentation] ベジエ曲線を用いたカーフベンディングのスリットパターンと梁要素による構造解析モデルの生成手法2022

    • Author(s)
      林 和希, 大崎 純
    • Organizer
      日本建築学会大会[北海道]
    • Related Report
      2022 Annual Research Report
  • [Presentation] Chi-tathon Kupwiwat, Kazuki Hayashi, Makoto Ohsaki2022

    • Author(s)
      Geometry optimization of lattice shells using GAT-DDPG with Bezier surface
    • Organizer
      日本建築学会大会[北海道]
    • Related Report
      2022 Annual Research Report
  • [Presentation] Assembly sequence optimization of spatial trusses using graph embedding and reinforcement learning2022

    • Author(s)
      Kazuki Hayashi, Makoto Ohsaki, Masaya Kotera
    • Organizer
      IASS 2022 Symposium affiliated with APCS 2022 conference
    • Related Report
      2022 Annual Research Report
    • Int'l Joint Research / Invited
  • [Presentation] Knowledge extraction of discrete cross-section optimization of planar steel frames using graph-based reinforcement learning2022

    • Author(s)
      Kazuki Hayashi and Makoto Ohsaki
    • Organizer
      Asian Congress of Structural and Multidisciplinary Optimization
    • Related Report
      2021 Research-status Report
    • Int'l Joint Research
  • [Presentation] Graph and machine learning-based approach to prediction of ultimate load of latticed shells considering geometric nonlinearity2022

    • Author(s)
      Kazuki Hayashi and Makoto Ohsaki
    • Organizer
      15th World Congress on Computational Mechanics (WCCM-XV) 8th Asian Pacific Congress on Computational Mechanics (APCOM-VIII)
    • Related Report
      2021 Research-status Report
    • Int'l Joint Research
  • [Presentation] 部材の逐次的な付加・除去過程を訓練した強化学習エージェントによる平面 トラスの位相最適化2021

    • Author(s)
      林 和希, 大崎 純
    • Organizer
      第44回情報・システム・利用・技術シンポジウム
    • Related Report
      2021 Research-status Report
  • [Presentation] 安定性の評価とグラフ埋め込みによるトラスの施工経路の強化学習2021

    • Author(s)
      小寺 正也, 林 和希, 大崎 純
    • Organizer
      第44回情報・システム・利用・技術シンポジウム
    • Related Report
      2021 Research-status Report
  • [Remarks] ASSEMBLY SEQUENCE PREDICTOR (by Kazuki_Hayashi)

    • URL

      https://www.food4rhino.com/en/app/assembly-sequence-predictor

    • Related Report
      2022 Annual Research Report

URL: 

Published: 2021-10-22   Modified: 2024-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi