資源描述：
《灘壩砂儲層預(yù)測方法分析與應(yīng)用》由會員上傳分享，免費在線閱讀，更多相關(guān)內(nèi)容在學(xué)術(shù)論文-天天文庫。
1、ResearchofGeologicalPredictiononBeach-BarSandstoneReservoirZhangBao·zhen(GeophysicalProspectingandInformationTechnology)DirectedbyAssociate··ProfessorYangGuo-·quanAbstractNowitisahotspottoresearchthin-layerorthininterbed．Firstly,wehaveanalyzedthecharacteristicsintimedomainandfrequencydo
2、mainofthereflectioncoefficientofthin·layerintheory,andtriedtoanalysisnewmethodstodescribethesubsurfacegeologicstructuremuchmoretrulyandmeticulousofbeach—barsandstonereservoirinfrequenciesdomain．Thebasicideaoftime—frequencyanalysisistodesignthejointfunctionoftimeandfrequency，whichisusedt
3、odescribethedifferenttimesandfrequencyoftheenergydistributionthatcanmakethelocalcharacteristicsofthetime··frequencytwo--dimensionalplaneclearlydisplayed．ThisthesisstartsfromtheFouriertransformtoanalyzetheadvantagesandlimitations；thenwithallexampleofashort-timeFouriertransform，wavelettra
4、nsformandWigner-Villedistribution，theyaremoreclassicmethodsoftime-frequencyanalysis，buttherearesomeproblemsinpracticalapplication．Therecentemergedadaptivetime-frequencyanalysis，whichseparatesthesignalwithouttheconstraintsofHeisenberguncertaintyprinciple，haveagoodtimeandfrequencyresoluti
5、on，andhighnoiseimmunity．Hilbert-Huangtransformhasendsflyingwingphenomenon．Inordertosolvethisproblem，severalapproachestorestrainingtheendeffectwereproposed，andallhaveacertaindegreeofrestrainteffectinanalyingseismicdata．Matchingpursuitisanothermethodofadaptivetime—frequencyanalysis．Inthis
6、thesis，wefocusontheresearchofimprovingthecomputationalefficiencyofMPbyusingparticleswarmalgorithmandmultichannelmatchingpursuit．Throughmodelanalysis，theimprovedadaptivetime-frequencyanalysishavegoodtime—ffquencyperformanceandcarlaccuratelyshowthelocalcharacteristicsofthesignal．Basedonth
7、eadaptivetime—frequencyanalysis，wehaveanalyedseveralseismicpredictionmethods，suchasdividingsedimentarycycles，multiwaveletstheoryandspectraldecomposition．Allofthemethodshavebeenusedinbeach—barsandstone，withcomparisonllofmultipleattributionclusteringanalyses．Theresultsindicatedth
