基于小波的NN-GARCH模型及其應用研究.pdf

1、獨創(chuàng)性聲明本人所呈交的學位論文是在導師指導下進行的研究I作及取得的成果。盡我所知，除特別加以標注的地方外，論文中不包含其他人的研究成果。與我一同工作的同志對本文的研究工作和成果的任何貢獻均已在論文中作了明確的說明并已致謝。本論文及其相關(guān)資料若有不實之處，由本人承擔一切相關(guān)責任。論文作者簽名逍訓年歲月瑚日學位論文使用授權(quán)本人作為學位論文作者了解并愿意遵守學校有關(guān)保留、使用學位論文的規(guī)定，即：在導師的指導下創(chuàng)作完成的學位論文的知識產(chǎn)權(quán)歸西安

2、理工大學所有，本人今后在使用或發(fā)表該論文涉及的研究內(nèi)容時，會注明西安理工大學。西安理工大學擁有學位論文的如下使用權(quán)，包括：學?？梢员４鎸W位論文；可以采用影印、縮印或其他復制手段保存論文；可以查閱或借閱。本人授權(quán)西安理工大學對學位論文全部內(nèi)容編入公開的數(shù)據(jù)庫進行檢索。本學位論文全部或部分內(nèi)容的公布(包括刊登)授權(quán)西安理工大學研究生學院辦理。經(jīng)過學校保密辦公室確定密級的涉密學位論文，按照相關(guān)保密規(guī)定執(zhí)行；需要進行技術(shù)保密的學位論文，按照《西

3、安理工大學學位論文技術(shù)保密申請表》內(nèi)容進行保密(附《西安理工大學學位論文技術(shù)保密申請表》)。保密的學位論文在解密后，適用本授權(quán)。做作者簽名』登翩魏莖垂i塑訓年弓月以日AbstractTitle：THENNGARCHMODELBASEDONⅥfAVELETANDITSAPPLICATlONMajor：AppliedMathematicsName：RuoxingWANGSupervisor：ProfDeshengZHANGAbstractS

4、ignature：＆竺∑!垃巡吣qSignature：一ThisthesismainlystudiedtheNNGARCHmodelbasedonwavelettheoryanditsapplicationinpredictionofTheShanghaiStockIndexThemodelcombineswavelettheoryandneuralnetworkinfinancialtimeseriessimultaneouslybo

5、ththewaveletbasedenoisingeffectsandthestrongfittingnonlinearfunctionofneuralnetwork，alsoconsiderthefinancialtimeseriesheteroskedasticitycharacteristics，thengetsmoreaccuratefittingforecastmodelThecontentsofthethesisareasf

6、ollows：1Thewaveletanalysisintimesequenceofapplicationwerediscussed，duetothewaveletanalysishasthenatureofgoodtime—frequencydomainpropertiesand”adaptive”and”mathematicsmicroscope”，consequentlybecomecommonfocus，alsoitbecome

7、saveryusefulmathematicaltoolstotheanalysisofstockmarketSecondlydiscussedtheapplicationofneuralnetworkintimesequence，becausethestrongfittingfunctionoftheneuralnetworkhasplasticityselforganizing，adaptandverypowerfultononli

8、nearSOitalsohaswideapplicationinthefinancialtimeseriesanalysis2FirststudiedthemodelingmethodandmodelingprocessofARNNmodelbasedonwaveletanalysisHereusingbottom—upmodelingmethods，beginningthesimplestARmodel，againconsummate

9、sthemodelunceasingly；Then，appliedthemodeltotheShanghaiIndexofpredictionresearch；FinallycomparedthemodelwithARNNthatdidnotaddwaveletanalysisandsimpleARmodelTheresultshowsthatcomparednowaveletandthesimplestARmodel，theNix／A

10、Rmodelwhichbasedonwaveletanalysishasmoreapproximatingandforecastinghigherprecision3AfterthediscussionoftheAR—NNmodelbasedonwaveletanalysis，wefurtherconsideredtheresidualheteroskedasticityofthesequence，thenestimatedtheNNG

11、ARCHmodelbasedonwaveletanalysisThemodelcombineswavelettheoryandneuralnetworkwiththeGARCHmodelFirstlydealwiththenoiseofdatabythewaveletanalysis；then，estimatedthemodelwhichcontainsnonlinearneuralnetworkparametersinmeanequa