预览加载中,请您耐心等待几秒...
1/2
2/2

在线预览结束,喜欢就下载吧,查找使用更方便

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

基于混合网络U-SegNet的地震初至自动拾取 Title:AutomaticPickingofSeismicOnsetsusingU-SegNetbasedHybridNetwork Abstract: Theautomaticpickingofseismiconsetsisofparamountimportanceinvariousseismologicalapplications.ThispaperproposesanovelapproachtoautomaticallypickseismiconsetsusingahybridnetworkbasedontheU-SegNetarchitecture.TheU-SegNetcombinesthestrengthsoftheU-NetandSegNetarchitecturestoeffectivelyidentifyseismiconsetsandaccuratelypicktheirarrivaltimes.Theproposedmethoddemonstratespromisingresultsintermsofaccuracyandefficiencycomparedtoexistingtechniques. 1.Introduction Seismiconsetpickingisacriticaltaskinseismology,asitprovidescrucialinformationforearthquakeanalysis,eventdetection,andseismicmonitoring.Traditionally,seismiconsetsaremanuallypickedbyexpertseismologists,whichisatime-consumingandsubjectiveprocess.Consequently,thereisagrowinginterestindevelopingautomatedtechniquesforseismiconsetpicking.ThispaperpresentsahybridU-SegNetmodelforautomaticpickingofseismiconsets. 2.Methodology 2.1U-SegNetArchitecture TheU-SegNetarchitecturecombinestheU-NetandSegNetarchitectures.TheU-Netisaconvolutionalneuralnetwork(CNN)thatconsistsofanencoderanddecoder.Itiswidelyusedinimagesegmentationtasks.TheSegNetarchitectureisalsoaCNN-basedarchitecturethatperformspixel-levelclassification.IntegrationofbotharchitecturesintheU-SegNetallowsforbetterfeatureextractionandimprovedclassificationaccuracy. 2.2DataPreprocessing Theseismicwaveformdataispreprocessedtoremovenoiseandenhancethesignal-to-noiseratio.Variouspreprocessingtechniquessuchasbandpassfiltering,normalization,anddenoisingareappliedtoensurethequalityoftheinputdata. 2.3TrainingandTesting TotraintheU-SegNetmodel,labeledseismicdatasetsarerequired.Human-labeledseismiconsetsareusedasgroundtruthforthetrainingprocess.TheU-SegNetmodelistrainedusingacombinationofsupervisedandunsupervisedlearningtechniques.Themodelistrainedonalargedatasettooptimizeitsperformanceandgeneralizationability. 3.ExperimentalResults Toevaluatetheproposedapproach,avarietyofseismicdatase