SSA分解预测校正模型在年径流预报中的应用  

Research of Dedium-long Terms Runoff Predictor-corrector Model Based on SSA Decomposition

作　　者：张强 王本德[1] 何斌[1] 彭勇[1] 

机构地区：[1]大连理工大学建设工程学部,大连116024

出　　处：《武汉理工大学学报》2010年第7期152-155,共4页Journal of Wuhan University of Technology

基　　金：国家自然科学基金(50579095);国家"十一五"科技支撑计划课题(2007BAB28B01)

摘　　要：针对中长期水文预报因果规律不清楚,预报准确率低的问题,该文引入奇异谱分析方法(Singular SpectrumAnalysis,简称SSA),结合ARIMA模型,建立了基于SSA的分解预测校正模型。该模型通过SSA方法从年径流时间序列中提取对应着某些大气低频振荡的显著主分量序列,然后运用ARIMA模型对各显著分量序列分别进行预测,并对各序列预测结果的和进行校正。最后以大连市碧流河水库的年径流预报为例,对建立的SSA分解预测校正模型进行了应用检验。Due to the generation mechanism of Medium-long terms runoff is unclear,its forecasting accuracy is poor.In order to solve this problem,a predictorcorrector model based on Singular Spectrum Analysis（SSA） is proposed for the simulation and prediction of medium-long terms runoff time series.In this model, the runoff time series is decomposed into several principal components corresponding to certain atmospheric low-frequency oscillations.Then each principal component time serial is predicted respectively through the ARIMA model,and the correction is conducted for the sum of the prediction results.In the end,the proposed predictor-corrector model is examined by the forecast simulation of the annual runoff of Biliuhe reservoir.

关 键 词：中长期水文预报 奇异谱分析 ARIMA 校正 碧流河水库 

分 类 号：P338[天文地球—水文科学]