摘要
近年来磨刀门水道咸潮上溯频发,引起广泛关注。对该地区建立了基于极大重叠离散小波变化(MODWT)和Elman神经网络的咸潮模拟模型(MOD-Elman),模拟预测日均氯度,并借助一维纵向扩散方程分析河段的含氯度分布。结果表明:(1)MOD-Elman模型对非平稳的含氯度变化模拟效果较优且泛化能力较好;(2)咸潮影响预测图在一定程度上能反映站点之间各点含氯度变化。利用MOD-Elman模型和一维纵向扩散方程,可预测未来一段时间内磨刀门的受咸潮影响情况,对预警压咸有一定帮助。
In recent years, there has been a frequent occurrence of saltwater intrusion in Modaomen estuary,causing widespread concern. In this paper, a MOD-Elman model based on MODWT and Elman ANN was established. It could simulate the average daily chlorinity and analyze the chlorinity distribution in the river with the one-dimensional longitudinal diffusion equation. Results showed as follow:(1)The MOD-Elman model performed well in simulating non-stationary chlorinity changes and had a good generalization ability;(2)The influence diagram of saltwater intrusion could reflect the variation of saltwater intrusion between Pinggang and Guadingjiao sites to a certain extent. This paper suggested that MOD-Elman model and one-dimensional longitudinal diffusion equation could be used to predict the variation of saltwater intrusion in Modaomen estuary, which had a practical significance for guaranteeing the quality security of water source.
引文
[1]刘祖发,丁波,关帅,等.磨刀门水道咸潮上溯数值模拟及其分析[J].中山大学学报(自然科学版),2016,55(6):1-9.
[2]GRIGG N.J.,IVEY G.N..A laboratory investigation into sheargenerated mixing in a salt wedge estuary[J].Geophysical and Astrophysical Fluid Dynamics,1997,85(1-2):65-95.
[3]刘祖发,关帅,张淦濠,等.基于FVCOM的虎门水道盐水入侵三维数值模拟[J].热带海洋学报,2016,35(2):10-18.
[4]刘德地,陈晓宏.基于偏最小二乘回归与支持向量机耦合的咸潮预报模型[J].中山大学学报(自然科学版),2007(4):89-92.
[5]杨兴果,张宏建,周洪亮.采用BP神经网络建立咸潮入侵预报模型——以钱塘江为例[J].中国建设信息,2010(11):49-51.
[6]沈汉堃,朱三华,谢淑琴,等.珠江河口BP神经网络的咸潮预报模型研究[J].珠江现代建设,2007(5):4-8.
[7]陈玲舫,陈子燊,黄强.珠江河口磨刀门水道的亚潮振荡特征及其对水文气象要素的响应[J].海洋通报,2014,33(2):126-131.
[8]陈丽棠,吕忠华.珠江河口治理[J].水利水电技术,2000(1):41-44.
[9]KLEME V.Operational testing of hydrological simulation models[J].Hydrological Sciences Journal,1986,31(1):13-24.
[10]WALDEN A T.Wavelet methods for time series analysis[M].Cambridge:Cambridge University Press,2000.
[11]JEFFREY L E.Finding structure in time[J].Cognitive Science,1990(14):179-211.