Deep learning-driven nonlinear reduced-order models for predicting wave-structure interaction
نویسندگان
چکیده
The feasibility of using a Long Short-Term Memory (LSTM) network-driven Non-Intrusive Reduced Order Model (NIROM) to predict the dynamics two-dimensional box floating and interacting with surface water waves is assessed in this study. ground for these wave-structure interactions (WSI) problems, namely displacements hydrodynamic forces arising from wave interaction corresponding particular profile, are computed single-phase Smoothed Particle Hydrodynamics (SPH). dimensionality system first reduced Discrete Empirical Interpolation Method (DEIM) LSTM applied resulting DEIM-LSTM network developing surrogate prediction. This further enhanced by incorporating physics information into loss function physics-informed (LSTM-PINN) predicting rigid body motion. performance networks compared as proof-of-concept demonstration.
منابع مشابه
A Hybrid Optimization Algorithm for Learning Deep Models
Deep learning is one of the subsets of machine learning that is widely used in Artificial Intelligence (AI) field such as natural language processing and machine vision. The learning algorithms require optimization in multiple aspects. Generally, model-based inferences need to solve an optimized problem. In deep learning, the most important problem that can be solved by optimization is neural n...
متن کاملComparisons between reduced order models and full 3D models for fluid-structure interaction problems in haemodynamics
When modeling the cardiovascular system, the effect of the vessel wall on the blood flow has great relevance. Arterial vessels are complex living tissues and three-dimensional specific models have been proposed to represent their behaviour. The numerical simulation of the 3D3D Fluid Structure Interaction (FSI) coupled problem has high computational costs in terms of required time and memory sto...
متن کاملA Hybrid Optimization Algorithm for Learning Deep Models
Deep learning is one of the subsets of machine learning that is widely used in Artificial Intelligence (AI) field such as natural language processing and machine vision. The learning algorithms require optimization in multiple aspects. Generally, model-based inferences need to solve an optimized problem. In deep learning, the most important problem that can be solved by optimization is neural n...
متن کاملReduced Order Modeling for Nonlinear Multi-component Models
Reduced order modeling plays an indispensible role in most real-world complex models. A hybrid application of order reduction methods, introduced previously, has been shown to effectively reduce the computational cost required to find a reduced order model with quantifiable bounds on the reduction errors, which is achieved by hybridizing the application of local variational and global sampling ...
متن کاملReduced-Order Model Design for Nonlinear Smart System Models
The use of finite element or finite difference techniques to discretize nonlinear smart material system models can yield full-order numerical models that accurately characterize the system dynamics but do so at significant computational cost. This can preclude the use of these full-order models for uncertainty analysis, sensitivity analysis, system design, or real-time control implementation. I...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Ocean Engineering
سال: 2023
ISSN: ['1873-5258', '0029-8018']
DOI: https://doi.org/10.1016/j.oceaneng.2023.114511