Page 111 - 2024年第55卷第8期
P. 111
interlayerlocationsandthicknesses.Theresultsshowthattheparticlesizehasasignificanteffectonthemigration -
depositionprocessoffineparticlesinsandcolumnswithinterlayers.Whentheparticlesizeratioislargerthan1 ,
thepeakvalueofthesuspendedparticlespenetrationcurveishigher,andtheamountoffineparticlesdepositedin
theinterlayerincreasesabruptly;theclosertheinterlayerpositionistothewaterinjectionend,thesmallerthe
thicknessis ,andthelargerthepeakvalueandthesuddenincreaseindepositionare.Whentheparticlesizeratio
islessthan1,thepeakvalueofthesuspendedparticlespenetrationcurveislower,andthereisnoparticleenrich
mentattheinterlayer;theclosertheinterlayeristothewateroutletend,thegreaterthethickness,andthehigher
thepeakvalueofthepenetrationcurve.Duringthetest,themigrationanddepositionofsuspendedparticleswill
causechangesinthepermeabilitycoefficient.Theprocessofpermeabilitycoefficientchangecorrespondstothe
changeofpenetrationcurve ,andtheprocessisdividedintothreekinds:deposition,strippingmigrationanddy
namicequilibrium.Fineparticlesaredepositedinsandysoilbyadsorptionandaccumulateattheadsorption.
Keywords:interlayer;particlesizeratio;migration;deposition;permeabilitycoefficient
(责任编辑:李 娜)
(上接第 976页)
Researchonmulti - outputseismictime - historylong - term sequencesprediction
modelforfreefieldofdam basedonvirtualsensors
1
1
2
1
1
3
SUZhe,LIUZongxian,YUHongling,TONGDawei,YUJia,WANGXiaoling
(1.StateKeyLaboratoryofHydraulicEngineeringIntelligentConstructionandOperation,TianjinUniversity,Tianjin 300072,China;
2.YalongRiverHydropowerDevelopmentCompany,Ltd,Chengdu 610051,China;
3.CollegeofWaterResourcesandCivilEngineering,ChinaAgriculturalUniversity,Beijing 100083,China)
Abstract:Themultidimensionallong - termpredictionofseismictime - historyindam areasholdssignificantim
portanceforrapiddamageanalysis.Virtualsensors ,ascomplementarysensingmechanismstoseismicphysical
sensors ,facilitateseismictime - historypredictions.However,existingvirtualsensorsfacechallengesineffec
tivelypredictinglong - term sequencesformultiplesignals ,leadingtodelaysinanalyzingdam seismicdamage.
Addressingtheaforementionedissue ,amulti - outputseismictime - historylong - termsequencespredictionmodel
basedonTFA - Seq2Seqvirtualsensorsisproposed.ThismodelenhancestheSeq2Seqvirtualsensorsusingmulti -
tasklearning ,restructuringthemintoan“Encoder - 3Decoder”architecture.Thisstructureestablishesthemap
pingrelationshipbetweenmultipledamphysicalsensorsignalsandlong - termseismictime - historyinthreefree -
fielddirections.Additionally ,anattentionmechanismisintegratedtocapturetemporaldependenciesamongmul
tipleinputsignals ,resolvingsynchronousmulti - outputpredictionissuesandenhancingpredictionaccuracy.Fur
thermore ,Time - Frequencytransform(TF)layersandtheirinversetransformationlayersareintroducedtoimprove
theEncoderandDecoder ,shorteningthetemporallengthofseismicsignalsandextractingfrequencydomainfea
tures.Correspondingly ,amodeltrainingstrategyinvolvingstochasticforcedlearningisproposedtoovercomethe
limitationsofexistingvirtualsensorsineffectivelypredictinglongsequences.Casestudiesdemonstratethatthe
proposedmethodachievesavirtualsenseof10secondsaheadforseismicsignalsinthreedirectionswithindam
freefield.Comparedtomodelswithoutattentionmechanismsandsingleoutputs ,theproposedmethodexhibitsan
enhancedpredictionaccuracyof6.88% and3.32%,respectively.Thisresearchpresentsnovelinsightsandap
proachesforadvancingtheanticipatorysenseofseismicinformationduringseismicevents.
Keywords:freefieldseismic;virtualsensors;multi - outputlong - term sequencesprediction;TFA - Seq2Seq;
multi - tasklearning
(责任编辑:韩 昆)
— 9 8 9 —