PREDICTION OF CROSSFLOW TRANSITION BASED ON DEEP NEURAL NETWORKS
-
摘要: 基于深度神经网络DNN构建了从层流流场无量纲速度梯度、流向涡强度等物理量到横流转捩模态下间歇因子间的映射关系, 获得一种新的数据驱动转捩模型. 通过将数据驱动转捩模型与SST k-ω湍流模型耦合, 有效简化了转捩模型输运方程求解, 实现高效、准确的亚音速三维边界层横流转捩流场计算. DNN训练数据来自变雷诺数的NLF(2)-0415无限展长后掠翼计算结果, 并以两种工况进行测试, 数据驱动转捩模型预测精度与γ-Reθ转捩模型近似. 将数据驱动转捩模型用于其他典型横流转捩算例的计算, 以验证其泛化能力. 对于变后掠角的NLF(2)-0415后掠翼, 数据驱动转捩模型与γ-Reθt-CF模型预测的转捩位置几乎一致, 并且能够预测出后掠角从45°增长到65°的过程中, 转捩位置先向前再向后移动的现象; 对于标准椭球体, 使用低分辨率网格进行计算, 数据驱动转捩模型依然能够实现转捩位置预测, 对椭球体表面Cf的计算结果与多个平台的横流转捩模型、实验结果基本一致. 研究表明, 以横流转捩相关物理量作为输入对DNN进行训练, 并将获得的数据驱动转捩模型与SST k-ω湍流模型耦合, 可以实现对横流转捩的有效预测, 且具有较强的泛化能力. 数据驱动转捩模型对网格分辨率要求更低, 在保证计算精度的前提下, 具有较高的计算效率.Abstract: This study based on deep neural networks (DNN), produces a mapping from the physical quantities such as dimensionless velocity gradient and streamwise vorticity of laminar flow field to the intermittency of cross flow transition, and obtains a new data driven transition model. The data driven transition model is coupled with the SST k-ω turbulence model, and the process of solving the transport equations is effectively simplified, which realize efficient and accurate numerical simulation of subsonic 3-D cross flow transition. The computational data of NLF(2)-0415 swept airfoil at different Reynolds numbers is used to train DNN, and two cases are used to test. The prediction accuracy of data driven transition model is similar to that of γ-Reθ transition model. Using the data driven transition model to compute other typical examples of cross flow transition, to verify its generalization ability. For the transition locations of NLF(2)-0415 swept airfoil with different swept angles, the simulation results of data driven transition model have similar accuracy to that of γ-Reθ transition model. Moreover, the phenomenon of transition position moving forward and then backward in the process of sweep Angle increasing from 45° to 65° can be predicted by data driven transition model. For the standard ellipsoid, although using low resolution mesh, the data driven transition model has the ability to compute the transition location, and the computed results of Cf are same to those of other transition models and experiments. The results show that coupling data driven transition model (which is obtained from the physical quantities related to cross flow transition) with SST k-ω transition model can realize the general prediction of cross flow transition. On the premise of ensuring computational accuracy, the data driven transition model requires lower resolution mesh and has higher computing efficiency.
-
Key words:
- crossflow transition /
- deep neural works /
- supervised learning /
- intermittency
-
表 1 NFL(2)-0415后掠翼计算条件
Table 1. Conditions of NFL(2)-0415 airfoil
Case Re/106 m−1 h/μm AoA
/(o)Λ
/(o)Tu∞ νt/ν 1 1.92 3.3 −4 45 0.1 5 2 2.19 3 2.37 4 2.73 5 3.27 表 2 交叉验证结果
Table 2. Result of cross validation
DNN Ltrain/10-4 Ltest/10-4 Ltest average/10-4 1 2.68 4.83 4.31 2 1.99 3.68 3 2.54 4.41 4 2.52 4.87 5 2.37 3.74 表 3 数据驱动转捩模型与γ-Reθt-CF转捩模型对转捩位置的计算结果对比
Table 3. Comparison of xtr between data driven transition model and γ-Reθt-CF transition model
Λ Data driven
transition modelγ-Reθt-CF
transition model45o 0.416 0.424 55 o 0.391 0.395 65 o 0.408 0.413 表 4 椭球体计算工况
Table 4. Conditions of ellipsoid
Case Re/m Ma h/μm AoA νt/ν 1 6.5×106 0.136 3.3 15o 5 -
