集对分析聚类预测法在区域用水量中的应用  

Application of Set Pair Analysis Classified Prediction Method in Regional Water Consumption Prediction

作　　者：袁朝阳[1] 吴成国[1] 张礼兵[1] 潘争伟[2] 

机构地区：[1]合肥工业大学土木与水利工程学院,安徽合肥230009 [2]安徽新华学院土木与环境工程学院,安徽合肥230088

出　　处：《华北水利水电大学学报（自然科学版）》2015年第4期32-35,共4页Journal of North China University of Water Resources and Electric Power：Natural Science Edition

基　　金：国家自然科学基金项目(51309072;51309004;51479045);水利部公益性行业专项经费项目(201301003)

摘　　要：精确预测用水量有利于水资源的规划和管理.本文利用集对分析联系度及聚类思想建立了集对分析聚类预测模型,并应用于山东省用水量预测中.结果表明,山东省2010年用水总量预测的计算值与实际值相对误差为0.67%,采用灰色GM(1,1)模型预测的相对误差为4.95%,采用BP神经网络预测的误差为4.77%.进一步对山东省2011—2013年用水量的年增长率进行预测,相对误差较小.可见,集对分析聚类预测模型精度较高,可用于区域产业用水量的预测研究中.An accurate prediction of water consumption is conducive to the planning and management of water resources. Based on the connection degrees of set pair analysis and clustering thought, a model of set pair analysis classified prediction is established ,and is applied into water demand forecast in Shandong Province. The results show that The total amount of water in Shandong Province in 2010 forecast The relative error between the calculated and predicted value and the actual value of the total water consumption of Shandong Province in 2010 is 0.67% , while the relative error is 4.95% and 4.77% by the prediction method of GM （1,1）and BP neural network, respectively. Furthermore, the growth rates of water consumption from 2011 to 2013 are predicted, and the relative error is also small. Thus, the set pair analysis classified prediction model has high forecast accuracy and can be used to predict for the regional industrial water consumption.

关 键 词：用水量预测 集对分析 聚类预测 联系度 山东省 

分 类 号：TV213.4[水利工程—水文学及水资源]