基于CCGA―BP凌情预报模型研究.docVIP

基于CCGA―BP凌情预报模型研究 　　摘要：凌汛是河道里的冰凌对水流的阻力作用而引起的一种涨水现象，它直接影响到水利工程的运行和维护、水利发电、冬季输水等问题，较高精度的凌情预报成果可为防凌减灾提供重要决策信息。针对遗传算法优化神经网络时不能同步优化网络结构和连接权值且计算量大的问题，采取分而治之的思路构建了基于合作型协同进化遗传算法的BP神经网络（CCGA-BP）凌情预报模型。运用MATLAB2009b进行建模，并将其应用于宁蒙段巴彦高勒封河日期的预报，结果表明，CCGA-BP模型历史样本拟合合格率达到89.3%（25/28），预报检验合格率为100%（5/5）。预报模型可以用于黄河宁蒙段冰凌预报实践。 　　关键词：凌汛；凌情预报；封开河日期；CCGA-BP；宁蒙河段 　　中图分类号：P338文献标志码：A文章编号： 　　1672-1683（2015）001-0207-04 　　Study on the ice condition forecasting model based on CCGA-BP 　　LI Fu-chuan1 ，WANG Fu-qiang2，PEI Ying-xin3，LI Zong-lin1 　　（1.South-to-North Water Diversion Project eastern shandong co.，LTD ，JiNan 250013 China； 　　2.North China University of Water Conservancy and Hydroelectric Power，Zhengzhou 450045 China； 　　3.Material company？of Liaohe Petroleum Exploration Bureau，Panjin 124000 China） 　　Abstract：Ice flood is a freshet phenomenon caused by the blocking effect of ice in the river channel，it will directly influence the operation and maintenance of water conservancy project，hydroelectric generation and water delivery in winter，etc.High-accuracy ice forecasting achievement can provide important decision information to ice flood control and disaster mitigation.To counter the problem that genetic algorithm cannot optimize network topology as well as connection weights and has a large amount of calculation when optimizing neural network，this paper，based on the co-evolutionary genetic algorithm，first take the policy of divide and rule and set up the BP neural network（CCGA-BP） ice forecasting model.Then use MATLAB 2009 b and applied the model to the forecasting of ice-locked date in Bayangol，Ningxia-Inner Mongolia reach.The results show that the historical sample fitting rates reach 893%（25/28），and the prediction test passing rate is 100%（5/5）.Thus the forecasting model can be put into practice in Ningxia-Inner Mongolia reach of Yellow River. 　　Key words：ice run；ice condition forecasting；ice-locked and break up date；CCGA-BP；Ningxia-Inner Mongolia reach 　　黄河以冰凌灾害频繁而著名，每年封开河都要