黄河下游利津断面年径流量预测研究 .pdfVIP

GA-SVM模型分别制

订了4种预报方案°对比发现，随着输入信息的增多，模型预测能力不断提升0在输入信息不变的情况下，通过结

合信号分解技术制定了方案四方案四下的6种模型较方案一在预测精度上有了明显提高其中以VMD-GA-SVM最

，

°

佳研究表明在不额外增加输入信息的情况下仅利用信号分解技术进行数据预处理便可有效提高模型预测精

，，

°

度可为水文时间序列的预测研究提供一种新的思路°

，

关键词年径流变分模态分解遗传算法支持向量机黄河流域

：；；；；

StudyonAnnualRunoffPredictionofLijinSectionintheLowerYellowRiver

DUYi,WANGDayang,LINYongen,WANGDagag

(SchoolofGeographyandPlanning,SunYat-senUniversity,Guangzhou510275,Guangdong,China)

Abstraci:BasedontheannualrunoffdataofLijinsectionofYellowRivesfrom1952to2018,fourschemeewithGA-SVM

moderarebuilttopredictannualrunoff.Theresult-showthat,withtheincreeseofinputinformation,thepredictioneffectof

themodelicimprovedccntinuously.Undertheccnditionofccnstantinputinformation,Scheme4icformulatedbyccmbining

signaldeccmpositiontechnolovy■ComparedwithScheme1,thepredictionaccuracyofthesiomodelsundeeScheme4is

significantlyimproved,andtheVMD-GA-SVMisthebest.Thestudyshowsthat,withoutadditionalinputinformation,only

usongsognaDdecomposotoonmethodooedatapeepeocessongcaneoectoeeyompeoeetheaccueacyooeunoopeedoctoon,andcan

provideanewideafoetimeseriespredictionreseerch.

KeyWordt:annualrunoff;variationalmodedeccmposition;geneticalgorithm;suppoevectoemachine;Yello