摘要
针对丹江口流域秋汛期(9、10月)径流长期预报,为了消除网络输入的复共线性与网络训练的过拟合现象,将最优子集回归(OSR)和BP神经网络进行耦合,综合考虑训练误差和检验误差,来确定网络训练的最佳训练次数和终止条件,在此基础上提出基于OSR-BP神经网络的径流长期预报技术,并对丹江口秋汛期入库径流量进行了模拟和试报,结果表明:建立的模型稳定性良好,不论模拟还是试报精度均令人满意,特别是对预报年份中的丰枯特征均具有较好的体现。
This paper put forward a new method called OSR-BP neural network for long-term runoff forecasting. In order to eliminate input multi-collinearity and the phenomenon of over-fitting of the neural network, the optimal subset regression（OSR） and BP neural network was coupled to an integrated. Meanwhile, the training and testing error was comprehensively considered to determine the best training. On this basis, runoff in September and October in Danjiangkou Reservoir, was simulated from 1956 to 2000, and was predicted from 2001 to 2008 by using OSR-BP neural network. The result shows that the stability of model is good and accuracy is satisfactory whether simulation or prediction, especially for forecasting the characteristics of drought and flood years.
出处
《水文》
CSCD
北大核心
2010年第6期32-36,共5页
Journal of China Hydrology
基金
"十一五"国家科技支撑计划重点项目(2006BAB04A0702
2006BAB14B02)
水利部现代水利科技创新项目(XDS2007-04)
水利部公益性行业科研专项经费项目(201001002)
关键词
OSR-BP神经网络
径流长期预报
秋汛期
丹江口水库
OSR-BP neural network
long-term runoff forecasting
autumn flood season
Danjiangkou Reservoir
