华东师范大学学报(自然科学版) ›› 2003, Vol. 2003 ›› Issue (2): 62-67.

• 论文 • 上一篇    下一篇

长江口河水氯离子浓度指标值的建模与预报

潘仁良1; 王新伟2; 茅志昌3;   

  1. 1 华东师范大学 数学系, 上海 200062; 2 华东师范大学 计算机系, 上海 200062; 3 华东师范大学 河口海岸研究所, 上海 200062
  • 收稿日期:2001-06-03 修回日期:2001-12-03 出版日期:2003-04-15 发布日期:2003-04-15

The Model and Prediction for the Density of Chlorine Ion in the Waters of the Yangtze River Estuary

PAN Ren-liang1; WANG Xin-wei2; MAO Zhi-chang3   

  1. 1 Department of Mathematics, East China Normal University, Shanghai 200062,China 2 Department of Computer, East China Normal University, Shanghai 200062,China 3 Institute of Estuarine and Coastal Research, East China Normal University, Shanghai 200062,China
  • Received:2001-06-03 Revised:2001-12-03 Online:2003-04-15 Published:2003-04-15

摘要: 长江口宽度大叉道多,影响潮流运动的因素错综复杂,且各种变化量的实测资料有限,面对如此一个复杂的非线性随时间变化且实测资料有限的系统,利用传统的方法分析研究系统的建模、预报有一定的强求和困难。作者尝试利用在八十年代后期得到飞速发展的人工神经网络现代技术分析研究长江口河水氯离子浓度指标值的动态变化规律。所采用的模型集是多层映射的BP神经网络,利用长江口水文观察站的水文观察资料,采用误差反向传播学习方法来调整BP网络参数连接权值和节点阀值。经过训练学习后的BP神经网络输出与量测输出是非常接近的。并考虑三峡水库建成后有不同的调水方案,用得到的BP神经网络模型分别来与博爱长江口河水日平均氯离子浓度值的变化情况,所得结果供上海有关部门自来水公司做决策参考。

关键词: 氯离子浓度, BP神经网络, 反向传播学习训练, 氯离子浓度, BP神经网络, 反向传播学习训练

Abstract: In this paper, the author use modern technology of the artificial neural network to analyse the law of dynamic change of the Chlorine ion in the water of the Yangtze river estuary. With the consider of different controlling water quantity project after the completion of the construction of the reservoir on Three Gorge Upward the Yangtze River, a BP neural network model is used for forecasting the denisity of Chlorine ion in Gaoqiao observation station. The results will be referred to related Shanghai department and water supply corporation for decision-making.

Key words: BP neural network, back-propagation training, the denisity of Chlorine ion, BP neural network, back-propagation training