基于CNN-LSTM融合深度神经网络的地震速度反演（英文）-豆柴文库

您所在位置：网站首页 / 基于CNN-LSTM融合深度神经网络的地震速度反演（英文）.docx / 文档详情

在线预览结束，喜欢就下载吧，查找使用更方便

5 金币

下载文档

/ 2

下载提示

如果您无法下载资料，请参考说明：

1、部分资料下载需要金币，请确保您的账户上有足够的金币

2、已购买过的文档，再次下载不重复扣费

3、资料包下载后请先用软件解压，在使用对应软件打开

文本预览

基于CNN-LSTM融合深度神经网络的地震速度反演（英文）
Title:SeismicVelocityInversionbasedonCNN-LSTMIntegratedDeepNeuralNetwork
Abstract:
Inrecentyears,theapplicationofdeeplearningtechniquesinthefieldofgeophysicshasshowngreatpotentialinsolvingcomplexproblems,suchasseismicvelocityinversion.Seismicvelocityinversionisacriticaltaskingeophysicalexploration,asitprovidesvaluableinformationaboutsubsurfacestructuresandhelpsinthecharacterizationofhydrocarbonreservoirs.Inthispaper,weproposeanovelapproachforseismicvelocityinversionusingaCNN-LSTMintegrateddeepneuralnetwork.Thisapproachcombinesthestrengthsofbothconvolutionalneuralnetworks(CNNs)andlongshort-termmemory(LSTM)networkstoeffectivelycapturespatialandtemporalfeaturesinseismicdata,leadingtoimprovedinversionresults.
Introduction:
Seismicvelocityinversionistheprocessofestimatingsubsurfacevelocitiesbasedonseismicdata.Itplaysacrucialroleinreservoircharacterization,facilitatingtheidentificationofpotentialhydrocarbonreservesandunderstandingthegeologicalstructures.Traditionalseismicvelocityinversionmethodsoftenrelyoniterativealgorithmsandrequireextensivemanualintervention,makingthemtime-consumingandsubjective.Inrecentyears,deeplearningtechniqueshavegainedattentioninthegeophysicalcommunityfortheirabilitytoautomaticallylearncomplexrepresentationsfromrawdata.Convolutionalneuralnetworks(CNNs)andrecurrentneuralnetworks(RNNs)aretwopopulardeeplearningarchitecturesthathavebeensuccessfullyappliedinvariousfields.Inthiswork,weproposeaCNN-LSTMintegrateddeepneuralnetworkforseismicvelocityinversiontoachieveaccurateandefficientresults.
Methodology:
TheproposedCNN-LSTMintegrateddeepneuralnetworkconsistsoftwomajorcomponents:theCNNmoduleandtheLSTMmodule.TheCNNmoduleisresponsibleforcapturingspatialfeaturesinseismicdatabyperforming2Dconvolutionsoninputseismicimages.Thishelpsinidentifyingkeypatternsandstructuresthatareindicativeofsubsurfacevelocities.TheLSTMmodule,ontheotherhand,capturesthetemporaldependenciesintheseismicdatabyprocessingthesequenceoffeaturemapsobtainedfromt