|
|
论文:2021,Vol:39,Issue(6):1212-1221 |
|
|
引用本文： |
|
|
贾续毅, 龚春林, 李春娜. 基于POD和BPNN的流场快速计算方法[J]. 西北工业大学学报 |
|
|
JIA Xuyi, GONG Chunlin, LI Chunna. Fast flow simulation method based on POD and BPNN[J]. Northwestern polytechnical university |
|
|
|
|
|
|
|
| 基于POD和BPNN的流场快速计算方法 |
|
| 贾续毅, 龚春林, 李春娜 |
|
| 西北工业大学 航天学院, 陕西 西安 710072 |
| 摘要: |
| 采用高精度CFD仿真进行大量流场分析存在计算成本高、耗时长的问题。提出了一种基于本征正交分解（proper orthogonal decomposition，POD）和反向传播神经网络（back propagation based neural network，BPNN）的流场快速计算方法。在几何参数化设计空间中抽样，然后利用POD将高维流场数据映射到低维基模态空间，并用BPNN建立几何参数到基模态系数的多层神经网络模型，实现流场快速预测。在POD和BPNN模型构建中分别引入分区和聚类取样策略，以提高建模效率，降低模型训练耗时。变几何翼型的定常流场案例结果表明：在亚声速情况下，训练所得的模型可以保证流场中等压线、翼面压力系数等信息的预测精度，其升阻力系数平均预测误差在0.4%之内；在跨声速情况下，训练所得的模型升阻力系数平均预测误差在1.4%之内，并且激波位置也可以得到较准确的预测。 |
| 关键词:
本征正交分解
反向传播神经网络
CFD
聚类
分区
|
|
| Fast flow simulation method based on POD and BPNN |
|
| JIA Xuyi, GONG Chunlin, LI Chunna |
|
| School of Astronautics, Northwestern Polytechnical University, Xi'an 710072, China |
| Abstract: |
| It is computationally expensive to perform a large number of flow analysis by using high-fidelity CFD simulation. The paper proposes a fast flow-analysis method based on the proper orthogonal decomposition(POD) and back propagation based neural network(BPNN). First samples are generated in the geometric parameter space. Then a POD model is built to map the high-dimensional flow-field data to low-dimensional base modal coefficients, and further a BPNN model is fitted from geometric parameters to base modal coefficients to achieve fast flow prediction. During constructing the POD and BPNN model, the partitioning strategy and K-means clustering are implemented to improve modeling efficiency, as well as to reduce model training time. The results of predicting the steady flow of variable geometries show that:at subsonic, the trained model possesses good accuracy, especially in predicting the pressure isolines of the flow field and the pressure coefficient distribution of the airfoil. The average prediction errors of the lift and drag coefficient are smaller than 0.4%, while they are smaller than 1.4% at transonic. The shock location can be well predicted as well. |
| Key words:
proper orthogonal decomposition
back propagation based neural network
CFD
clustering
partition strategy
|
|
| 收稿日期: 2021-03-31
修回日期:
|
| DOI: 10.1051/jnwpu/20213961212 |
| 通讯作者: 龚春林(1980-),西北工业大学教授,主要从事飞行器总体设计研究。e-mail:leonwood@nwpu.edu.cn
Email：leonwood@nwpu.edu.cn |
| 作者简介: 贾续毅(1999-),西北工业大学博士研究生,主要从事智能流体技术研究。
|
|
|
|
|
|
|
|
参考文献: |
|
|
[1] GHOREYSHI M, JIRASEK A, MILLER T, et al. Implementation and verification of gust modeling in an open-source flow solver[J]. Aerospace Science and Technology, 2019, 92(6):777-789
[2] 乔磊, 白俊强, 邱亚松, 等. 一种跨声速定常流场求解加速方法[J]. 北京航空航天大学学报, 2018, 44(3):470-479 QIAO Lei, BAI Junqiang, QIU Yasong, et al. High-efficiency solving method for steady transonic flow field[J]. Journal of Beijing University of Aeronautics and Astronautics, 2018, 44(3):470-479(in Chinese)
