.png){kind=icon}
Return
A convolutional neural network-based model for reconstructing free surface flow fields
.png){kind=icon}
PRE
AI
DOI：10.1063/5.0248883.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This paper introduces hydrological computational fluid dynamics model (HydroCFD), a deep learning model based on the convolutional neural network U-Net framework designed for reconstructing free surface flow fields. With well-posed boundary and initial conditions, the model rapidly generates a result that approximates the two-dimensional (2D) shallow water equations, significantly improving the efficiency of obtaining free surface flow fields compared to traditional computational fluid dynamics methods. The framework features an input layer that integrates water depth and terrain (hydrological element variables), and incorporates a new loss function, coefficient of variation loss function (CVLoss), based on the variation coefficient to improve accuracy and stability. HydroCFD is trained and validated on two different datasets, open channel flows with a groin, and with an abrupt expansion. Error analysis demonstrated that HydroCFD achieves high precision in reconstructing 2D free surface flow fields. Furthermore, a comparison of six different loss functions reveals that CVLoss contributes to improved accuracy and computational stability.
Keywords:
DEEP
OPTIMIZATION
RECOGNITION
Journal
IF:
4.3
Papers:
2.9W
Citations:
8.0W
Organization
Cited Papers
Online in situ prediction of 3-D flame evolution from its history 2-D projections via deep learning
PREAI
ShareSaveShareSaveShareSaveShareSaveA fast three-dimensional flow field prediction around bluff bodies using deep learning.png)
PHYSICS OF FLUIDS
PHYSICS OF FLUIDSIF4.3
PREAI
ShareSaveShareSave
View more