REPRESENTATION AND RECONSTRUCTION OF FLOW AROUND BRIDGE DECK USING TIME HISTORY DEEP LEARNING
-
摘要: 流场特性的研究是结构风工程的核心问题，而高分辨率的流场数据对解决风致振动问题、探索流固耦合机理具有着重要意义。受测量方法、计算效率等因素限制，高空间分辨率的流场时程数据的直接获取仍有一定困难。该文基于流场时程数据的表征模型，提出了桥面非定常流动时程重构的深度学习方法。基于一维卷积方法建立了非定常桥面绕流场的表征模型，得到了物理空间与表征模型的编码空间之间的映射关系，最后利用表征模型的解码器生成未知测点处的流场时程数据。对较低雷诺数桥梁主梁的非定常绕流流场进行了研究与验证，实现了桥面绕流的时程数据重构，验证了方法的准确性与可行性。该文所提方法基于流场的时程数据进行表征与重构，可广泛应用于工程中基于一点的传感器数据处理，是一种桥面流场数据分析的新方法。Abstract: High-resolution flow field data has a great significance to the study of fluid induced vibration and vortex induced vibration mechanism. Limited by measurement methods and calculation efficiency, it is still difficult to obtain high-resolution flow fields. Thusly, the low-dimensional representation model of flow time history data is adopted, and a deep learning method is proposed for the reconstruction of unsteady flow time history data. A low-dimensional representation model is established for the unsteady flow field based on the one-dimensional convolution method; The mapping relationship is developed between the physical space and the encoding space; The decoder in the representation model is utilized to generate the flow field time history data at any position. The problem of unsteady flow around bridge deck is verified, and the accuracy of the method is proved. The method proposed is a high-precision flow field data reconstruction method in the time dimension, and it is an unsupervised training method. It is a brand-new method that can be widely used in point-based sensor data processing.
-
Key words:
- flow reconstruction /
- flow time history /
- deep learning /
- feature extraction /
- unsupervised model
-
表 1 流动时程自动编码特征分类模型参数
Table 1. FTH-AE model parameters
网络层 通道数 激活函数 参数个数 Input ? ? 0 Conv 1 64 ReLU 2688 Conv 2 32 ReLU 84000 Conv 3 10 ReLU 13130 Code 10 ReLU 185710 Dense 18570 ReLU 204270 Conv_T 1 10 ReLU 4110 Conv_T 2 32 ReLU 13152 Conv_T 3 64 ReLU 84032 Output 1 ReLU 2625 下载: 导出CSV
-
[1] 葛耀君. 大跨度桥梁抗风的技术挑战与精细化研究[J]. kb88凯时集团官网, 2011(增刊 2): 11 ? 23.GE Yaojun. Technical challenges and refinement research on wind resistance of long-span bridges [J]. Engineering Mechanics, 2011(Suppl 2): 11 ? 23. (in Chinese) [2] 李永乐, 陈星宇, 汪斌, 等. 扁平箱梁涡激共振阻塞效应及振幅修正[J]. kb88凯时集团官网, 2018, 35(11): 45 ? 52. doi: 10.6052/j.issn.1000-4750.2017.07.0576LI Yongle, CHEN Xingyu, WANG Bin, et al. Blockage-Effects and amplitude conversion of vortex- induced vibration for flat-box girder [J]. Engineering Mechanics, 2018, 35(11): 45 ? 52. (in Chinese) doi: 10.6052/j.issn.1000-4750.2017.07.0576 [3] 刘剑寒, 马文勇. 旋转圆柱气动力特性风洞试验研究[J]. kb88凯时集团官网, 2021, 38(增刊): 89 ? 92. doi: 10.6052/j.issn.1000-4750.2020.05.S016LIU Jianhan, MA Wenyong. Wind tunnel test on aerodynamic characteristics of a rotating cylinder [J]. Engineering Mechanics, 2021, 38(Suppl): 89 ? 92. (in Chinese) doi: 10.6052/j.issn.1000-4750.2020.05.S016 [4] 刘庆宽, 孙一飞, 张磊杰, 等. 凹痕对斜拉桥斜拉索气动性能影响研究[J]. kb88凯时集团官网, 2019, 36(增刊): 272 ? 277. doi: 10.6052/j.issn.1000-4750.2018.05.S053LIU Qingkuan, SUN Yifei, ZHANG Leijie, et al. Study on the influence of dent on aerodynamic performance of stay cables of cable-stayed bridge [J]. Engineering Mechanics, 2019, 36(Suppl): 272 ? 277. (in Chinese) doi: 10.6052/j.issn.1000-4750.2018.05.S053 [5] 金晓威, 赖马树金, 李惠. 物理增强的流场深度学习建模与模拟方法[J]. 力学学报, 2021, 53(10): 2616 ? 2629. doi: 10.6052/0459-1879-21-373JIN Xiaowei, LAIMA Shujin, LI Hui. Physics-enhanced deep learning methods for modelling and simulating flow fields [J]. Chinese Journal of Theoretical and Applied Mechanics, 2021, 53(10): 2616 ? 2629. (in Chinese) doi: 10.6052/0459-1879-21-373 [6] KUTZ J N. Deep learning in fluid dynamics [J]. Journal of Fluid Mechanics, 2017, 814: 1 ? 