卷积神经网络方法在涌潮水动力特性演变中的应用研究

doi:10.3969/j.issn.1001-909X.2024.03.011

海洋学研究 ›› 2024, Vol. 42 ›› Issue (3): 131-141.DOI: 10.3969/j.issn.1001-909X.2024.03.011

卷积神经网络方法在涌潮水动力特性演变中的应用研究

王智弘¹(), 屈科¹^,³^,⁴^,^*(), 杨元平², 王旭¹^,³, 高榕泽¹

1.长沙理工大学水利与环境工程学院,湖南长沙 410114
2.浙江省水利河口研究院,浙江杭州 310017
3.洞庭湖水环境治理与生态修复湖南省重点实验室,湖南长沙 410114
4.水沙科学与水灾害防治湖南省重点实验室,湖南长沙 410114

收稿日期:2023-10-12 修回日期:2024-02-26 出版日期:2024-09-15 发布日期:2024-11-25
通讯作者: ^*屈科(1985—),男,副教授,主要从事计算流体力学、海岸工程、海洋工程等方面的研究,E-mail:kqu@csust.edu.cn。
作者简介:王智弘(2003—),男,湖南省娄底市人,主要从事波浪水动力方面的研究,E-mail: 202104330130@stu.csust.edu.cn。
基金资助:
国家重点研发计划(2022YFC3103601);浙江省河口海岸重点实验室开放基金项目(ZIHE21009);浙江省自然科学基金项目(LY22E090007);浙江省水利厅科技计划项目(RC2020);湖南省省级大学生创新创业训练计划项目(S202310536108)

Application of convolutional neural network method in evolution of tidal bore hydrodynamic characteristics

WANG Zhihong¹(), QU Ke¹^,³^,⁴^,^*(), YANG Yuanping², WANG Xu¹^,³, GAO Rongze¹

1. School of Hydraulic and Environmental Engineering, Changsha University of Science & Technology, Changsha 410114, China
2. Zhejiang Water Conservancy and Estuary Research Institute, Hangzhou 310017, China
3. Key Laboratory of Dongting Lake Aquatic Eco-Environmental Control and Restoration of Hunan Province, Changsha 410114, China
4. Key Laboratory of Water-Sediment Sciences and Water Disaster Prevention of Hunan Province, Changsha 410114, China

Received:2023-10-12 Revised:2024-02-26 Online:2024-09-15 Published:2024-11-25

摘要/Abstract

摘要：

该文基于开源软件OpenFOAM求解雷诺平均的Navier-Stokes方程,实现了对于波状涌潮(Fr=1.2~1.3)的数值模拟,并与物理实验数据比较,验证了数值模拟的准确性。使用CONV1D卷积神经网络模型对数值模拟数据进行了学习,预测出在具有斜坡地形上的涌潮水动力发展过程。对比涌潮到达x=30.0 m 测点处的用时和该测点的最高水位的模型预测结果与数值模拟结果可知:到达用时的平均相对误差为2.28%,最高水位的平均相对误差为3.73%。较小的相对误差证明了CONV1D模型的准确性。该文对于涌潮的水动力过程模拟、与涌潮相关的灾害预警以及初生涌潮未来发展趋势研究都具有一定意义。

关键词: 涌潮, 钱塘江, OpenFOAM, 水动力过程, 斜坡地形, 卷积神经网络, CONV1D, 数值模拟

Abstract:

The open source software OpenFOAM was used to solve the Reynolds-averaged Navier-Stokes equations, numerical simulations of wave surges (Fr=1.2~1.3) was implemented, and the results were compared with physical experimental data to verify the accuracy of the numerical simulations. The data was studied based on the CONV1D convolutional neural network model to predict the dynamic development process of tidal bore on terrain with slope. Comparison of the model prediction and numerical simulation results for the arrival time and the maximum water level of the tidal bore at x=30.0 m shows that the average relative error in the arrival time is 2.28%, and the average relative error in the maximum water level is 3.73%. The smaller relative error proves the accuracy of the CONV1D model. This result has certain research significance for the simulation of the hydrodynamic process of tidal bore, early warning of tidal bore disasters, and the study of future development trends of primary tidal bore.

Key words: tidal bore, Qiantang River, OpenFOAM, hydrodynamic process, slope, convolutional neural network, CONV1D, numerical simulation

中图分类号:

P731.23

王智弘, 屈科, 杨元平, 王旭, 高榕泽. 卷积神经网络方法在涌潮水动力特性演变中的应用研究[J]. 海洋学研究, 2024, 42(3): 131-141.

