Research Reports of National Institute of Technology, Nagaoka College
Online ISSN : 2432-3241
Print ISSN : 0027-7568
ISSN-L : 0027-7568
Current issue
Displaying 1-10 of 10 articles from this issue
Cover, Contents
Paper
  • Takahiko Kurahashi, Sakura Ichikawa, Masayuki Kishida, Mizuki Ikarashi ...
    2024 Volume 59 Pages 1-5
    Published: 2024
    Released on J-STAGE: November 13, 2024
    JOURNAL FREE ACCESS
    Herein, we present numerical results obtained after performing density-based topology optimization for a maximally stiff structure problem using two-phase materials. A maximally stiff structure was modeled to minimize the strain energy of the structure. The Young’s modulus expressed based on a two-phase material, such as a functionally graded material at the junction point, was used in the density-based topology optimization. The two-phase material was completely separated in the optimized structure, and the strain energies before and after the separation were compared.
    Download PDF (986K)
  • (Considerations for Introducing a Shape Selection Process Using Deep Reinforcement Learning)
    Yudai Sugiyama, Takahiko Kurahashi, Toshihiko Eto
    2024 Volume 59 Pages 6-11
    Published: 2024
    Released on J-STAGE: November 13, 2024
    JOURNAL FREE ACCESS
    In this study, we introduce and consider a shape optimization process using deep reinforcement learning for the drag minimization of an object in an incompressible viscous fluid. Most previous studies on shape optimization have used the adjoint variable method. However, it has been demonstrated that shape optimization results are dependent on the initial shape and that the final shape is not necessarily the optimal shape1). Deep reinforcement learning is a method that learns only from the interactions between an agent and its environment. We believe that this is appropriate for optimization where the correct answer is unknown. Furthermore, if the shape design variables increase, the behavior patterns that an agent can consider will significantly increase. Therefore, we believe that deep learning is more suitable for this purpose.
    Download PDF (811K)
  • Doan Quang Thuan, Takahiko Kurahashi, Hiroaki Arata, Tetsuro Iyama
    2024 Volume 59 Pages 12-19
    Published: 2024
    Released on J-STAGE: November 13, 2024
    JOURNAL FREE ACCESS
    In this study, we present a new practical optimum design method that consists of two steps: finite element analysis (FEA) and design of experiments. The design of experiments is used to generate approximate evaluation functions for controlling the behavior depending on the changes in the design variables of an object structure by finite element analyses. Here, we used a design of experiments to determine the optimal combination of design parameters in the texture analysis for friction coefficient reduction. The effects of the design variables can be calculated based on an orthogonal array of design variable combinations. The approximate evaluation functions were then generated by these effects based on the analysis of variance. The proposed method was found to be an effective and powerful tool for the optimum design of various practical design problems.
    Download PDF (957K)
  • Takahiko Kurahashi, Manaya Nakazawa, Masayuki Kishida, Kousuke Shimizu ...
    2024 Volume 59 Pages 20-25
    Published: 2024
    Released on J-STAGE: November 13, 2024
    JOURNAL FREE ACCESS
    This study examined the effect of varying the approach to obtain the topological derivative and presents the numerical results of level set-based topology optimization for a maximally stiff structure problem. To perform a topology optimization analysis, the performance function was first defined by the strain energy of the structure. The problem was to determine the optimal topology to minimize the performance function under constraint conditions, that is, the governing equation and boundary condition. The adjoint variable method was introduced to address the minimization problem of the performance function under constraint conditions. The optimal topology of the structure was obtained by updating the level-set function, which was achieved by solving the reaction-diffusion equation. The reaction term of the reaction-diffusion equation was expressed by the topological derivative, that is, the gradient of the performance function extended by the adjoint variable and the governing equation with respect to the level-set function. In this study, we varied the method to obtain the topological derivative in level-set-based topology optimization and performed numerical experiments. The finite element method was applied to solve the structural deformation problem.
    Download PDF (1100K)
Note
  • Shinsuke Hoshii
    2024 Volume 59 Pages 26-35
    Published: 2024
    Released on J-STAGE: November 13, 2024
    JOURNAL FREE ACCESS
    本校における独自制度であるプレラボは,低学年からの研究活動によって学習意欲を高めることを主目的として平成27年度から始まった制度であり,萌芽的テーマやセミナーなどを学科横断・学年横断的に全学生および教職員に向けて周知・提案し,取り組みに参加するメンバーを募集して活動が行われることに大きな特徴がある.プレラボは年間10件以上のテーマが提起され,活発な取り組みが進められている.関心のあるテーマであれば学科の枠を越えて他学科の教員らが提案する研究活動にも参加でき,学科横断的な活動が展開されている.プレラボ活動の成果は各種コンテストや学会などでの受賞歴もあり,多くの学生が活躍している.筆者は令和4年度に「テキストマイニングの実践 -社会の中のことばを調べよう-」というテーマのもとで2名の学生とともにプレラボ活動を実施した.テキストマイニングとは,文章データなどを対象として,テキスト同士の関連性や連鎖を見いだすための技術の総称であり,このプレラボ活動では,参加した学生自身が気になる事柄や好きなテーマを選んで,テキストマイニングの手法の一つであるテキスト分析を取り上げ,実践した.本報では,当該テーマのプレラボ活動の実施状況ならびにテキストマイニングによる計量テキスト分析の結果について報告する.
    Download PDF (1030K)
Colophon, Backcover
feedback
Top