泾河年径流量BP神经网络模型研究  

Research on BP Neural Network Model for Runoff in Jinghe River Basin

作　　者：王峥[1] 马孝义[1] 吕静渭[2] 尹京川[1] 胡杰华[1] 

机构地区：[1]西北农林科技大学旱区农业水土工程教育部重点实验室,陕西杨凌712100 [2]甘肃省水利厅水利工程建设质量与安全管理中心,甘肃兰州730030

出　　处：《人民黄河》2012年第4期26-29,共4页Yellow River

基　　金：国家"863"计划项目(2011AA100509);"十二五"国家科技支撑计划项目(2011BAD25B03;2012BAD08B01)

摘　　要：以1957—2000年的实测降水序列和泾河年径流量序列为研究对象,利用EMD法和GA建立了泾河年径流量的BP神经网络模型。分析结果表明:泾河流域年降水量变化可能存在准2～3、5～7、10～13、18～22 a的周期;基于EMD的年径流量BP神经网络模型预测值的相对误差为-4.71%～8.21%,基于GA的年径流量BP神经网络模型预测值的相对误差为2.25%～12.22%。An annual precipitation-runoff forecasting BP neural network model based on EMD and GA was established,which using the rainfall data from 1957 to 2000 of 4 typical hydrologic stations in Jinghe River basin and the annual runoff as a research object.The results show that the precipitation series has periods that about 2-3,5-7,10-13 and 18-22 years.And using the BP neural network model based on EMD to predict annual runoff,the relative error of predictive value between 4.71% and 8.21%,while using BP neural network model based on genetic algorithm,the relative error of predictive value between 2.25% and 12.22%.

关 键 词：EMD GA BP神经网络 年径流量 泾河 

分 类 号：S161.9[农业科学—农业气象学]