48278.大地電磁測深信號的高階譜估計及應用研究

1、中南大學碩士學位論文大地電磁測深信號的高階譜估計及應用研究姓名：王通申請學位級別：碩士專業(yè)：地球探測與信息技術(shù)指導教師：湯井田20070101ABSTRACTInthemagnetotelluricdataprocessingitiSoftendi佑cultbecausetheresponsefunctions(suchasapparentresistivityandphase)appeardispersionatindividualf

2、requenciesratherbigerrorbarandshapeabnormityetoAlthoughwecanusemuchmoresamplingandstackingaveragemeansinfeldtoimprovedataqualitybutcannotradicallySOlvetheproblemsAdditionallymanyhypothesisandlimitsWasintroducedonsignalan

3、dgeologymodelinpreviouskindsofdataprocessingmethodsbasedonpowerspectrumsuchassignalisGausssignal，noiseisGausswhitenoiseandteUuricsystemiSlinearnon—minimumphasesystemButactualsituationdoesnotlikethisMoreover’powerspectrum

4、doesnotsuppressnoiseverywellandlosesphaseinformationofsignalwhenprocessingdatabecauseoftheinherentcomputationalmodeofpowerspectrumStudyshowedthathigherorderstatisticsCanextractmanyinformationfromsignalthatcallnotbytwo—or

5、derstatistic(correlation)ItcannotonlyautomaticallysuppresstheinfluenceofGausscolornoise，butalsosuppresstheinfluenceofnonGausscolornoiseIthasmoreexcellencethanclassicalpowerspectrumestimationandcorrectionfunctionItiSsaidt

6、halitisworthresortingtohigherorderstatisticsmethodtoanalyzeandprocessanyproblemsthatcannotbesolvedsatisfactorilybyusingpowerspectrulnorcorrelationfunctionhlt11ispaperthroughtheUSeofMAlⅡ，ABasthenumericalplatformthestudyof

7、powerspectrumestimationformagnetotelluricbyusinghigherorderstatisticanalysisiSperformedFromreadingthetimeserieasfilesofEH4highfrequentcymagnetotelluricimagesysternthepowerspectumandcrosspowerspectrumwerereconstructedusin

8、ghigherorderstatisticsthepowersDectrulnfileisformedandtheresultgeneratedbyusingclassicalpowersogctrunlestimationmethodsuchasperiodogramiscomparedForthespecialpropertythatthehigherorderstatisties(higherordermomentsandhigh

9、erordercumulants)ofGaussianprocessiszero，sopowerspectrumestimationbasedonhigherordercumulantsCallsuppresscolorgaussnoisenaturelySotliSmethodCanbeusedtoimprovethequalityofhighfrequentcyNrrdataprocessingKeywords：magnetotel