[1] 郝子辉, 阎超, 周玲. k-ω-γ模式对转捩影响因素的预测性能研究. 力学学报, 2015, 47(2): 215-222 (Hao Zihui, Yan Chao, Zhou Ling. Parametric study of a k-ω-γ model in predicting hypersonic boundary-layer flow transition. Chinese Journal of Theoretical and Applied Mechanics, 2015, 47(2): 215-222 (in Chinese) doi: 10.6052/0459-1879-14-294 [2] Zhu WK. Notes on the hypersonic boundary layer transition. Advances in Aerodynamics, 2022, 4(1): 23-24 doi: 10.1186/s42774-022-00117-x [3] 向星皓, 张毅锋, 陈坚强等. 横流转捩模型研究进展. 空气动力学学报, 2018, 36(2): 254-264 (Xiang Xinghao, Zhang Yifeng, Chen Jianqiang et al. Progress in transition models for cross-flow instabilities. Acta Aerodynamica Sinica, 2018, 36(2): 254-264 (in Chinese) [4] 陈久芬, 徐洋, 蒋万秋等. 升力体外形高超声速边界层转捩红外测量实验. 实验流体力学, 2022, 36: 1-9 (Chen Jiufen, Xu Yang, Jiang Wanqiu et al. Infrared thermogram measurement experiment of hypersonic boundary-layer transition of a lifting body. Journal of Experiments in Fluid Mechanics, 2022, 36: 1-9 (in Chinese) doi: 10.11729/syltlx20220005 [5] 刘是成, 姜应磊, 董昊. 高超声速圆锥边界层不稳定性及转捩实验研究. 实验流体力学, 2022, 36(2): 122-130 (Liu Shicheng, Jiang Yinglei, Dong Hao. Experimental study on instability and transition over hypersonic boundary layer on a straight cone. Journal of Experiments in Fluid Mechanics, 2022, 36(2): 122-130 (in Chinese) [6] 易仕和, 刘小林, 牛海波等. 高超声速边界层流动稳定性实验研究. 空气动力学学报, 2020, 38(1): 137-142 (Yi Shihe, Liu Xiaolin, Niu Haibo et al. Experimental study on flow stability of hypersonic boundary layer. Acta Aerodynamica Sinica, 2020, 38(1): 137-142 (in Chinese) [7] Dong SW, Chen JQ, Yuan XX. Wall pressure beneath a transitional hypersonic boundary layer over an inclined straight circular cone. Advances in Aerodynamics, 2020, 2(1): 29-48 doi: 10.1186/s42774-020-00057-4 [8] Sun D, Chen JQ, Yuan XX, et al. On the wake structure of amicro-ramp vortex generator in hypersonic flow. Physics of fluids, 2020, 32(12): 126111 doi: 10.1063/5.0030975 [9] Han Q, Li XL, Yu CP, et al. Direct numerical simulation of hypersonic boundary layer transition over a liftingbody model HyTRV. Advances in Aerodynamics, 2021, 3(1): 1-21 doi: 10.1186/s42774-020-00055-6 [10] 陈坚强, 涂国华, 张毅锋等. 高超声速边界层转捩研究现状与发展趋势. 空气动力学学报, 2017, 35(3): 311-337Chen Jianqiang, Tu Guohua, Zhang Yifeng et al. Hypersonic boundart layer transition: what we know, where shall we go. Acta Aerodynamica Sinica, 2017, 35(03): 311-337 (in Chinese) [11] 谢晨月, 袁泽龙, 王建春等. 基于人工神经网络的湍流大涡模拟方法. 力学学报, 2021, 53(1): 1-16 (Xie Chenyue, Yuan Zelong, Wang Jianchun et al. Artificial neural network-based subgrid-scale models for large-eddy simulation of turbulence. Chinese Journal of Theoretical and Applied Mechanics, 2021, 53(1): 1-16 (in Chinese) [12] 张伟伟, 寇家庆, 刘溢浪. 智能赋能流体力学展望. 航空学报, 2021, 42(4): 26-71 (Zhang Weiei, Kou Jiaqing, Liu Yilang. Prospect of artificial intelligence empowering fluid mechanics. Acta Aeronautica et Astronautica Sinica, 2021, 42(4): 26-71 (in Chinese) [13] Chen SS, Li JP, Li Z, et al. Anti-dissipation pressure correction under low mach numbers for godunov-type schemes. Journal of Computational Physics, 2022, 456: 111027 doi: 10.1016/j.jcp.2022.111027 [14] Abbasi S, Pirker S, Lichtenegger T. Application of recurrence CFD(rCFD) to species transport in turbulent vortex shedding. Computers & Fluids, 2020, 196: 104348 [15] Steelant J, Dick E. Modelling of bypass transition with conditioned navier–stokes equations coupled to an intermittency transport equation. International Journal for Numerical Methods in Fluids, 1996, 23(3): 193-220 doi: 10.1002/(SICI)1097-0363(19960815)23:3<193::AID-FLD415>3.0.CO;2-2 [16] Langtry RB, Menter FR. Correlation-based transition modeling for unstructured parallelized computational fluid dynamics codes. AIAA Journal, 2009, 47(12): 2894-2906 doi: 10.2514/1.42362 [17] 王亮, 符松. 