基于未确知盲数的BP模型及其在年径流预测中的应用  

BP Neural Network Model based on Unascertained Blind Number and Its Application to Annual Runoff Prediction

作　　者：黄钰凯[1] 王文圣[1,2] 李跃清[2] 

机构地区：[1]四川大学水利水电学院,四川成都610065 [2]中国气象局成都高原气象研究所,四川成都610072

出　　处：《水电能源科学》2011年第8期1-3,34,共4页Water Resources and Power

基　　金：科技基础性工作专项基金资助项目(2009IM020100);教育部新世纪优秀人才支持计划基金资助项目(2009SCU-NCET-09-0573)

摘　　要：针对未确知盲数可展现信息真实的数量及状态,将未确知盲数与BP网络耦合,提出了以未确知盲数为网络输入、预测对象为输出的BP模型,模型预测结果为具有置信度与置信区间的盲数,并将基于未确知盲数的BP模型用于年径流预测。结果表明,该模型有效,预测结果可供借鉴。The unascertained blind number can denote the real number and state of the information. By coupling unas- certained blind number and BP network, this paper proposes the BP model based on the unascertained blind number, which takes the unascertained blind number as model input and the forecasted object as output. The prediction results are the blind number with confidence interval and confidence level. The proposed model is used to forecast annual runoff. The results show that the model is effective and the forecasting results can be referenced.

关 键 词：未确知盲数 BP模型 耦合 年径流预测 

分 类 号：TV212[水利工程—水文学及水资源] P338.2[天文地球—水文科学]