WANG Zhihong, QU Ke, YANG Yuanping, WANG Xu, GAO Rongze. Application of convolutional neural network method in evolution of tidal bore hydrodynamic characteristics[J]. Journal of Marine Sciences, 2024, 42(3): 131-141.

图/表 14

图1 物理实验验证工况布置图

Fig.1 Diagram of physical experimental verification conditions layout

表1 物理实验验证工况

Tab.1 Verification of physical experimental conditions

工况序号	潮前水深 h₀/m	波高H/m	潮后水深 h₁/m	潮后流速 u₁/(m·s^-1)
1	0.05	0.07	0.12	0.83
2	0.05	0.03	0.08	0.38

图2 物理实验验证工况下x=10 m处的水位时程曲线

Fig.2 Temporal evolution of water level in verification of physical experimental conditions at x=10 m

图3 物理实验验证工况下 x=10 m,y=0.05 m处的水平速度时程曲线

Fig 3. Temporal evolution of horizontal flow velocity in verification of physical experimental conditions at x=10 m,y=0.05 m

图4 数值模拟工况布置示意图

Fig.4 Diagram of numerical simulation conditions layout

表2 数值模拟工况

Tab.2 Numerical simulation conditions

工况序号	潮前水深 h₀/m	波高H/m	潮后水深 h₁/m	潮后流速 u₁/(m·s^-1)	Fr	工况序号	潮前水深 h₀/m	波高H/m	潮后水深 h₁/m	潮后流速 u₁/(m·s^-1)	Fr
1	0.36	0.14	0.50	0.678	1.288	14	0.42	0.16	0.58	0.718	1.282
2	0.38	0.14	0.52	0.662	1.273	15	0.44	0.16	0.60	0.703	1.269
3	0.40	0.14	0.54	0.647	1.259	16	0.37	0.12	0.49	0.579	1.241
4	0.42	0.14	0.56	0.633	1.247	17	0.38	0.12	0.50	0.572	1.234
5	0.44	0.14	0.58	0.620	1.236	18	0.40	0.13	0.53	0.603	1.241
6	0.37	0.15	0.52	0.715	1.300	19	0.42	0.13	0.55	0.590	1.230
7	0.38	0.15	0.53	0.706	1.292	20	0.44	0.13	0.57	0.578	1.219
8	0.40	0.15	0.55	0.690	1.278	21	0.37	0.11	0.48	0.533	1.221
9	0.42	0.15	0.57	0.676	1.265	22	0.38	0.11	0.49	0.527	1.215
10	0.44	0.15	0.59	0.662	1.253	23	0.40	0.12	0.52	0.559	1.223
11	0.37	0.13	0.50	0.624	1.260	24	0.42	0.12	0.54	0.547	1.212
12	0.38	0.13	0.51	0.617	1.254	25	0.44	0.12	0.56	0.535	1.203
13	0.40	0.16	0.56	0.734	1.296

图5 在1~5号数值模拟工况下x=10 m处的水位时程曲线

Fig.5 Temporal evolution of water level in numerical simulation conditions 1-5 at x=10 m

图6 在1~5号数值模拟工况下x=15 m处的水位时程曲线

Fig.6 Temporal evolution of water level in numerical simulation conditions 1-5 at x=15 m

图7 卷积神经网络结构图

Fig.7 Structure diagram of convolutional neural network

表3 测试工况设置

Tab.3 Test conditions

工况序号	1	2	3	4	5	6	7	8	9	10
水深/m	0.40	0.40	0.38	0.37	0.42	0.39	0.41	0.45	0.46	0.38
波高/m	0.16	0.13	0.14	0.11	0.12	0.13	0.12	0.15	0.17	0.15

图8 不同测试工况下CONV1D模型预测结果与数值模拟结果对比

Fig.8 Comparison of water level from CONV1D model predictions and numerical simulations under different test conditions

图9 不同测试工况下CONV1D模型预测结果与数值模拟结果的误差平方和对比

Fig.9 Comparison of the sum of squares of errors between CONV1D model predictions and numerical simulations under different test conditions

表4 不同测试工况下涌潮到达 x=30.0 m 测点处的用时和该测点最高水位的CONV1D模型预测值与数值模拟值的误差分析

Tab.4 Error analysis between CONV1D model predicted and numerical simulations of the time taken for the tidal bore reaching and maximum water level at x=30.0 m under different test conditions

