Convolutional neural network and long short-term memory based reduced order surrogate for minimal turbulent channel flow

被引:150
作者
Nakamura, Taichi [1 ]
Fukami, Kai [1 ,2 ]
Hasegawa, Kazuto [1 ,3 ]
Nabae, Yusuke [1 ]
Fukagata, Koji [1 ]
机构
[1] Keio Univ, Dept Mech Engn, Yokohama, Kanagawa 2238522, Japan
[2] Univ Calif Los Angeles, Dept Mech & Aerosp Engn, Los Angeles, CA 90095 USA
[3] Politecn Milan, Dipartimento Sci & Tecnol Aerosp, I-20156 Milan, Italy
关键词
NONLINEAR MODE DECOMPOSITION; SUPERRESOLUTION RECONSTRUCTION; PREDICTION;
D O I
10.1063/5.0039845
中图分类号
O3 [力学];
学科分类号
08 ; 0801 ;
摘要
We investigate the applicability of the machine learning based reduced order model (ML-ROM) to three-dimensional complex flows. As an example, we consider a turbulent channel flow at the friction Reynolds number of R e tau = 110 in a minimum domain, which can maintain coherent structures of turbulence. Training datasets are prepared by direct numerical simulation (DNS). The present ML-ROM is constructed by combining a three-dimensional convolutional neural network autoencoder (CNN-AE) and a long short-term memory (LSTM). The CNN-AE works to map high-dimensional flow fields into a low-dimensional latent space. The LSTM is, then, utilized to predict a temporal evolution of the latent vectors obtained by the CNN-AE. The combination of the CNN-AE and LSTM can represent the spatiotemporal high-dimensional dynamics of flow fields by only integrating the temporal evolution of the low-dimensional latent dynamics. The turbulent flow fields reproduced by the present ML-ROM show statistical agreement with the reference DNS data in time-ensemble sense, which can also be found through an orbit-based analysis. Influences of the population of vortical structures contained in the domain and the time interval used for temporal prediction on the ML-ROM performance are also investigated. The potential and limitation of the present ML-ROM for turbulence analysis are discussed at the end of our presentation.
引用
收藏
页数:15
相关论文
共 65 条
[21]  
Guibas L.J, 2020, ARXIV201009469
[22]   CNN-LSTM based reduced order modeling of two-dimensional unsteady flows around a circular cylinder at different Reynolds numbers [J].
Hasegawa, Kazuto ;
Fukami, Kai ;
Murata, Takaaki ;
Fukagata, Koji .
FLUID DYNAMICS RESEARCH, 2020, 52 (06)
[23]   Machine-learning-based reduced-order modeling for unsteady flows around bluff bodies of various shapes [J].
Hasegawa, Kazuto ;
Fukami, Kai ;
Murata, Takaaki ;
Fukagata, Koji .
THEORETICAL AND COMPUTATIONAL FLUID DYNAMICS, 2020, 34 (04) :367-383
[24]  
Hasegawa K, 2019, ASME JSME KSME JOINT
[25]   Reducing the dimensionality of data with neural networks [J].
Hinton, G. E. ;
Salakhutdinov, R. R. .
SCIENCE, 2006, 313 (5786) :504-507
[26]  
Hochreiter S, 1997, NEURAL COMPUT, V9, P1735, DOI [10.1162/neco.1997.9.1.1, 10.1007/978-3-642-24797-2]
[27]   Modeling of transitional channel flow using balanced proper orthogonal decomposition [J].
Ilaka, Milos ;
Rowley, Clarence W. .
PHYSICS OF FLUIDS, 2008, 20 (03)
[28]   Reynolds number effect on wall turbulence: toward effective feedback control [J].
Iwamoto, K ;
Suzuki, Y ;
Kasagi, N .
INTERNATIONAL JOURNAL OF HEAT AND FLUID FLOW, 2002, 23 (05) :678-689
[29]   THE MINIMAL FLOW UNIT IN NEAR-WALL TURBULENCE [J].
JIMENEZ, J ;
MOIN, P .
JOURNAL OF FLUID MECHANICS, 1991, 225 :213-240
[30]   Unsupervised deep learning for super-resolution reconstruction of turbulence [J].
Kim, Hyojin ;
Kim, Junhyuk ;
Won, Sungjin ;
Lee, Changhoon .
JOURNAL OF FLUID MECHANICS, 2021, 910