TY - GEN
T1 - A PRELIMINARY STUDY OF LEARNING A WAVE ENERGY CONVERTER SYSTEM USING PHYSICS-INFORMED NEURAL NETWORK METHOD
AU - Chen, Bo Chen
AU - Yu, Yi HSiang
N1 - Publisher Copyright:
Copyright © 2023 by ASME.
PY - 2023
Y1 - 2023
N2 - Physics-informed neural network (PINN) is a new type of neural network method that can be used to solve the physical problem by providing a given data set in the machine learning process with embedded physics information directly described by differential equations. A PINN model was applied in this study to solve the governing equation of motion for the analysis of a floating sphere wave energy converter (WEC), and a time-series segmentation approach was implemented to effectively handle the time-dependent problem. A series of PINN simulations are presented in this paper, including a decay test and a set of wave conditions, where the PINN solutions are verified against those obtained from other time-domain numerical models. This work aims to investigate the feasibility of using PINN for WEC applications with a focus on numerical benchmarking, and the studies of collection point resolution, the overall model accuracy, and the corresponding computational efficiency are also investigated.
AB - Physics-informed neural network (PINN) is a new type of neural network method that can be used to solve the physical problem by providing a given data set in the machine learning process with embedded physics information directly described by differential equations. A PINN model was applied in this study to solve the governing equation of motion for the analysis of a floating sphere wave energy converter (WEC), and a time-series segmentation approach was implemented to effectively handle the time-dependent problem. A series of PINN simulations are presented in this paper, including a decay test and a set of wave conditions, where the PINN solutions are verified against those obtained from other time-domain numerical models. This work aims to investigate the feasibility of using PINN for WEC applications with a focus on numerical benchmarking, and the studies of collection point resolution, the overall model accuracy, and the corresponding computational efficiency are also investigated.
KW - Physics-informed neural networks
KW - floating sphere
KW - mass-spring-damper system
KW - time-domain simulation
KW - wave energy converter
UR - http://www.scopus.com/inward/record.url?scp=85173622040&partnerID=8YFLogxK
U2 - 10.1115/OMAE2023-105123
DO - 10.1115/OMAE2023-105123
M3 - Conference contribution
AN - SCOPUS:85173622040
T3 - Proceedings of the International Conference on Offshore Mechanics and Arctic Engineering - OMAE
BT - Ocean Renewable Energy
PB - American Society of Mechanical Engineers (ASME)
T2 - ASME 2023 42nd International Conference on Ocean, Offshore and Arctic Engineering, OMAE 2023
Y2 - 11 June 2023 through 16 June 2023
ER -