Session: 08-06-01 non-presentations

Paper Number: 127868

127868 - Research on Fluid-Induced Vibration of a Flexible Cylinder Based on the Physics-Informed Deep Learning 

The fluid-induced vibration (FIV) exhibited by offshore pipelines represents a quintessential and intricate engineering challenge pertaining to fluid-structure interaction (FSI). The conventional approach for physical modeling is highly precise but necessitates substantial computational resources. Conversely, classical data-driven artificial intelligence (AI) models offer computational efficiency but lack physical interpretability. Thus, the integration of data-driven deep learning techniques and physics-based computational fluid dynamics methods becomes imperative. This amalgamation entails the incorporation of the governing equations describing the internal and external fluid and flexible cylinder coupling system into the deep learning model as a regularization term. Subsequently, a physics-informed deep learning model is formulated for the investigation of forward and inverse problems concerning flexible cylinder with internal and external fluid. Within this framework, deep subnetworks are employed to represent higher-order derivative operators, thereby enhancing both the computational efficiency and precision of the established deep learning model. The model, trained using discrete and sparse label data, is employed for prediction (or reconstruction) of flow field and structure information and inversion of unknown parameters (damping, stiffness and flow friction factor). The physics-informed deep learning model, adhering to the principles of physical equations and data distribution, emerges as a promising tool for the exploration of diverse complex FSI problems. This can transcend the constraints imposed by monitoring equipment and methodologies in real-world projects, thereby facilitating a more profound exploration of FIV.

Presenting Author: Yangyang Liao Tongji University

Presenting Author Biography: Yangyang Liao, PhD candidate
Bachelor, Civil Engineering, Central South University, China
Master, Civil Engineering, Tianjin University, China
Doctoral candidate, Civil Engineering, Tongji University, China
Research interests: Fluid-structure interaction, Vortex-induced vibration, AI for science

Authors:

Yangyang Liao Tongji University
Hesheng Tang Tongji University