Session: 02-06-01: Machine Learning in Structural Dynamics and Aeroelasticity

Paper Number: 159109

159109 - Graph Neural Network Assisted Flexible Structure Modeling 

Flexible structures present considerable challenges, primarily due to their geometric large deformations and highly nonlinear dynamics. Conventional finite element methods, while precise, are computationally onerous and frequently prove ineffective in meeting the real-time demands of contemporary applications. To address these challenges, we propose a novel framework for predicting the deformations of flexible structures. This framework employs a Graph Neural Network (GNN) to effectively model the intricate interactions within the structure. Additionally, it incorporates Neural Ordinary Differential Equations (Neural ODE) along with Physics-Informed Neural Network (PINN) to ensure the fidelity of physical laws during long-term simulations. In order to validate the effectiveness of the proposed method, experiments are conducted on commonly used flexible plates. The simulation data for these structures is generated using the Discrete Elastic Plate model. The framework is applied to deformation prediction and optimization control tasks. The experimental results demonstrate that the proposed method significantly accelerates the simulation of flexible structure deformations, reduces computational overhead, and achieves real-time control of flexible structures. The developed machine learning method shows better prediction accuracy and generalizability than traditional pure data-driven approaches. The proposed framework can be used to solve problems in modeling soft robotics, renewable energy systems, and the aeroelasticity of aerospace structures in the future. 

Presenting Author: Leixin Ma Arizona State University

Presenting Author Biography: Leixin Ma is an assistant professor in the Mechanical and Aerospace Engineering program at Arizona State University. She was a postdoctoral fellow in the Department of Mechanical and Aerospace Engineering at UCLA. She was also a senior research personnel at UCLA Clean Energy Smart Manufacturing Innovation Institute. She received her BSc in Naval Architecture & Ocean Engineering from Shanghai Jiao Tong University in 2015, an S.M. degree, and a Ph.D. degree in mechanical engineering from MIT in 2017 and 2021, respectively. Her research interest is in the physics-constrained data-driven approach to studying fluid-structure interaction and mechanics problems. She was awarded the Ho-Ching and Han-Ching Fund Award from the MIT Office of Graduate Education in 2019. She also served as a Teaching Development Fellow at MIT from 2020-2021.