Identification of viscoelastic vibration systems expressed by third-order differential equations using neural networks

2023 Fiscal Year Final Research Report

• Project/Area Number: 22K20409
• Research Category: Grant-in-Aid for Research Activity Start-up
• Allocation Type: Multi-year Fund
• Review Section: 0301:Mechanics of materials, production engineering, design engineering, fluid engineering, thermal engineering, mechanical dynamics, robotics, aerospace engineering, marine and maritime engineering, and related fields
• Research Institution: Toyohashi University of Technology
• Principal Investigator: Tajiri Daiki 豊橋技術科学大学, 工学(系)研究科(研究院), 助教 (90944124)
• Project Period (FY): 2022-08-31 – 2024-03-31
• Keywords: 粘弾性材 / 非線形 / 同定 / ニューラルネットワーク / 3階微分方程式

Outline of Final Research Achievements

We developed a method to represent a viscoelastic vibration system using a mathematical model that introduces mass into the three-element model most commonly used in materials engineering, and to identify its parameters using a neural network. In numerical simulations, the steady-response obtained from the equation of motion was input into the neural network, and linear parameters and nonlinear forces were identified, confirming that valid results could be obtained. In experiments, the target system was a beam made of cold-rolled steel plate with urethane rubber bonded to it, and parameters were identified from the measured steady-response. In this verification, parameters that seem correct were obtained, but the rubber used in the verification had high rigidity, and it was found that the system would not generate nonlinear vibrations. In the future, we will proceed with experimental verification using soft viscoelastic materials.

Free Research Field

振動工学

Academic Significance and Societal Importance of the Research Achievements

本研究課題は，機械や構造物の動的設計に関する研究であり，産業の基盤をつくることに役立つ技術課題である．学術的には，粘弾性材の動的特性をどのような数学モデルで表現するか，どのように部材特性を同定するかという課題を解決するもので，逆問題の研究に属する．本研究は対象とする系の応答から部材特性を同定するが，一意に同定することは非常に難しい．その困難を，対象系の特徴を適当に表現する数学モデルとニューラルネットワークにより解決している．社会的には，粘弾性材を含む制振鋼板が用いられる機械や構造物の動的設計において，材料工学と振動工学の数学モデルを共通化することで，部材特性の容易な評価が期待できる．