[3] 李大力. 结构网格高精度CFD高效隐式求解及众核并行计算[D]. 长沙:国防科技大学, 2017 LL Dali. Research on efficient implicit solver and many-core parallel optimization for structured high-order CFD[D]. Changsha:National University of Defense Technology, 2017(in Chinese)
[4] DENIL M, SHAKIBI B, DINH L, et al. Predicting parameters in deep learning[C]//Proceedings of the 26th International Conference on Neural Information Processing Systems, Lake Tahoe, 2013
[5] 杨文慧. 基于参数和特征冗余的神经网络模型压缩方法[D]. 西安:西安电子科技大学, 2018 YANG Wenhui. Neural network model compression methods based on parameters and features redundancy[D]. Xi'an:Xidian University, 2018(in Chinese)
[6] DU J, FANG F, PAIN C C, et al. POD reduced-order unstructured mesh modeling applied to 2D and 3D fluid flow[J]. Computers & Mathematics with Applications,2013,65(3):362-379
[7] JOHN Burkardt, MAX Gunzburger, LEE Hyungchun. POD and CVT-based reduced-order modeling of Navier-Stokes flows[J]. Computer Methods in Applied Mechanics and Engineering, 2006, 196(1):337-355
[8] 邱亚松, 白俊强, 华俊. 基于本征正交分解和代理模型的流场预测方法[J]. 航空学报, 2013, 34(6):1249-1260 QIU Yasong, BAI Junqiang, HUA Jun. Flow field estimation method based on proper orthogonal decomposition and surrograte model[J]. Acta Aeronautica et Astronautica Sinica, 2013, 34(6):1249-1260(in Chinese)
[9] 王晨,白俊强, JAN Shehaven, 等. POD-Kriging降阶方法在串联双圆柱流场预测中的应用[J]. 西北工业大学学报, 2018, 36(2):220-228 WANG Chen, BAI Junqiang, JAN Shehaven, et al. POD-Kriging reduced method's application in tandem cylinders' flow[J]. Journal of Northwestern Polytechnical University, 2018, 36(2):220-228(in Chinese)
[10] 陈海昕, 邓凯文, 李润泽. 机器学习技术在气动优化中的应用[J]. 航空学报, 2019, 40(1):522480 CHEN Haixin, DENG Kaiwen, LI Runze. Utilization of machine learning technology in aerodynamic optimization[J]. Acta Aeronautica et Astronautica Sinica, 2019, 40(1):522480(in Chinese)
[11] LI K, KOU J, ZHANG W. Deep neural network for unsteady aerodynamic and aeroelastic modeling across multiple Mach numbers[J]. Nonlinear Dynamics, 2019, 96(5):1-21
[12] SALEEM W, KHARAL A, AHMAD R, et al. Comparison of ACO and GA techniques to generate neural network based Bezier-Parsec parameterized airfoil[C]//The 201511th International Conference on Natural Computation, 2015
[13] ZHU L, ZHANG W, KOU J, et al. Machine learning methods for turbulence modeling in subsonic flows around airfoils[J]. Physics of Fluids, 2019, 31(1):015105
[14] 刘长龙. 从机器学习到深度学习:基于scikit-learn与TensorFlow的高效开发实战[M]. 北京:电子工业出版社, 2019:173-179 LIU Changlong. From machine learning to deep learning:effective development based on scikit-learn and TensorFlow[M]. Beijing:Electronic Industry Press, 2019:173-179(in Chinese)
[15] KIM P. MATLAB deep learning:with machine learning neural networks and artificial intelligence[M]. Berkeley:Apress, 2017, 54-60
[16] LIANG R, LI T, LI L, et al. Knowledge consistency between neural networks and beyond[C]//International Conference on Learning Representations, Addis Ababa, 2019
[17] PAN V Y, CHEN Z Q. The complexity of the matrix eigenproblem[C]//Proceedings of the 31st Annual ACM Symposium on Theory of Computing, Atlanta, 1999 |
|
|
|
|
|
|
|
