基于稀疏环境监测点的流动时程重构模型精度研究

战庆亮; 刘鑫; 晁阳; 葛耀君

PDF(8876 KB)

中国环境科学 ›› 2023, Vol. 43 ›› Issue (12) : 6592-6600.

环境生态

基于稀疏环境监测点的流动时程重构模型精度研究

战庆亮¹, 刘鑫¹, 晁阳¹, 葛耀君²

作者信息 +

Accuracy of environmental flows time history reconstruction model based on sparse observation

ZHAN Qing-liang¹, LIU Xin¹, CHAO Yang¹, GE Yao-jun²

Author information +

文章历史 +

摘要

根据有限数量且稀疏分布的环境流场监测点处的观测数据,得到更多空间位置处的环境测点的流动时变信息,能够为大气监测、水体监测和污染物扩散等问题的控制与研究提供更加丰富的数据.本文研究了物理约束对机器学习流场时程表征模型的精度影响规律,研究了可用测点数以及物理约束权重对重构精度的影响.以低雷诺数方柱大气绕流为例,开展了基于稀疏环境监测点的数据进行高空间分辨率流场时程重构的精度研究.研究发现当监测点数量增至500时精度不再提高;当监测点数量仅有50时,约束权重为10可得到最优结果.结果表明,通过选择合适的物理约束影响权重,可以有效弥补可用数据较少的问题,为环境流动问题的数据处理和高分辨率流场重构提供了新的方法与依据.

Abstract

Obtaining more environmental flow data at more monitoring sites based on limited and sparse available flow monitoring points can provide data for the study of atmospheric monitoring, water monitoring and pollutant dispersion issues. In this study, the influence of physical constraints on the accuracy of the machine learning flow field time history representation model was investigated, and the results of different available measurement points and physical constraint influence weights were compared. The environmental flow around a low Reynolds number square column was tested as an example. Results show that when the number of monitoring points was increased to 500, the accuracy did not improve. When the number of monitoring points was reduced to only 50, the error was minimised when the physical constraint weights were set to 10. The results indicate that the problem of less available data can be effectively compensated by choosing appropriate physical constraint influence weights. Providing a new method and basis for data processing and high-resolution flow field reconstruction for environmental flow problems.

导出引用

战庆亮, 刘鑫, 晁阳, 葛耀君. 基于稀疏环境监测点的流动时程重构模型精度研究[J]. 中国环境科学. 2023, 43(12): 6592-6600

ZHAN Qing-liang, LIU Xin, CHAO Yang, GE Yao-jun. Accuracy of environmental flows time history reconstruction model based on sparse observation[J]. China Environmental Science. 2023, 43(12): 6592-6600