工况序号	涌潮到达 x=30.0 m 测点处的用时			x=30.0 m 测点处的最高水位
工况序号	数值模拟值/s	CONV1D模型预测值/s	相对误差/%	数值模拟值/m	CONV1D模型预测值/m	相对误差/%
1	11.0	11.1	0.91	0.611	0.596	2.45
2	12.0	12.0	0.00	0.575	0.553	3.83
3	11.5	11.5	0.00	0.567	0.531	6.35
4	12.4	14.0	12.90	0.530	0.498	6.04
5	12.0	12.0	0.00	0.595	0.569	4.37
6	11.7	11.7	0.00	0.575	0.547	4.87
7	12.0	12.0	0.00	0.585	0.562	3.93
8	11.2	11.2	0.00	0.658	0.649	1.37
9	10.8	10.7	0.93	0.684	0.690	0.88
10	11.2	12.1	8.04	0.562	0.544	3.20

图10 不同测试工况下涌潮到达 x=30.0 m 测点处的用时和该测点的最高水位的相对误差

Fig.10 Relative error of maximum water level and time taken for the tidal bore reaching under different test conditions at x=30.0 m

参考文献 28

[1]	中国水利百科编委会. 中国水利百科全书[M].第二版. 北京: 水利水电出版社,2006:1623.
	Chinese Water Conservancy Encyclopedia Editorial Board. Chinese water conservancy encyclopedia[M]. 2nd ed. Beijing: Water Conservancy and Hydropower Press, 2006: 1623.
[2]	林炳尧, 潘存鸿. 涌潮与激波[J]. 自然杂志, 2015, 37(2):134-142.
	LIN B Y, PAN C H. Tidal bore and shock wave[J]. Chinese Journal of Nature, 2015, 37(2): 134-142.
[3]	黄婷, 张怀, 石耀霖. 基于Boussinesq型方程的钱塘江涌潮数值模拟[J]. 地球物理学报, 2022, 65(1):79-95. DOI
	HUANG T, ZHANG H, SHI Y L. Numerical simulation of the tidal bore in the Qiantang River based on Boussinesq-type equations[J]. Chinese Journal of Geophysics, 2022, 65(1): 79-95.
[4]	DE SAINT-VENANT A J C B. Théorie et équations générales du mouvement non permanent des eaux courantes[J]. Comptes Rendus des séances de l’Académie des Sciences, 1871, 73: 147-154.
[5]	苏铭德, 徐昕, 朱锦林, 等. 数值模拟在钱塘江涌潮分析中的应用:Ⅰ.数值计算方法[J]. 力学学报, 1999, 31(5):521-533. DOI
	SU M D, XU X, ZHU J L, et al. The application of numerical simulation in analysis of Qiantangjiang tide——Ⅰ. Numerical method[J]. Acta Mechanica Sinica, 1999, 31(5): 521-533.
[6]	林炳尧, 黄世昌, 毛献忠. 波状水跃和波状涌潮的分析[J]. 水动力学研究与进展:A辑, 1998, 13(1):106-115.
	LIN B Y, HUANG S C, MAO X Z. Analyses of undular hydraulic jump and undular bore[J]. Journal of Hydrody-namics, 1998, 13(1): 106-115.
[7]	林炳尧, 黄世昌, 周潮生. 涌潮的反射和碰撞:回头潮和交叉潮[J]. 浙江水利水电专科学校学报, 2000, 12(1): 6-10.
	LIN B Y, HUANG S C, ZHOU C S. Reflection and collision of tidal bore: Turning tide and crossing tide[J]. Journal of Zhejiang Water Conservancy and Hydropower College, 2000, 12(1): 6-10.
[8]	FAN J, TAO A F, SHI M Q, et al. Flume experiment investigation on propagation characteristics of tidal bore in a curved channel[J]. China Ocean Engineering, 2023, 37(1): 131-144.
[9]	戚蓝, 肖厅厅, 张芝永, 等. 涌潮水流CFD数值模拟[J]. 水利水运工程学报, 2019(3):32-40.
	QI L, XIAO T T, ZHANG Z Y, et al. Numerical simulation of tidal bore based on CFD method[J]. Hydro-Science and Engineering, 2019(3): 32-40.
[10]	PUTRA Y S, NOVIANI E, NURHASANAH, et al. A numerical study of hydro-hydraulic energy on undular tidal bore phenomenon[J]. Journal of Physics: Conference Series, 2021, 1816(1): 012067.
[11]	陈志根. 钱江涌潮的自然文化特征和保护利用[J]. 浙江水利水电学院学报, 2021, 33(2):6-9.
	CHEN Z G. On natural and cultural characteristics and protection and utilization of Qianjiang tidal bore[J]. Journal of Zhejiang University of Water Resources and Electric Power, 2021, 33(2): 6-9.
[12]	鲁海燕, 潘存鸿, 陈刚. 杭州湾天文潮长周期变化对涌潮的影响[J]. 水利水电科技进展, 2023, 43(4):25-30.
	LU H Y, PAN C H, CHEN G. Influence of long-period variation of astronomical tide on tidal bore in Hangzhou Bay[J]. Advances in Science and Technology of Water Resources, 2023, 43(4): 25-30.
