Wind and Structures

Techno-Press (테크노프레스)
권호 표지
DOI QR코드

DOI QR Code

A deep learning framework for wind pressure super-resolution reconstruction

  • Xiao Chen (Artificial Intelligence for Wind Engineering (AIWE) Lab, School of Civil and Environmental Engineering, Harbin Institute of Technology) ;
  • Xinhui Dong (Artificial Intelligence for Wind Engineering (AIWE) Lab, School of Civil and Environmental Engineering, Harbin Institute of Technology) ;
  • Pengfei Lin (Artificial Intelligence for Wind Engineering (AIWE) Lab, School of Civil and Environmental Engineering, Harbin Institute of Technology) ;
  • Fei Ding (NatHaz Modeling Laboratory, University of Notre Dame) ;
  • Bubryur Kim (Department of Robot and Smart System Engineering, Kyungpook National University) ;
  • Jie Song (Research Center of Urban Disasters Prevention and Fire Rescue Technology of Hubei Province, School of Civil Engineering, Wuhan University) ;
  • Yiqing Xiao (Artificial Intelligence for Wind Engineering (AIWE) Lab, School of Civil and Environmental Engineering, Harbin Institute of Technology) ;
  • Gang Hu (Artificial Intelligence for Wind Engineering (AIWE) Lab, School of Civil and Environmental Engineering, Harbin Institute of Technology)
  • Received : 2022.08.29
  • Accepted : 2023.01.12
  • Published : 2023.06.25
⟨ Previous Next ⟩

Abstract

Strong wind is the main factors of wind-damage of high-rise buildings, which often creates largely economical losses and casualties. Wind pressure plays a critical role in wind effects on buildings. To obtain the high-resolution wind pressure field, it often requires massive pressure taps. In this study, two traditional methods, including bilinear and bicubic interpolation, and two deep learning techniques including Residual Networks (ResNet) and Generative Adversarial Networks (GANs), are employed to reconstruct wind pressure filed from limited pressure taps on the surface of an ideal building from TPU database. It was found that the GANs model exhibits the best performance in reconstructing the wind pressure field. Meanwhile, it was confirmed that k-means clustering based retained pressure taps as model input can significantly improve the reconstruction ability of GANs model. Finally, the generalization ability of k-means clustering based GANs model in reconstructing wind pressure field is verified by an actual engineering structure. Importantly, the k-means clustering based GANs model can achieve satisfactory reconstruction in wind pressure field under the inputs processing by k-means clustering, even the 20% of pressure taps. Therefore, it is expected to save a huge number of pressure taps under the field reconstruction and achieve timely and accurately reconstruction of wind pressure field under k-means clustering based GANs model.

Keywords

Acknowledgement

This study is supported by National Key R&D Program of China (2021YFC3100702), National Natural Science Foundation of China (52108451), Shenzhen Science and Technology Program (SGDX20210823103202018), Shenzhen Science and Technology Innovation Commission (GXWD20201230155427003-20200823230021001), Shenzhen Science and Technology Program (KQTD20210811090112003), and Guangdong-Hong KongMacao Joint Laboratory for Data-Driven Fluid Mechanics and Engineering Applications (2020B1212030001). The authors highly appreciate the aerodynamic database of Tokyo Polytechnic University.

