This website requires JavaScript.

Forecasting through deep learning and modal decomposition in multi-phase concentric jets

Le\'on MataRodrigo Abad\'ia-HerediaManuel Lopez-MartinJos\'e M. P\'erezSoledad Le Clainche
Dec 2022
摘要
This work presents a set of neural network (NN) models specifically designedfor accurate and efficient fluid dynamics forecasting. In this work, we showhow neural networks training can be improved by reducing data complexitythrough a modal decomposition technique called higher order dynamic modedecomposition (HODMD), which identifies the main structures inside flowdynamics and reconstructs the original flow using only these main structures.This reconstruction has the same number of samples and spatial dimension as theoriginal flow, but with a less complex dynamics and preserving its mainfeatures. We also show the low computational cost required by the proposed NNmodels, both in their training and inference phases. The core idea of this workis to test the limits of applicability of deep learning models to dataforecasting in complex fluid dynamics problems. Generalization capabilities ofthe models are demonstrated by using the same neural network architectures toforecast the future dynamics of four different multi-phase flows. Data setsused to train and test these deep learning models come from Direct NumericalSimulations (DNS) of these flows.
展开全部
图表提取

暂无人提供速读十问回答

论文十问由沈向洋博士提出,鼓励大家带着这十个问题去阅读论文,用有用的信息构建认知模型。写出自己的十问回答,还有机会在当前页面展示哦。

Q1论文试图解决什么问题?
Q2这是否是一个新的问题?
Q3这篇文章要验证一个什么科学假设?
0
被引用
笔记
问答