水力发电学报2012，Vol.31Issue(6)：14-21,8.
基于修正组合模型的河川径流中长期预报
Mid-long term river runoff forecast based on modified combination model
摘要
Abstract
Grey model has strength in single variable forecasting and neural network shows advantage in nonlinear problem,but these two methods′ uncertainty in calculation of the weight coefficients of linear combination may deteriorate their forecasting results.This paper adopts Markov chain and the idea of combination to modify the error sequences forecasted by grey model,BP neural network and their combination model.The modified combination model of mid-long term river runoff forecast is brought forward through a comparison of its prediction accuracy to those of two modified single models.These methods are applied to forecasting of annual runoffs at the controlling hydrological stations on four tributaries of the middle Yellow River,i.e.Kuye River,Tuwei River,Wuding Rriver and Gushan River.The results show that the combination model with markov chain modification is better,more accurate,and more effective in runoff forecasting.关键词
水文学/修正组合预测模型/灰色神经网络/马尔科夫链/径流预报Key words
hydrology system/modified combination forecasting model/grey neural network/Markov chain/runoff forecast分类
建筑与水利引用本文复制引用
景亚平,张鑫..基于修正组合模型的河川径流中长期预报[J].水力发电学报,2012,31(6):14-21,8.基金项目
国家高技术研究发展计划（“863”计划）项目 （）
国家重大科技支撑计划项目 （）
西北农林科技大学博士科研启动基金项目 （）
西北农林科技大学科研专项 （）