References

  1. Bao, Y. and Li, H. (2020), "Machine learning paradigm for structural health monitoring", J. Civil Struct. Health Monit., 20, 1353-1372. https://doi.org/10.1177/1475921720972416. 
  2. Bengio, Y., Simard, P. and Frasconi, P. (1994), "Learning long-term dependencies with gradient descent is difficult", IEEE Transact. Neural Netw., 5, 157-166. https://doi.org/10.1109/72.279181. 
  3. Bruna, J., Sprechmann, P. and LeCun, Y. (2016), "Super-resolution with deep convolutional sufficient statistics", The 4th International Conference on Learning Representations, San Juan, May. https://doi.org/10.48550/arXiv.1511.05666. 
  4. Cao, S., Wang, J., Cao, J., Zhao, L. and Chen, X. (2015), "Experimental study of wind pressures acting on a cooling tower exposed to stationary tornado-like vortices", J. Wind Eng. Indust. Aerodyn., 145, 75-86. https://doi.org/10.1016/j.jweia.2015.06.004. 
  5. Chen, Y., Kopp, G.A. and Surry, D. (2003), "Prediction of pressure coefficients on roofs of low buildings using artificial neural networks", J. Wind Eng. Indust. Aerodyn., 91, 423-441. https://doi.org/10.1016/S0167-6105(02)00381-1. 
  6. Cheng, L., Lam, K. and Wong, S. (2015), "Pod analysis of crosswind forces on a tall building with square and h-shaped cross-sections", Wind Struct. Int. J., 21, 63-84. http://dx.doi.org/10.12989/was.2015.21.1.063. 
  7. Deng, Z., He, C., Liu, Y. and Kim, K.C. (2019), "Super-resolution reconstruction of turbulent velocity fields using a generative adversarial network-based artificial intelligence framework", Phys. Fluid., 31, 125111. https://doi.org/10.1063/1.5127031. 
  8. Dong, C., Loy, C.C., He, K. and Tang, X. (2016a), "Image super-resolution using deep convolutional networks", IEEE Transact. Pattern Anal. Mach. Intell., 38, 295-307. https://doi.org/10.1109/TPAMI.2015.2439281. 
  9. Dong, C., Loy, C.C. and Tang, X. (2016b), "Accelerating the super-resolution convolutional neural network", Lecture Notes Comput. Sci., 9906, 391-407. https://doi.org/10.1007/978-3-319-46475-6_25. 
  10. Diez, A., Khoa, N.L.D., Alamdari, M.M., Wang, Y., Chen, F. and Runcie, P. (2016), "A clustering approach for structural health monitoring on bridges", J. Wind Eng. Indust. Aerodyn., 6, 429-445. https://doi.org/10.1007/s13349-016-0160-0. 
  11. Duthinh, D., Main, J.A., Gierson, M.L. and Phillips, B.M. (2018), "Analysis of wind pressure data on components and cladding of low-rise buildings", ASCE-ASME J. Risk Uncertain. Eng. Syst., Part A: Civil Eng., 4(1), 04017032. https://doi.org/10.1061/AJRUA6.0000936. 
  12. Feng, R., Gu, J., Qiao, Y. and Dong, C. (2019), "Suppressing model overfitting for image super-resolution networks", IEEE Comput. Soc. Conf. Comput. Vision Pattern Recog. Workshops, 1964-1973. 
  13. Feng, R. qiang, Zhu, B. and Wang, X. (2015), "A mode contribution ratio method for seismic analysis of large-span spatial structures", Int. J. Steel Struct., 15, 835-852. https://doi.org/10.1007/s13296-015-1206-6. 
  14. Fu, J.Y., Liang, S.G. and Li, Q.S. (2007), "Prediction of wind-induced pressures on a large gymnasium roof using artificial neural networks", Comput. Struct., 85, 179-192. https://doi.org/10.1016/j.compstruc.2006.08.070. 
  15. Fukami, K., Nabae, Y., Kawai, K. and Fukagata, K. (2019), "Synthetic turbulent inflow generator using machine learning", Phys. Rev. Fluids, 4, 064603. https://doi.org/10.1103/PhysRevFluids.4.064603. 
  16. Glorot, X. and Bengio, Y. (2010), "Understanding the difficulty of training deep feedforward neural networks", J. Mach. Learn. Res., 249-256. 
  17. Goodfellow, I., Pouget-Abadie, J., Mirza, M., Xu, B., WardeFarley, D., Ozair, S., Courville, A. and Bengio, Y. (2014), "Generative adversarial networks", Communication. ACM, 63(11), 139-144. https://doi.org/10.1145/3422622. 
  18. Han, D. (2013), "Comparison of commonly used image interpolation methods", The 2nd International Conference on Computer Science and Electronics Engineering (ICCSEE 2013), Hangzhou, March. 
