Data-driven reduced order model with temporal convolutional neural network

Abstract This paper presents a novel model reduction method based on proper orthogonal decomposition and temporal convolutional neural network. The method generates basis functions of the flow field by proper orthogonal decomposition, and the coefficients are taken as the low-dimensional features. Temporal convolutional neural network is used to construct the model for predicting low-dimensional features. In this work, the training data are obtained from high fidelity numerical simulation. Compared with recurrent networks, temporal convolutional neural network is more effective with fewer parameters. The model reduction method developed here depends only on the solution of flow field. The performance of the new reduced order model is evaluated using numerical case: flow past a cylinder. Experimental results illustrate that time cost is reduced by three orders of magnitude, and convolutional architecture is beneficial to construct reduced order model. The speed-up ratio is linear with the computational scale of the numerical simulation.

[1]  Yunlong Li,et al.  Uncertain reduced-order modeling for unsteady aerodynamics with interval parameters and its application on robust flutter boundary prediction , 2017 .

[2]  Li Yueming.  ADVANCES AND PROSPECTS OF THE REDUCED ORDER MODEL FOR UNSTEADY FLOW AND ITS APPLICATION , 2011 .

[3]  Ionel M. Navon,et al.  Non-intrusive reduced order modelling of the Navier-Stokes equations , 2015 .

[4]  Danny C. Sorensen,et al.  Nonlinear Model Reduction via Discrete Empirical Interpolation , 2010, SIAM J. Sci. Comput..

[5]  Yi Li,et al.  A public turbulence database cluster and applications to study Lagrangian evolution of velocity increments in turbulence , 2008, 0804.1703.

[6]  Yoshihide Tominaga,et al.  CFD Simulation of Near-Field Pollutant Dispersion in the Urban Environment: A Review of Current Modeling Techniques , 2013 .

[7]  Guo Jun,et al.  Drop test and crash simulation of a civil airplane fuselage section , 2015 .

[8]  A. Mannarino,et al.  Nonlinear aeroelastic reduced order modeling by recurrent neural networks , 2014 .

[9]  Nikolaos Doulamis,et al.  Deep Learning for Computer Vision: A Brief Review , 2018, Comput. Intell. Neurosci..

[10]  J. Nathan Kutz,et al.  Deep learning in fluid dynamics , 2017, Journal of Fluid Mechanics.

[11]  Weiwei Zhang,et al.  Multi-kernel neural networks for nonlinear unsteady aerodynamic reduced-order modeling , 2017 .

[12]  Yao Zhang,et al.  Application of Convolutional Neural Network to Predict Airfoil Lift Coefficient , 2017, ArXiv.

[13]  Guangjun Gao,et al.  Impact of ground and wheel boundary conditions on numerical simulation of the high-speed train aerodynamic performance , 2016 .

[14]  Christopher C. Pain,et al.  Non-intrusive reduced order modelling of fluid–structure interactions , 2016 .

[15]  Jürgen Schmidhuber,et al.  Long Short-Term Memory , 1997, Neural Computation.

[16]  P. Nair,et al.  Nonintrusive reduced‐order modeling of parametrized time‐dependent partial differential equations , 2013 .

[17]  Min Chen,et al.  A non-intrusive reduced-order model for compressible fluid and fractured solid coupling and its application to blasting , 2017, J. Comput. Phys..

[18]  Boo Cheong Khoo,et al.  Fast flow field prediction over airfoils using deep learning approach , 2019, Physics of Fluids.

[19]  Han Chen,et al.  Blackbox Stencil Interpolation Method for Model Reduction , 2012 .

[20]  Göran Sandberg,et al.  Reduced order modelling of liquid-filled pipe systems , 2016 .

[21]  Shan Tang,et al.  Clustering discretization methods for generation of material performance databases in machine learning and design optimization , 2019, Computational Mechanics.

[22]  Jan S. Hesthaven,et al.  Reduced order modeling for nonlinear structural analysis using Gaussian process regression , 2018, Computer Methods in Applied Mechanics and Engineering.

[23]  David A Steinman,et al.  High-resolution CFD detects high-frequency velocity fluctuations in bifurcation, but not sidewall, aneurysms. , 2013, Journal of biomechanics.

[24]  M. Winter,et al.  Reduced-Order Modeling of Unsteady Aerodynamic Loads using Radial Basis Function Neural Networks , 2014 .

[25]  Charbel Farhat,et al.  Stabilization of projection‐based reduced‐order models , 2012 .

[26]  C. Pain,et al.  Non‐intrusive reduced‐order modelling of the Navier–Stokes equations based on RBF interpolation , 2015 .

[27]  Weiwei Zhang,et al.  Reduced-Order Modeling for Nonlinear Aeroelasticity with Varying Mach Numbers , 2018, Journal of Aerospace Engineering.

[28]  Omer San,et al.  Machine learning closures for model order reduction of thermal fluids , 2018, Applied Mathematical Modelling.

[29]  C. Pain,et al.  Model identification of reduced order fluid dynamics systems using deep learning , 2017, International Journal for Numerical Methods in Fluids.

[30]  Jan S. Hesthaven,et al.  Non-intrusive reduced order modeling of nonlinear problems using neural networks , 2018, J. Comput. Phys..

[31]  Rui Huang,et al.  Parametric reduced-order modeling of unsteady aerodynamics for hypersonic vehicles , 2019, Aerospace Science and Technology.

[32]  Erik Cambria,et al.  Recent Trends in Deep Learning Based Natural Language Processing , 2017, IEEE Comput. Intell. Mag..

[33]  Michael C. Romanowski.  Reduced order unsteady aerodynamic and aeroelastic models using Karhunen-Loeve eigenmodes , 1996 .

[34]  Qian Wang,et al.  Non-intrusive reduced order modeling of unsteady flows using artificial neural networks with application to a combustion problem , 2019, J. Comput. Phys..

[35]  David J. Lucia,et al.  Projection methods for reduced order models of compressible flows , 2003 .