中图分类号： X83

参考文献

[1] 周琪,于洋,刘苗苗,等.基于机器学习和非参数估计的PM_2.5风险评估[J]. 中国环境科学, 2022,42(8):3554-3560. Zhou Q, Yu Y, Liu M M, et al. Risk assessment of PM_2.5 pollution based on machine learning and nonparametric estimation[J]. China Environmental Science, 2022,42(8):3554-3560.
[2] 周恒左,陈恒蕤,廖鹏,等.兰州市CMAQ近地面臭氧模拟结果的订正方法——基于机器学习方法[J]. 中国环境科学, 2022,42(12): 5472-5483. Zhou H Z, Chen H R, Liao P, et al. A revised approach to CMAQ near-surface ozone modelling results in Lanzhou-Based on machine learning methods[J]. China Environmental Science, 2022,42(12): 5472-5483.
[3] 嵇晓燕,姚志鹏,杨凯,等.基于MSLSTM-DA模型的水质自动监测异常数据报警[J]. 中国环境科学, 2022,42(4):1877-1883. Ji X Y, Yao Z P, Yang K, et al. Water quality alert with automatic monitoring data based on MSLSTM-DA model[J]. China Environmental Science, 2022,42(4):1877-1883.
[4] 任永琴,金柱成,俞真元,等.基于双向门控循环单元的地表水氨氮预测[J]. 中国环境科学, 2022,42(2):672-679. Ren Y Q, Kim J S, Yu J W, et al. Ammonia nitrogen prediction in surface water based on bidirectional gated recurrent unit[J]. China Environmental Science, 2022,42(2):672-679.
[5] 张寒博,杨骥,荆文龙,等.多种特征因子结合GBDT的降水数据降尺度方法研究[J]. 中国环境科学, 2023,43(4):1867-1882. Zhang H B, Yang J, Jing W L, et al. Downscaling method of precipitation data based on GBDT combined with multiple eigenfactors[J]. China Environmental Science, 2023,43(4):1867-1882.
[6] 李静波,张莹,盖荣丽.基于机器学习的星载短波红外CO₂柱浓度估算[J]. 中国环境科学, 2023,43(4):1499-1509. Li J B, Zhang Y, Gai R L. Estimation of the column concentration of carbon dioxide using spaceborne shortwave infrared spectrometer[J]. China Environmental Science, 2023,43(4):1499-1509.
[7] 葛渊博,卢文喜,白玉堃,等.基于SSA-BP与SSA的地下水污染源反演识别[J]. 中国环境科学, 2022,42(11):5179-5187. Ge Y B, Lu W X, Bai Y K, et al. Inversion and identification of groundwater pollution sources based on SSA-BP and SSA[J]. China Environmental Science, 2022,42(11):5179-5187.
[8] 潘紫东,卢文喜,范越,等.基于模拟-优化方法的地下水污染源溯源辨识[J]. 中国环境科学, 2020,40(4):1698-1705. Pan Z D, Lu W X, Fan Y, et al. Inverse Identification of groundwater pollution source based on simulation-optimization approach[J]. China Environmental Science, 2020,40(4):1698-1705.
[9] 张伟伟,窦家庆,刘溢浪.智能赋能流体力学展望[J]. 航空学报, 2021,42(4):524689. Zhang W W, Kou J Q, Liu Y L. Prospect of artificial intelligence empowered fluid mechanics[J]. Acta Aeronautica et Astronautica Sinica, 2021,42(4):524689.
[10] Zhou L, Lu X, Yang L. A local structure adaptive super-resolution reconstruction method based on BTV regularization[J]. Multimedia Tools and Applications, 2014,71(3):1879-1892.
[11] Lim B, Son S, Kim H, et al. Enhanced deep residual networks for single image super-resolution[A]//O'Conner L.2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)[C]. Piscataway: IEEE, 2017:1132-1140.
[12] Tai Y, Yang J, Liu X. Image super-resolution via deep recursive residual network[A]//O'Conner L.2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)[C]. Piscataway: IEEE, 2017:2790-2798.
[13] Lai W S, Huang J B, Ahuja N, et al. Deep laplacian pyramid networks for fast and accurate super-resolution[A]//O'Conner L.2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR)[C]. Piscataway: IEEE, 2017:5835-5843.
[14] 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):25105.
[15] Xie Y, Franz E, Chu M, et al. Tempo GAN: A temporally coherent, volumetric GAN for super-resolution fluid flow[J]. ACM Trans. Graph., 2018,37(4):1-15.
[16] Xu W, Luo W, Wang Y, et al. Data-driven three-dimensional super-resolution imaging of a turbulent jet flame using a generative adversarial network[J]. Applied Optics, 2020,59(19):5729-5736.
[17] Lee J, Lee S, You D. Deep learning approach in multi-scale prediction of turbulent mixing-layer[J]. arXiv: Computational Physics, 2018.doi:10.48550/arXiv.1809.07021.
[18] Lin J, Lensink K, Haber E. Fluid flow mass transport for generative networks[A]//RUSH A, TECH C.ICLR 2020 Conference Blind Submission.[C]. Addis Ababa, Ethiopia: Open Review. net, 2020: 1-10.
[19] 战庆亮,白春锦,葛耀君.基于时程深度学习的复杂流场流动特性表征方法[J]. 物理学报, 2022,71(22):224701. Zhan Q L, Bai C J, Ge Y J. Deep learning representation of flow time history for complex flow field[J]. Acta Physica Sinica, 2022,71(22):224701.
[20] 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 Q L, Bai C J, Ge Y J. Fluid feature analysis based on time history deep learning[J]. Chinese Journal of Theoretical and Applied Mechanics, 2022,54(3):822-828.
[21] 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):526531. Zhan Q L, Bai C J, Zhang N et al. Feature extraction method of flow around airfoil based on time-history convolutional autoencoder[J]. Acta Aeronautica et Astronautica Sinica, 2022,43(11):526531.
[22] 战庆亮,葛耀君,白春锦.基于深度学习的流场时程特征提取模型研究[J]. 物理学报, 2022,71(7):074701. Zhan Q L, Bai C J, Ge Y J. Flow feature extraction models based on deep learning[J]. Acta Physica Sinica, 2022,71(7):074701.
[23] 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]. 工程力学, 2023,40(2):17-24. Zhan Q L, Bai C J, Ge Y J. Deep learning method for flow feature recognition based on dimensionless time history[J]. Engineering Mechanics, 2023,40(2):17-24.
[24] Raissi M, Karniadakis G E. Hidden physics models: Machine learning of nonlinear partial differential equations[J]. Journal of Computational Physics, 2018,357:125-141.
[25] Raissi M, Perdikaris P, Karniadakis G E. Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations[J]. Journal of Computational Physics, 2019,378:686-707.
[26] 战庆亮,刘鑫,白春锦,等.考虑物理方程约束的机器学习流场时程表征方法[J]. 工程力学, 2023.doi:10.6052/j.issn.1000-4750.2022. 12.1067. Zhan Q L, Liu X, Bai C J, et al. Physical constrained flow representation model using machine learning for flow time history[J]. Engineering Mechanics[J]. Engineering Mechanics, 2023.doi:10. 6052/j.issn.1000-4750.2022.12.1067.