  19. Harris, J.L. (1964), "Diffraction and resolving power", J. Optic. Soc. America, 54(7), 931-936. https://doi.org/10.1364/JOSA.54.000931. 
  20. He, K., Zhang, X., Ren, S. and Sun, J. (2016), "Deep residual learning for image recognition", IEEE Comput. Soc. Conf. Comput. Vision Pattern Recog., 770-778. 
  21. Hu, G. and Kwok, K.C.S. (2020), "Predicting wind pressures around circular cylinders using machine learning techniques", J. Wind Eng. Indust. Aerodyn., 198, 104099. https://doi.org/10.1016/j.jweia.2020.104099. 
  22. Hu, G., Liu, L., Tao, D., Song, J., Tse, K.T. and Kwok, K.C.S. (2020), "Deep learning-based investigation of wind pressures on tall building under interference effects", J. Wind Eng. Indust. Aerodyn., 201, 104138. https://doi.org/10.1016/j.jweia.2020.104138. 
  23. Huang, D., He, S., He, X. and Zhu, X. (2017), "Prediction of wind loads on high-rise building using a BP neural network combined with POD", J. Wind Eng. Indust. Aerodyn., 170, 1-17. https://doi.org/10.1016/j.jweia.2017.07.021. 
  24. Isola, P., Zhu, J.Y., Zhou, T. and Efros, A.A. (2017), "Image-to-image translation with conditional adversarial networks", The 30th IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, July. 
  25. Jin, X., Laima, S., Chen, W.L. and Li, H. (2020), "Time-resolved reconstruction of flow field around a circular cylinder by recurrent neural networks based on non-time-resolved particle image velocimetry measurements", Exp. Fluids, 61, 114. https://doi.org/10.1007/s00348-020-2928-6. 
  26. Karras, T., Laine, S. and Aila, T. (2018), "A style-based generator architecture for generative adversarial networks", IEEE Conference on Computer Vision and Pattern Recognition, Long Beach, CA, June. 
  27. Ke, S.T., Ge, Y.J., Zhao, L. and Tamura, Y. (2012)m "A new methodology for analysis of equivalent static wind loads on super-large cooling towers", J. Wind Eng. Indust. Aerodyn., 111, 30-39. https://doi.org/10.1016/j.jweia.2012.08.001. 
  28. Kim, B., Yuvaraj, N., Sri Preethaa, K.R., Hu, G. and Lee, D.E. (2021a), "Wind-induced pressure prediction on tall buildings using generative adversarial imputation network", Sensors, 21(7), 2515. https://doi.org/10.3390/s21072515. 
  29. Kim, B., Yuvaraj, N., Tse, K.T., Lee, D. and Hu, G. (2021b), "Pressure pattern recognition in buildings using an unsupervised machine-learning algorithm", J. Wind Eng. Indust. Aerodyn., 214, 104629. https://doi.org/10.1016/j.jweia.2021.104629. 
  30. Kim, H., Kim, J., Won, S. and Lee, C. (2021), "Unsupervised deep learning for super-resolution reconstruction of turbulence", J. Fluid Mech., 910, A29. https://doi.org/10.1017/jfm.2020.1028. 
  31. Kim, J. and Lee, C. (2020), "Deep unsupervised learning of turbulence for inflow generation at various Reynolds numbers", J. Comput. Phys., 406, 109216. https://doi.org/10.1016/j.jcp.2019.109216. 
  32. Kim, S.P., Bose, N.K. and Valenzuela, H.M. (1990), "Recursive reconstruction of high resolution image from noisy undersampled multiframes", IEEE Transact. Acous. Speech Sign. Processing, 38(6), 1013-1027. https://doi.org/10.1109/29.56062. 
  33. Kim, S. Y., Oh, J. and Kim, M. (2019), "Deep SR-ITM: Joint learning of super-resolution and inverse tone-mapping for 4K UHD HRR applications", IEEE/CVF International Conference on Computer Vision, Seoul, October-November.
  34. Kingma, D.P. and Ba, J.L. (2015), "Adam: A method for stochastic optimization", The 3rd International Conference on Learning Representations, San Diego, CA, May. https://doi.org/10.48550/arXiv.1412.6980. 
  35. Koza, J.R., Bennett, F.H., Andre, D. and Keane, M.A. (1996), "Automated design of both the topology and sizing of analog electrical circuits using genetic programming", Artif. Intell. Des., 96, 151-170. https://doi.org/10.1007/978-94-009-0279-4_9. 
  36. Lai, W.S., Huang, J. Bin, Ahuja, N. and Yang, M.H. (2017), "Deep laplacian pyramid networks for fast and accurate super-resolution, The 30th IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, July. 
