Considering the demand for predicting density fluctuation in turbulence of in prediction of the aero-optical effect, this paper develops a neural network method to predict the density fluctuation in supersonic turbulence. A model for the density fluctuation with 5 hidden layers is built by mining the regularity from the flow field data of supersonic turbulent boundary layers, which are simulated by the Direct Numerical Simulation (DNS) method. The experimental results show that the neural network method proposed is able to predict the mean square values of density fluctuation accurately. The method can predict the training samples well, and can get the prediction of test samples with better accuracy and stability than traditional models, and has certain capability of generalization. By selecting features and adding prior information, the 7 features of input parameters of the density fluctuation model are determined to further improve the capability of generalization and practicability of the model.
WANG Zhengkui
,
JIN Xuhong
,
ZHU Zhibin
,
CHENG Xiaoli
. Neural network method for predicting density fluctuations in supersonic turbulence[J]. ACTA AERONAUTICAET ASTRONAUTICA SINICA, 2018
, 39(10)
: 122244
-122244
.
DOI: 10.7527/S1000-6893.2018.22244
[1] GIERLOFF J, ROBERTSON S, BOUSKA D. Computer analysis of aero-optic effects:AIAA-1992-2794[R]. Reston,VA:AIAA, 1992.
[2] CLARK R, FARRIS R. A numerical method to predict aero-optical performance in hypersonic flight:AIAA-1987-1396[R]. Reston,VA:AIAA, 1987.
[3] LUTZ S A. Modeling of density fluctuations in supersonic turbulent boundary layer[J]. AIAA Journal, 1989, 27(6):822-823.
[4] POND J E, SUTTON G W. Aero-optic performance of an aircraft forward-facing optical turret[J]. Journal of Aircraft, 2015, 43(3):600-607.
[5] RODI W. A new algebraic relation for calculating the Reynolds stresses[C]//Gesellschaft fuer angewandte Mathematik und Mechanik, Wissenschaftliche Jahrestagung, 1976:219-221.
[6] WEI H, CHEN C. A second-order algebraic model for turbulent density fluctuation:AIAA-1996-0427[R]. Reston, VA:AIAA, 1996.
[7] WEATHERITT J, PICHLER R, SANDBERG R D, et al. Machine learning for turbulence model development using a high-fidelity HPT cascade simulation:GT2017-63497[R]. New York:ASME, 2017.
[8] LING J, TEMPLETON J. Evaluation of machine learning algorithms for prediction of regions of high Reynolds averaged Navier Stokes uncertainty[J]. Physics of Fluids, 2015, 27(8):42032-42094.
[9] LING J. Using machine learning to understand and mitigate model form uncertainty in turbulence models[C]//14th International Conference on Machine Learning and Applications. Piscataway, NJ:IEEE Press, 2015:813-818.
[10] WANG J X, WU J L, XIAO H. Physics-informed machine learning for predictive turbulence modeling:Using data to improve RANS modeled Reynolds stresses[J]. Physical Review Fluids, 2016, 2:034603.
[11] EDELING W N, IACCARINO G, CINNELLA P. Data-free and data-driven RANS predictions with quantified uncertainty[J]. Flow, Turbulence & Combustion, 2018, 100(3):593-616.
[12] PARISH E J, DURAISAMY K. A paradigm for data-driven predictive modeling using field inversion and machine learning[J]. Journal of Computational Physics, 2016, 305:758-774.
[13] SINGH A P, DURAISAMY K. Using field inversion to quantify functional errors in turbulence closures[J]. Physics of Fluids, 2016, 28(4):045110.
[14] TRACEY B, DURAISAMY K, ALONSO J J. A machine learning strategy to assist turbulence model development:AIAA-2015-1287[R]. Reston, VA:AIAA, 2015.
[15] YARLANKI S, RAJENDRAN B, HAMANN H. Estimation of turbulence closure coefficients for data centers using machine learning algorithms[C]//Thermal and Thermomechanical Phenomena in Electronic Systems. Piscataway, NJ:IEEE Press, 2012:38-42.
[16] RAY J, LEFANTZI S, ARUNAJATESAN S, et al. Bayesian parameter estimation of a k-ε model for accurate jet-in-crossflow simulations[J]. AIAA Journal, 2016, 54(8):2432-2448.
[17] RAY J, LEFANTZI S, ARUNAJATESAN S, et al. Bayesian calibration of a k-ε turbulence model for predictive jet-in-crossflow simulations:AIAA-2014-2085[R]. Reston, VA:AIAA, 2014.
[18] DURAISAMY K, ZHANG Z J, SINGH A P. New approaches in turbulence and transition modeling using data-driven techniques:AIAA-2015-1284[R]. Reston,VA:AIAA, 2015.
[19] ZHANG Z J, DURAISAMY K. Machine learning methods for data-driven turbulence modeling:AIAA-2015-2460[R]. Reston,VA:AIAA, 2015.
[20] ZHANG X, SHU C. Positivity-preserving high order finite difference WENO schemes for compressible Euler equations[J]. Journal of Computational Physics, 2012, 231(5):2245-2258.
[21] JIANG G, SHU C. Efficient Implementation of weighted ENO schemes[J]. Journal of Computational Physics, 1996, 126(1):202-228.
[22] SCHLATTER P, STOLZ S, KLEISER L. Large-eddy simulation of transition in wall-bounded flow[C]//40th JAXA Workshop on "Investigation and Control of Boundary-Layer Transition". Tokyo:Japan Aerospace Exploration Agency, 2007:13-16.
[23] GATSKI T B, ERLEBACHER G. Numerical simulation of a spatially evolving supersonic turbulent boundary layer:NASA/TM-2002-211934[R]. Washington, D.C.:NASA, 2002.
[24] PIROZZOLI S, GRASSO F, GATSKI T B. Direct numerical simulation and analysis of a spatially evolving supersonic turbulent boundary layer at M=2.25[J]. Physics of Fluids, 2004, 16(3):530-545.
[25] WANG K, WANG M. Aero-optics of subsonic turbulent boundary layers[J]. Journal of Fluid Mechanics, 2012, 696:122-151.
CJA