Browse > Article
http://dx.doi.org/10.5407/jksv.2022.20.3.109

Prediction of aerodynamics using VGG16 and U-Net  

Bo Ra, Kim (Division of Mechanical Engineering, Korea Maritime and Ocean University)
Seung Hun, Lee (Division of Mechanical Engineering, Korea Maritime and Ocean University)
Seung Hyun, Jang (Division of Mechanical Engineering, Korea Maritime and Ocean University)
Gwang Il, Hwang (Division of Mechanical Engineering, Korea Maritime and Ocean University)
Min, Yoon (Division of Mechanical Engineering, Korea Maritime and Ocean University)
Publication Information
Journal of the Korean Society of Visualization / v.20, no.3, 2022 , pp. 109-116 More about this Journal
Abstract
The optimized design of airfoils is essential to increase the performance and efficiency of wind turbines. The aerodynamic characteristics of airfoils near the stall show large deviation from experiments and numerical simulations. Hence, it is needed to perform repetitive analysis of various shapes near the stall. To overcome this, the artificial intelligence is used and combined with numerical simulations. In this study, three types of airfoils are chosen, which are S809, S822 and SD7062 used in wind turbines. A convolutional neural network model is proposed in the combination of VGG16 and U-Net. Learning data are constructed by extracting pressure fields and aerodynamic characteristics through numerical analysis of 2D shape. Based on these data, the pressure field and lift coefficient of untrained airfoils are predicted. As a result, even in untrained airfoils, the pressure field is accurately predicted with an error of within 0.04%.
Keywords
Airfoil; Aerodynamic characteristics; Computational fluid dynamics; Deep learning; Convolutional neural network;
Citations & Related Records
Times Cited By KSCI : 2  (Citation Analysis)
연도 인용수 순위
1 Hore, A., and Ziou, D., 2010, "Image quality metrics: PSNR vs. SSIM." 2010 20th international conference on pattern recognition. IEEE.  
2 Seong-Uk, K., Seung-Hui, O., Jin-A, Y., 2012, "Comparison of Aerodynamic Characteristics of a Thick Airfoil for Wind Turbines using XFOIL and EDISON CFD." Korea Institute of Science and Technology Information, pp. 65-68.
3 Morgado, J., Vizinho, R., Silvestre, M. A. R., & Pascoa, J. C., 2016, "XFOIL vs CFD performance predictions for high lift low Reynolds number airfoils." Aerospace Science and Technology, 52, pp. 207-214.   DOI
4 Miyanawala, T. P., & Jaiman, R. K., 2017, "An efficient deep learning technique for the Navier-Stokes equations: Application to unsteady wake flow dynamics." arXiv preprint arXiv: 1710.09099.
5 Drela, M., 1989, "XFOIL: An analysis and design system for low Reynolds number airfoils." In Low Reynolds Number Aerodynamics, Springer, Berlin, Heidelberg, pp.1-12.
6 Xflr5., 2019, http://www.xflr5.tech/xflr5.html.
7 Gunel, O., Koc, E., & Yavuz, T., 2016, November, "CFD vs. XFOIL of airfoil analysis at low reynolds numbers." In 2016 IEEE International Conference on Renewable Energy Research and Applications, pp. 628-632.
8 Guo, X., Li, W., & Iorio, F., 2016, August, "Convolutional neural networks for steady flow approximation." In Proceedings of the 22nd ACM SIGKDD international conference on knowledge discovery and data mining, pp. 481-490.
9 Yilmaz, E., and German, B., 2017, "A convolutional neural network approach to training predictors for airfoil performance." 18th AIAA/ISSMO multidisciplinary analysis and optimization conference.
10 Szegedy, C., Ioffe, S., Vanhoucke, V., & Alemi, A. A., 2017, "Inception-v4, inception-resnet and the impact of residual connections on learning." In Thirty-first AAAI conference on artificial intelligence.
11 Srikanth Tammina. , 2019, "Transfer learning using vgg-16 with deep convolutional neural network for classifying images." International Journal of Scientific and Research Publications, 9(10), pp.143~150.
12 Walters, D. K., and Leylek, J.H., 2004, "A new model for boundary layer transition using a single-point RANS approach.", J. Turbomach. 126.1.
13 Walters, D. K., and Cokljat, D., 2008, "A three-equation eddy-viscosity model for Reynoldsaveraged Navier-Stokes simulations of transitional flow.", Journal of fluids engineering 130.12.
14 Ronneberger, Olaf, Philipp Fischer, and Thomas Brox., 2015, "U-net: Convolutional networks for biomedical image segmentation." International Conference on Medical image computing and computer-assisted intervention. Springer, Cham, pp.234~241.
15 Azmi, A. R. S., Sapit, A., Mohammed1,A. N., Razali, M. A., Sadikin, A., and Nordin, N., 2007, "Study on airflow characteristics of rear wing of F1 car.", IOP Conference Series: Materials Science and Engineering. Vol. 243.
16 Bora, Kim., 2022, " Prediction of the pressure field over airfoils using convolutional neural network.", Korea Maritime Ocean University., Master's thesis.
17 Walters, D. K., & Cokljat, D., 2008, "A three-equation eddy-viscosity model for Reynolds-averaged Navier-Stokes simulations of transitional flow." Journal of Fluids Engineering, 130(12), p.121401.   DOI
18 Simonyan, K., & Zisserman, A., 2014, "Very deep convolutional networks for large-scale image recognition." arXiv preprint arXiv: 1409.1556.
19 Azmi, A. R. S., Sapit, A., Mohammed, A. N., Razali, M. A., Sadikin, A., & Nordin, N., 2017, September, "Study on airflow characteristics of rear wing of F1 car." In IOP Conference Series: Materials Science and Engineering, 243(1), 012030.   DOI
20 Selig, M. S., & McGranahan, B. D., 2004, "Wind tunnel aerodynamic tests of six airfoils for use on small wind turbines." Journal of Solar Energy Engineering., 126(4), pp. 986-1001.   DOI
21 Selig, M. S., and McGranahan, B, D., 2004, "Wind tunnel aerodynamic tests of six airfoils for use on small wind turbines." , J. Sol. Energy Eng. 126.4.
22 Chicco, D., Warrens, M. J., & Jurman, G., 2021, "The coefficient of determination R-squared is more informative than SMAPE, MAE, MAPE, MSE and RMSE in regression analysis evaluation." PeerJ Computer Science, 7, p. 623.