4. doi: 10.1017/jfm.2016.803 [7] MURATA T, FUKAMI K, FUKAGATA K. Nonlinear mode decomposition with convolutional neural networks for fluid dynamics [J]. Journal of Fluid Mechanics, 2019, 882: 1 ? 15. doi: 10.1017/jfm.2019.822 [8] FUKAMI K, NAKAMURA T, FUKAGATA K. Convolutional neural network based hierarchical autoencoder for nonlinear mode decomposition of fluid field data [J]. Physics of Fluids, 2020, 32(9): 095110. doi: 10.1063/5.0020721 [9] RAISSI M, KARNIADAKIS G E. Hidden physics models: Machine learning of nonlinear partial differential equations [J]. Journal of Computational Physics, 2018, 357: 125 ? 141. doi: 10.1016/j.jcp.2017.11.039 [10] RAISSI M, WANG Z, TRIANTAFYLLOU M S, et al. Deep learning of vortex-induced vibrations [J]. Journal of Fluid Mechanics, 2019, 861: 119 ? 137. doi: 10.1017/jfm.2018.872 [11] MAULIK R, SAN O. A neural network approach for the blind deconvolution of turbulent flows [J]. Journal of Fluid Mechanics, 2017, 831: 151 ? 181. doi: 10.1017/jfm.2017.637 [12] FUKAMI K, FUKAGATA K, TAIRA K. Super-resolution reconstruction of turbulent flows with machine learning [J]. Journal of Fluid Mechanics, 2019, 870: 106 ? 120. doi: 10.1017/jfm.2019.238 [13] LIU B, TANG J, HUANG H, et al. Deep learning methods for super-resolution reconstruction of turbulent flows [J]. Physics of Fluids, 2020, 32(2): 025105. doi: 10.1063/1.5140772 [14] KIM H, KIM J, WON S, et al. Unsupervised deep learning for super-resolution reconstruction of turbulence [J]. Journal of Fluid Mechanics, 2021, 910: 1 ? 14. doi: 10.1017/jfm.2020.1028 [15] 叶舒然, 张珍, 宋旭东, 等. 自动编码器在流场降阶中的应用[J]. 空气动力学学报, 2019, 37(3): 498 ? 504.YE Shuran, ZHANG Zhen, SONG Xudong, et al. Applications of autoencoder in reducedGorder modeling of flow field [J]. Acta Aerodynamica Sinica, 2019, 37(3): 498 ? 504. (in Chinese) [16] 惠心雨, 袁泽龙, 白俊强, 等. 基于深度学习的非定常周期性流动预测方法[J]. 空气动力学学报, 2019, 37(3): 462 ? 469.HUI Xinyu, YUAN Zelong, BAI Junqiang, et al. A method of unsteady periodic flow field prediction based on the deep learning [J]. Acta Aerodynamica Sinica, 2019, 37(3): 462 ? 469. (in Chinese) [17] 战庆亮, 葛耀君, 白春锦. 基于尾流时程目标识别的流场参数选择研究[J]. 力学学报, 2021, 10(53): 2692 ? 2702.ZHAN Qingliang, GE Yaojun, BAI Chunjin. Study on flow field parameters of wake time history target recognition [J]. Chinese Journal of Theoretical and Applied Mechanics, 2021, 10(53): 2692 ? 2702. (in Chinese) [18] 战庆亮, 白春锦, 张宁, 等. 基于时程卷积自编码的机翼绕流特征识别方法[J]. 航空学报, 2022, 43(11): 126531. doi: 10.7527/S1000-6893.2021.26531ZHAN Qingliang, BAI Chunjin, ZHANG Ning, et al. Feature extraction method of flow around wing based on time history convolutional autoencoder [J]. Acta Aeronauticaet Astronautica Sinica, 2022, 43(11): 126531. (in Chinese) doi: 10.7527/S1000-6893.2021.26531 [19] 战庆亮, 周志勇, 葛耀君. Re=3900圆柱绕流的三维大涡模拟[J]. 哈尔滨工业大学学报, 2015, 47(12): 75 ? 79.ZHAN Qingliang, ZHOU Zhiyong, GE Yaojun. 3-Dimensional large eddy simulation of circular cylinder at Re=3900 [J]. Journal of Harbin Institute of Technology, 2015, 47(12): 75 ? 79. (in Chinese) [20] 战庆亮, 葛耀君, 白春锦. 流场特征识别的无量纲时程深度学习方法[J]. kb88凯时集团官网, 2023, 40(2): 17 ? 24. doi: 10.6052/j.issn.1000-4750.2021.08.0638Zhan Qingliang, Ge Yaojun, Bai Chunjin. Deep learning method for flow field feature recognition based on dimensionless time history [J]. Engineering Mechanics, 2023, 40(2): 17 ? 24. (in Chinese) doi: 10.6052/j.issn.1000-4750.2021.08.0638 [21] 战庆亮, 白春锦, 葛耀君. 基于时程深度学习的流场特征分析方法[J]. 力学学报, 2022, 54(3): 822 ? 828.ZHAN Qingliang, BAI Chunjin, GE Yaojun. Fluid feature analysis based on time history deep learning [J]. Chinese Journal of Theoretical and Applied Mechanics, 2022, 54(3): 822 ? 828. (in Chinese) -
点击查看大图
图(11) / 表(1)
计量
- 文章访问数: 324
- HTML全文浏览量: 58
- PDF下载量: 99
- 被引次数: 0
