熵权法并联组合建模在径流预测中的应用  

Entropy-based Parallel Combinational Model and Its Application in Runoff Prediction

作　　者：黎育红[1] 陈玥[1] 周建中[1] 姜福厚[1] 

机构地区：[1]华中科技大学水电与数字化工程学院,湖北武汉430074

出　　处：《水电能源科学》2012年第2期13-16,共4页Water Resources and Power

基　　金：中央高校基本科研业务费HUST基金资助项目(2010QN032);国家自然科学基金资助项目(10971225);教育部留学回国人员启动基金资助项目;华中科技大学高层次人才引进基金资助项目;华中科技大学科学研究基金联合资助基金资助项目

摘　　要：针对径流预报的具体特征和相关问题,建立了多元时变灰色预测模型,并基于信息熵原理,用熵权法将多元时变灰色预测模型、非时变的免疫神经网络模型、最小二乘支持向量机模型三者进行并联组合集成建模,以充分发挥多种模型各自优点且相互补充,并将此组合模型应用于新疆伊犁河雅马渡水文站的年径流预测。结果表明,与三个单项模型的预测结果相比,并联组合预测模型更合理、普适和可靠。In view of characteristics and related problem of runoff forecasting,a multi-attributes time-varying grey model for runoff prediction is established,which is proposed together with non-time-varying immune neural network model,least-square support vector machine model for a novel Entropy-based parallel combinational model based on the information entropy theory.The integrated parallel combination model is applied to predict the annual runoff of Yamadou hydrological station in Ili River in Xinjiang province in China.The results show that the combination prediction model is more rational,universal and reliable comparing with the original three individual models.

关 键 词：径流预报 信息熵 并联组合建模 GM(1 N) IEA-BP LS-SVM 

分 类 号：TV124[水利工程—水文学及水资源] P338.8[天文地球—水文科学]