[13]	沈阿四. 钱塘江涌潮形成的历史探索[J]. 浙江水利科技, 2002, 30(3):75-76.
	SHEN A S. Historical exploration on the formation of tidal bore in Qiantang River[J]. Zhejiang Hydrotechnics, 2002, 30(3): 75-76.
[14]	金晓剑, 谢骏, 沈利勤. 杭州市钱塘江防潮安全长效管理对策研究[J]. 水利发展研究, 2011, 11(6):64-66.
	JIN X J, XIE J, SHEN L Q. Study on long-term manage-ment countermeasures of moisture-proof safety of Qiantang River in Hangzhou[J]. Water Resources Development Research, 2011, 11(6): 64-66.
[15]	葛路, 张善亮, 许月萍, 等. 耦合BP神经网络的MIKE11模型预报无资料断面水位[J]. 水利水运工程学报, 2023(3):57-67.
	GE L, ZHANG S L, XU Y P, et al. Investigation of water level prediction at ungauged sections by coupled BP neural network and MIKE11 model[J]. Hydro-Science and Engineering, 2023(3): 57-67.
[16]	霍文博, 朱跃龙, 李致家, 等. 新安江模型和支持向量机模型实时洪水预报应用比较[J]. 河海大学学报:自然科学版, 2018, 46(4):283-289.
	HUO W B, ZHU Y L, LI Z J, et al. Comparison of Xin’anjiang model and support vector machine model in the application of real-time flood forecasting[J]. Journal of Hohai University: Natural Sciences, 2018, 46(4): 283-289.
[17]	姬战生, 章国稳, 张振林. 基于卷积神经网络的东苕溪瓶窑水文站水位预报[J]. 水电能源科学, 2021, 39(8):46-49.
	JI Z S, ZHANG G W, ZHANG Z L. Water level forecast of Pingyao hydrological station in Dongtiaoxi based on convolu-tional neural network[J]. Water Resources and Power, 2021, 39(8): 46-49.
[18]	MENTER F R. Two-equation eddy-viscosity turbulence models for engineering applications[J]. AIAA Journal, 1994, 32(8): 1598-1605.
[19]	MENTER F R. Improved two-equation k-ω turbulence models for aerodynamic flows[R]//NASA Technical Memorandum 103975, 1992.
[20]	ZHANG Z Y, PAN C H, ZENG J, et al. Hydrodynamics of tidal bore overflow on the spur dike and its influence on the local scour[J]. Ocean Engineering, 2022, 266: 113140.
[21]	潘存鸿, 郑君, 胡成飞, 等. 钱塘江涌潮影响因素研究[J]. 泥沙研究, 2023, 48(1):13-20.
	PAN C H, ZHENG J, HU C F, et al. Study on the impact factors of tidal bore on the Qiantang River[J]. Journal of Sediment Research, 2023, 48(1): 13-20.
[22]	WARNER J C, GEYER W R, LERCZAK J A. Numerical modeling of an estuary: A comprehensive skill assessment[J]. Journal of Geophysical Research: Oceans, 2005, 110: C05001.
[23]	WILLMOTT C J. On the validation of models[J]. Physical Geography, 1981, 2: 184-194.
[24]	谢东风, 潘存鸿, 陆波, 等. 基于实测资料的钱塘江涌潮水动力学特性研究[J]. 水动力学研究与进展A辑, 2012, 27(5):501-508.
	XIE D F, PAN C H, LU B, et al. A study on the hydrody-namic characteristics of the Qiantang tidal bore based on field data[J]. Chinese Journal of Hydrodynamics, 2012, 27(5): 501-508.
[25]	CHANSON H. Current knowledge in hydraulic jumps and related phenomena: A survey of experimental results[J]. European Journal of Mechanics-B/Fluids, 2008, 28(2): 191-210.
[26]	MAAS A L, HANNUN A Y, NG A Y. Rectifier nonlinearities improve neural network acoustic models[C]//Proceedings of the 30^th International Conference on Machine Learning, Atlanta, Georgia, USA, 2013.
[27]	SRIVASTAVA N, HINTON G, KRIZHEVSKY A, et al. Dropout: A simple way to prevent neural networks from overfitting[J]. Journal of Machine Learning Research, 2014, 15: 1929-1958.
[28]	KINGMA D P, BA J. Adam: A method for stochastic optimization[Z/OL]. arXiv, 2014: 1412.6980. https://arxiv.org/abs/1412.6980v9.