一种适用于超音速边界层的湍流转捩模式. 力学学报, 2009, 41(2): 162-168 (Wang Liang, Fu Song. A new transition/turbulence model for the flow transition in supersonic boundary layer. Chinese Journal of Theoretical and Applied Mechanics, 2009, 41(2): 162-168 (in Chinese) doi: 10.3321/j.issn:0459-1879.2009.02.003 [18] 盛蕾, 陈希亮, 康凯. 神经网络非梯度优化方法研究进展. 计算机工程与应用, 2022, 58(17): 34-49Sheng Lei, Chen Xiliang, Kang Kai. Research progress of neural network based on non-gradient optimization method. Computer Engineering and Applications, 2022, 58(17): 34-49 (in Chinese) [19] 袁先旭, 陈坚强, 杜雁霞等. 国家数值风洞(NNW)工程中的CFD基础科学问题研究进展. 航空学报, 2021, 42(9): 31-48 (Yuan Xianxu, Chen Jianqiang, Du Yanxia, et al. Research progress on fundamental CFD issues in national numerical windtunnel project. Acta Aeronautica et Astronautica Sinica, 2021, 42(9): 31-48 (in Chinese) [20] Brunton SL, Noack BR, Koumoutsakos P. Machine learning for fluid mechanics. Annual Review of Fluid Mechanics, 2020, 52(1): 477-508 doi: 10.1146/annurev-fluid-010719-060214 [21] Ling JL, Kurzawski A, Templeton J, et al. Reynolds averaged turbulence modelling using deep neural networks with embedded invariance. Journal of Fluid Mechanics, 2016, 807: 155-166 doi: 10.1017/jfm.2016.615 [22] 任海杰, 袁先旭, 陈坚强等. 宽速域下神经网络对雷诺应力各向异性张量的预测. 力学学报, 2022, 54(2): 347-358 (Ren Haijie, Yuan Xianxu, Chen Jianqiang et al. Prediction of Reynolds stress anisotropic tensor by neural network within wide speed range. Chinese Journal of Theoretical and Applied Mechanics, 2022, 54(2): 347-358 (in Chinese) doi: 10.6052/0459-1879-21-518 [23] Pope SB. A more general effective-viscosity hypothesis. Journal of Fluid Mechanics, 1975, 72(2): 331-340 [24] Wei SY, Jin XW, Li H. General solutions for nonlinear differential equations: a rule-based self-learning approach using deep reinforcement learning. Computational Mechanics. 2019, 64(5): 1361-1374 [25] Sekar V, Khoo BC. Fast flow field prediction over airfoils using deep learning approach. Physics of Fluids, 2019, 31(5): 057103 doi: 10.1063/1.5094943 [26] Zhu LY, Zhang WW, Sun XX, et al. Turbulence closure for high Reynolds number airfoil flows by deep neural networks. Aerospace Science and Technology, 2020, 110: 106452 [27] 张珍, 叶舒然, 岳杰顺等. 基于组合神经网络的雷诺平均湍流模型多次修正方法. 力学学报, 2021, 53(6): 1532-1542 (Zhang Zhen, Ye Shuran, Yue Jieshun et al. A combined neural network and multiple modification strategy for Reynolds-Averaged Navier-Stokes turbulence modeling. Chinese Journal of Theoretical and Applied Mechanics, 2021, 53(6): 1532-1542 (in Chinese) doi: 10.6052/0459-1879-21-073 [28] Beetham S, Capecelatro J. Formulating turbulence closures using sparse regression with embedded form invariance. Physical Review Fluids, 2020, 5(8): 084611 doi: 10.1103/PhysRevFluids.5.084611 [29] 郑天韵, 王圣业, 王光学等. 基于深度残差网络的高精度自然转捩模拟方法. 物理学报, 2020, 69(20): 275-286 (Zheng Tianyun, Wang Shengye, Wang Guangxue, et al. High-order natural transition simulation method based on deep residual netwo. Acta Physica Sinica, 2020, 69(20): 275-286 (in Chinese) doi: 10.7498/aps.69.20200563 [30] Foroozan F, Guerrero V, Ianiro A, et al. Unsupervised modelling of a transitional boundary layer. Journal of Fluid Mechanics, 2021, 929: 1-25 [31] Xiang XH, Ren HJ, Zhang YF, et al. Transition prediction with hypersonic cross-flow model on HIFiRE-5. Journal of Physics, 2021, 1786: 012051 [32] 向星皓, 张毅峰, 袁先旭等. C-γ-Reθ高超声速三维边界层转捩预测模型. 