  37. Ledig, C., Theis, L., Huszar, F., Caballero, J., Cunningham, A., Acosta, A., Aitken, A., Tejani, A., Totz, J., Wang, Z. and Shi, W. (2017), "Photo-realistic single image super-resolution using a generative adversarial network", The 30th IEEE Conference on Computer Vision and Pattern Recognition, Honolulu, HI, July. 
  38. Lee, Y.J., Hall, D., Liu, Q., Liao, W.W. and Huang, M.C. (2021), "Interpretable tropical cyclone intensity estimation using Dvorak-inspired machine learning techniques", Eng. Appl. Artif. Intell., 101, 104233. https://doi.org/10.1016/j.engappai.2021.104233. 
  39. Letchford, C.W. and Scott Norville, H. (1994), "Wind pressure loading cycles for wall cladding during hurricanes", J. Wind Eng. Indust. Aerodyn., 53, 189-206. https://doi.org/10.1016/0167-6105(94)90026-4. 
  40. Li, S., Laima, S. and Li, H. (2018), "Data-driven modeling of vortex-induced vibration of a long-span suspension bridge using decision tree learning and support vector regression", J. Wind Eng. Indust. Aerodyn., 172, 196-211. https://doi.org/10.1016/j.jweia.2017.10.022. 
  41. Li, W., Laima, S., Jin, X., Yuan, W. and Li, H. (2020), "A novel long short-term memory neural-network-based self-excited force model of limit cycle oscillations of nonlinear flutter for various aerodynamic configurations", Nonlinear Dyn., 100, 2071-2087. https://doi.org/10.1007/s11071-020-05631-5. 
  42. Li, D., Liu, B. and Cheng, Y. (2020a), "Wind pressure coefficients zoning method based on an unsupervised learning algorithm", Math. Probl. Eng., 2020, 1670128. https://doi.org/10.1155/2020/1670128. 
  43. Liao, H., Mei, H., Hu, G., Wu, B. and Wang, Q. (2021), "Machine learning strategy for predicting flutter performance of streamlined box girders", J. Wind Eng. Indust. Aerodyn., 209, 104493. https://doi.org/10.1016/j.jweia.2020.104493. 
  44. Lin, P., Hu, G., Li, C., Li, L., Xiao, Y., Tse, K.T. and Kwok, K.C.S. (2021), "Machine learning-based prediction of crosswind vibrations of rectangular cylinders", J. Wind Eng. Indust. Aerodyn., 211, 104549. https://doi.org/10.1016/j.jweia.2021.104549. 
  45. Lin, P., Ding, F., Hu, G., Li, C., Xiao, Y., Tse, K.T., Kwok, K.C.S. and Kareem, A. (2022), "Machine learning-enabled estimation of crosswind load effect on tall buildings", J. Wind Eng. Indust. Aerodyn., 220, 104860. https://doi.org/10.1016/j.jweia.2021.104860. 
  46. Liu, B., Tang, J., Huang, H. and Lu, X.Y. (2020), "Deep learning methods for super-resolution reconstruction of turbulent flows", Phys. Fluid., 32, 025105. https://doi.org/10.1063/1.5140772. 
  47. Liu, M., Li, Q.S., Huang, S.H., Shi, F. and Chen, F. (2018), "Evaluation of wind effects on a large span retractable roof stadium by wind tunnel experiment and numerical simulation", J. Wind Eng. Indust. Aerodyn., 179, 39-57. https://doi.org/10.1016/j.jweia.2018.05.014. 
  48. Liu, Z. and Ishihara, T. (2015), "A study of tornado induced mean aerodynamic forces on a gable-roofed building by the large eddy simulations", J. Wind Eng. Indust. Aerodyn., 146, 39-50. https://doi.org/10.1016/j.jweia.2015.08.002. 
  49. Noh, J., Bae, W., Lee, W., Seo, J. and Kim, G. (2019), "Better to follow, follow to be better: Towards precise supervision of feature super-resolution for small object detection", IEEE International Conference on Computer Vision, Seoul, October-November.
  50. Quan, Y., Tamura, Y. and Matsui, M. (2007), "Mean wind pressure coefficients on surfaces of gable-roofed low-rise buildings", Adv. Struct. Eng., 10(3), 259-271. https://doi.org/10.1260/1369433077814222. 
  51. Radford, A., Metz, L. and Chintala, S. (2016), "Unsupervised representation learning with deep convolutional generative adversarial networks", The 4th International Conference on Learning Representations, San Juan, May. https://doi.org/10.48550/arXiv.1511.06434. 