卷积神经网络方法在涌潮水动力特性演变中的应用研究

Application of convolutional neural network method in evolution of tidal bore hydrodynamic characteristics

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 14

参考文献 28

相关文章 15

编辑推荐

Metrics

[1]	徐广珺, 施宇诚, 余洋, 谢华荣, 谢文鸿, 刘婧媛, 林夏艳, 刘宇, 董昌明. 海洋涡旋智能检测研究进展[J]. 海洋学研究, 2024, 42(3): 38-50.
[2]	吕曌, 伍志元, 蒋昌波, 张浩键, 高凯, 颜仁. 基于海气耦合模式的超强台风“山竹”数值模拟[J]. 海洋学研究, 2023, 41(4): 21-31.
[3]	张良斌, 屈科, 黄竞萱, 王旭, 虢磊. 风对孤立波海堤越浪特性影响的数值模拟研究[J]. 海洋学研究, 2023, 41(4): 32-45.
[4]	张宸浩, 张明亮, 柴崇顼, 等. 基于OpenFOAM波浪—植物相互作用的数值模拟研究[J]. 海洋学研究, 2022, 40(1): 42-52.
[5]	张舒羽, 丁涛, 刘进宝, 等. 基于文献计量法的涌潮研究动态分析[J]. 海洋学研究, 2022, 40(1): 89-100.
[6]	徐浩波, 管清胜, 许明珠, 刁云云, . 蛇纹岩化作用对超慢速洋中脊拆离断层发育的影响[J]. 海洋学研究, 2021, 39(3): 21-30.
[7]	潘存鸿, 潘冬子, 郑君, 陈刚. 台风对钱塘江涌潮影响研究[J]. 海洋学研究, 2020, 38(4): 40-47.
[8]	周炜, 宣基亮, 黄大吉. 桑沟湾邻近海域岬角潮余流涡对及其产生机理[J]. 海洋学研究, 2020, 38(3): 10-20.
[9]	李志永, 程文龙, 金建峰, 张玉伦, 黄姿菡. 堤线调整对九溪涌潮景观影响评价研究[J]. 海洋学研究, 2019, 37(3): 79-85.
[10]	伍志元, 蒋昌波, 何智勇, 陈杰, 邓斌, 谢振东. 大气-海浪耦合模式及其在理想台风模拟中的应用研究[J]. 海洋学研究, 2019, 37(2): 9-15.
[11]	苏银秋, 潘国富, 俞亮亮, 杨万康. 海湾水交换对取排水工程的响应[J]. 海洋学研究, 2018, 36(4): 76-83.
[12]	张拂坤, 邹川玲, 刘淑静, 徐显, 刘伟. 晋江海域海水淡化排海盐度场三维数值模拟[J]. 海洋学研究, 2018, 36(2): 12-18.
[13]	黄宗伟, 邓斌, 蒋昌波, 刘晓建. 环抱式港区水体交换能力数值研究——以闸坡渔港为例[J]. 海洋学研究, 2018, 36(1): 66-74.
[14]	黄潘阳, 来向华, 季有俊, 胡涛骏, 王友忠. 舟山东港新城溃堤洪水演进数值模拟[J]. 海洋学研究, 2017, 35(4): 61-68.
[15]	邹逸航, 马旭林, 姜胜, 何海伦, 郭欢. COSMIC掩星资料同化对台风“天兔”预报影响的试验[J]. 海洋学研究, 2017, 35(3): 9-19.