航空学报, 2021, 42(9): 196-204 (Xiang Xinghao, Zhang Yifeng, Yuan Xianxu et al. C-γ-Reθ model for hypersonic three-dimensional boundary layer transition prediction. Acta Aeronautica et Astronautica Sinica, 2021, 42(9): 196-204 (in Chinese) [33] 张毅锋, 雷净, 张益荣等. 高超声速数值模拟平台转捩模型的标定. 空气动力学学报, 2015, 33(1): 42-47 (Zhang Yifeng, Lei Jing, Zhang Yirong et al. Calibration of transition model for hypersonic numerical simulation platform. Acta Aerodynamica Sinica, 2015, 33(1): 42-47 (in Chinese) [34] Langtry RB, Sengupta K, Yeh DT, et al. Extending the γ-Reθt local correlation based transition model for crossflow effects//45th AIAA Fluid Dynamics Conference, Dallas TX, 2015: 393-404 [35] 向星皓, 张毅锋, 陈坚强等. 横流转捩模型参数不确定度量化分析与应用研究. 宇航学报, 2020, 41(9): 1141-1150 (Xiang Xinghao, Zhang Yifeng, Chen Jianqiang, et al. Uncertainty quantification analysis and application research on cross-flow transition model parameters. Journal of Astronautics, 2020, 41(9): 1141-1150 (in Chinese) doi: 10.3873/j.issn.1000-1328.2020.09.004 [36] 孔维萱, 张辉, 阎超. 适用于高超声速边界层的转捩准则预测方法. 导弹与航天运载技术. 2013, 5: 54-58Kong Weixuan, Zhang Hui, Yan Chao. Transition criterion prediction method for hypersonic boundary layer. Acta Aeronautica et Astronautica Sinica, 2013, 5: 54-58 (in Chinese) [37] 周玲, 阎超, 郝子辉, 孔维萱等. 转捩模式与转捩准则预测高超声速边界层流动. 航空学报, 2016, 37(4): 1092-1102 (Zhou Ling, Yan Chao, Hao Zihui, et al. Transition model and transition criteria for hypersonic boundary layer flow. Acta Aeronautica et Astronautica Sinica, 2016, 37(4): 1092-1102 (in Chinese) [38] 陈坚强, 涂国华, 万兵兵等. HyTRV流场特征与边界层稳定性特征分析. 航空学报, 2021, 42(6): 264-279 (Chen Jianqiang, Tu Guohua, Wan Bingbing, et al. Characteristics of flow field and boundary-layer stability of HyTRV. Acta Aeronautica et Astronautica Sinica, 2021, 42(6): 264-279 (in Chinese) [39] 王玉轩, 徐家宽, 张扬等. 横流驻波增长因子模式在跨声速边界层的应用, http://kns.cnki.net/kcms/detail/51.1192.TK.20211220.1139.010.htmWang Yuxuan, Xu Jiakuan, Zhang Yang, et al. Applications of transition model based on amplification factor of stationary crossflow waves in transonic boundary layers, http://kns.cnki.net/kcms/detail/51.1192.TK.20211220.1139.010.htm (in Chinese)) [40] Menter FR. Two-equation eddy-viscosity turbulence models for engineering applications. AIAA Journal, 1994, 32: 1598-1605 [41] Radeztsky, Ronald H, Reibert MS, et al. Effect of isolated micron-sized roughness on transition in swept-wing flows. AIAA Journal, 1999, 37: 1370-1377 doi: 10.2514/2.635 [42] 史亚云, 白俊强, 华俊等. 基于当地变量的横流转捩预测模型的研究与改进. 航空学报, 2016, 37(3): 780-789Shi Yayun, Bai Junqiang, Hua Jun, et al. Study and modification of cross-flow induced transition model based on local variables. Acta Aeronautica et Astronautica Sinica, 2016, 37(3): 780-789 (in Chinese) [43] 蔡林峰, 李昊歌, 陈伟芳. 无限展长后掠翼边界层定常横流不稳定转捩分析. 计算物理, 2019, 36(2): 175-181 (Cai Linfeng, Li Haoge, Chen Weifang. Analysis of steady cross-flow instability transition in infinite swept wing boundary layer. Chinese Journal of Computational Physics, 2019, 36(2): 175-181 (in Chinese) doi: 10.19596/j.cnki.1001-246x.7830 [44] Grabe C, Krambein A. Correlation-based transition transport modeling for three-dimensional aerodynamic configurations. Journal of Aircraft, 2013, 50(5): 1533-1539 doi: 10.2514/1.C032063 -