  52. Schraudolph, N.N. (1998), Accelerated Gradient Descent by Factor-Centering Decomposition. 
  53. Seiler, M.C. and Seiler, F.A. (1989), "Numerical recipes in C: the art of scientific computing", Risk Anal., 9, 415-416. https://doi.org/10.1111/j.1539-6924.1989.tb01007.x. 
  54. Shi, W., Caballero, J., Huszar, F., Totz, J., Aitken, A.P., Bishop, R., Rueckert, D. and Wang, Z. (2016), "Real-time single image and video super-resolution using an efficient sub-pixel convolutional neural network", IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Las Vegas, NV, June-July. 
  55. Silva, M., Santos, A., Santos, R., Figueiredo, E., Sales, C. and Costa, J.C. (2017), "Agglomerative concentric hypersphere clustering applied to structural damage detection", Mech. Syst. Signal Processing, 92, 196-212. https://doi.org/10.1016/j.ymssp.2017.01.024. 
  56. Sohn, H., Kim, S.D. and Harries, K. (2008), "Reference-free damage classification based on cluster analysis", Comput.-Aided Civil Infrastruct. Eng., 23(5), 324-338. https://doi.org/10.1111/j.1467-8667.2008.00541.x. 
  57. Srivastava, R.K., Greff, K. and Schmidhuber, J. (2015), "Training very deep networks", The 28th International Conference on Neural Information Processing Systems, Montreal, December. 
  58. Sun, J., Zheng, N.N., Tao, H. and Shum, H.Y. (2003), "Image hallucination with primal sketch priors", IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Madison, WI, June. 
  59. Tian, J., Gurley, K.R., Diaz, M.T., Fernandez-Caban, P.L., Masters, F.J. and Fang, R. (2020), "Low-rise gable roof buildings pressure prediction using deep neural networks", J. Wind Eng. Indust. Aerodyn., 196, 104026. https://doi.org/10.1016/j.jweia.2019.104026. 
  60. Van Ouwerkerk, J.D. (2006), "Image super-resolution survey", Image Vision Comput., 24(10), 1039-1052. https://doi.org/10.1016/j.imavis.2006.02.026. 
  61. Wang, H., Zhang, Y.M., Mao, J.X. and Wan, H.P. (2020), "A probabilistic approach for short-term prediction of wind gust speed using ensemble learning", J. Wind Eng. Indust. Aerodyn., 202, 104198. https://doi.org/10.1016/j.jweia.2020.104198. 
  62. Wang, Q., Yan, L., Hu, G., Li, C., Xiao, Y., Xiong, H., Rabault, J. and Noack, B.R. (2022), "DRLinFluids-an open-source Python platform of coupling deep reinforcement learning and OpenFOAM", Phys. Fluids, 34, 081801. https://doi.org/10.1063/5.0103113. 
  63. Wang, Y., Huang, X., Yan, Y. and Zhen, Y. (2009), "A new method for motion-blurred image blind restoration based on huber Markov random field", The 5th International Conference on Image and Graphics, Beijing, January. 
  64. Yang, J., Wright, J., Huang, T.S. and Ma, Y. (2010), "Image super-resolution via sparse representation", IEEE Transact. Image Processing, 19(11), 2861-2873. https://doi.org/10.1109/TIP.2010.2050625 
  65. Yang, Z., Sarkar, P. and Hu, H. (2010), "Visualization of flow structures around a gable-roofed building model in tornado-like winds", J. Vis., 13, 285-288. https://doi.org/10.1007/s12650-010-0052-z. 
  66. Yoo, J., Ahn, N. and Sohn, K.A. (2020), "Rethinking data augmentation for image super-resolution: A comprehensive analysis and a new strategy", IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Seattle, WA, June. 
  67. Zhang, A., Lipton, Z., Li, M. and Smola, A. (2021), "Dive into deep learning", https://doi.org/10.48550/arXiv.2106.11342. 
  68. Zhang, K., Van Gool, L. and Timofte, R. (2020), "Deep unfolding network for image super-resolution", IEEE Computer Society Conference on Computer Vision and Pattern Recognition, Seattle, WA, June. 
  69. Zhu, J.Y., Park, T., Isola, P. and Efros, A.A. (2017), "Unpaired image-to-image translation using cycle-consistent adversarial networks", IEEE International Conference on Computer Vision, Venice, October. 
  70. Zhou, H.B. and Gao, J.T. (2014), "Automatic method for determining cluster number based on silhouette coefficient", Adv. Mater. Res,, 951, 227-230. https://doi.org/10.4028/www.scientific.net/amr.951.227.