Predicting Next Steps of a CFD Simulation using Deep Learning
Resumo
As simulações de dinâmica dos fluidos computacional (CFD) possuem aplicações em diversas áreas como indústria aeronáutica, automobilística e energética. Os fluxos de fluido são descritos por equações cuja resolução ainda é limitada pelo custo computacional. O objetivo desse trabalho é complementar as simulações de CFD em termos de previsão dos instantes subsequentes da simulação utilizando aprendizado profundo. A metodologia proposta modela esse problema como a previsão do próximo quadro de um vídeo. A entrada para a rede profunda são as primeiras imagens da simulação CFD e a saída são as próximas imagens. Os resultados obtidos para a simulação de um fluxo através de um cilindro fixo apresentaram valores de SSIM maiores que 0.83.
Referências
Avila, M., Gargallo-Peiró, A., and Folch, A. (2017). A cfd framework for offshore and onshore wind farm simulation. In Journal of Physics: Conference Series, volume 854, page 012002. IOP Publishing.
Bhatnagar, S., Afshar, Y., Pan, S., Duraisamy, K., and Kaushik, S. (2019). Prediction of aerodynamic flow fields using convolutional neural networks. Computational Mechanics, 64:525–545.
Kochkov, D., Smith, J. A., Alieva, A., Wang, Q., Brenner, M. P., and Hoyer, S. (2021). Machine learning–accelerated computational fluid dynamics. Proceedings of the National Academy of Sciences, 118(21):e2101784118.
Mathias, M. S., de Almeida, W. P., Coelho, J. F., de Freitas, L. P., Moreno, F. M., Netto, C. F. D., Cozman, F. G., Reali Costa, A. H., Tannuri, E. A., Gomi, E. S., and Dottori, M. (2022a). Augmenting a physics-informed neural network for the 2d burgers equation by addition of solution data points. In Xavier-Junior, J. C. and Rios, R. A., editors, Intelligent Systems, pages 388–401. Springer International Publishing.
Mathias, M. S., de Barros, M. R., Coelho, J. F., de Freitas, L. P., Moreno, F. M., Netto, C. F. D., Cozman, F. G., Costa, A. H. R., Tannuri, E. A., Gomi, E. S., and Dottori, M. (2022b). A physics-informed neural network to model port channels.
Ozaki, H. and Aoyagi, T. (2022). Prediction of steady flows passing fixed cylinders using deep learning. Scientific Reports 2022 12:1, 12:1–12.
Shi, X., Chen, Z., Wang, H., Yeung, D.-Y., Wong, W.-K., and Woo, W.-c. (2015). Convolutional lstm network: A machine learning approach for precipitation nowcasting. Advances in neural information processing systems, 28.
Sirignano, J., MacArt, J. F., and Freund, J. B. (2019). Dpm: A deep learning pde augmentation method (with application to large-eddy simulation). Journal of Computational Physics, 423.
Usman, A., Rafiq, M., Saeed, M., Nauman, A., Almqvist, A., and Liwicki, M. (2021). Machine learning computational fluid dynamics. In 2021 Swedish Artificial Intelligence Society Workshop (SAIS), pages 1–4.
Vinuesa, R. and Brunton, S. L. (2022a). Enhancing computational fluid dynamics with machine learning. Nature Computational Science 2022 2:6, 2:358–366.
Vinuesa, R. and Brunton, S. L. (2022b). Enhancing computational fluid dynamics with machine learning. Nature Computational Science, 2(6):358–366.
Wang, Z., Bovik, A., Sheikh, H., and Simoncelli, E. (2004). Image quality assessment: from error visibility to structural similarity. IEEE Transactions on Image Processing, 13(4):600–612.
Ye, S., Zhang, Z., Song, X., Wang, Y., Chen, Y., and Huang, C. (2020). A flow feature detection method for modeling pressure distribution around a cylinder in non-uniform flows by using a convolutional neural network. Scientific Reports, 10:4459–4459.
Yeung, P. K., Zhai, X. M., and Sreenivasan, K. R. (2015). Extreme events in computational turbulence. Proceedings of the National Academy of Sciences of the United States of America, 112:12633–12638.
Zhou, Y., Dong, H., and El Saddik, A. (2020). Deep learning in next-frame prediction: A benchmark review. IEEE Access, 8:69273–69283.