Page 59 - 2022年第53卷第9期
P. 59
Discussionon“TheYellowRiverProtectionLawofthe
People ’sRepublicofChina(Draft)”
1
2,3
1
CHENGYeqi,PENGHao ,SHIXueqing
(1.ChinaAgriculturalUniversity,Beijing 100091,China;2.KeyLaboratoryofWaterCycleandRelatedLandSurfaceProcesses,
InstituteofGeographicSciencesandNaturalResourcesResearch,ChineseAcademyofSciences,Beijing 100101,China;
3.CollegeofResourcesandEnvironment,UniversityofChineseAcademyofSciences,Beijing 100049,China)
Abstract:Theexistingissuesanddifficultiesthatlegislationmustbeawareofinthegovernanceandpreservationof
theYellowRiveraresummarized.Basedonthis,theimperfectionsinthe“thelawofthePeople’sRepublicof
ChinaontheProtectionofYellowRiver(Draft)”areanalyzedandexplored,andthelegislativesuggestionsisex
plainedaccordingtotherelevantprovisions.Forinstance,thefoundationofecologicalprotectionandrestorationin
theYellowRiverBasinistopromotethesustainabledevelopmentoftheYellowRiverBasinthroughscientificplan
ningandprudentmanagement.Toachievehigh - qualitydevelopment ,itisvitaltoinsistonbothgovernanceand
preservationfortheYellowRiver ,auniqueriverthatiscomplicatedanddifficulttogovern.Inordertomaintain
theauthorityoflawenforcement,thereshouldberoomforthewordingofarticlesthatneedtoimprovethetechnical
levelandaredifficulttouseasthebasisforlawenforcement.Legislationshouldpayattentiontothereformofgov
ernancesystem.Oncethedepartmentsresponsibleforpolicyformulation ,projectapprovalandlawenforcementare
relativelydetachedandnolongerdirectlyparticipateinrelevantenterprisebehaviors,thesocialinfluencecansig
nificantlyimprovethegovernanceefficiencyoftheYellowRiver.Fromtheperspectiveofensuringthesafetyofthe
peopleinthefloodplain ,itisneitherrealisticnornecessarytorelocatealltheresidentsinthefloodplainarea.
Keywords:theYellowRiver;legislation;protection;harnessing;management;high - qualitydevelopment
(责任编辑:李福田)
(上接第 1063页)
Multimodalperception - drivenhighrockfilldam constructionsimulation
inputmodelingusinganensembledeeplearningmodel
1
1
2
2
1
1
ZHANGJun,WANGJinguo,YUJia,ZHAOHao,ZHANGDongming,WANGXiaoling
(1.StateKeyLaboratoryofHydraulicEngineeringSimulationandSafety,TianjinUniversity,Tianjin 300072,China;
2.YalongRiverHydropowerDevelopmentCompany,LTD,Chengdu 610000,China)
Abstract:Thereal - timeandaccuratemodelingofconstructionsimulationparametersisthekeytoensuretheac
curacyofconstructionsimulation.Theexistingconstructionsimulationparameterupdatemethodsmainlyrelyon
singlemodaldatawhiletheexistingmulti - modaldataacquisitionprocesshasaproblem oflagging ,resultingin
insufficientreal - timeandaccuracyofsimulation.Tosolvethoseproblems,thispaperproposesanensembledeep
learningmodelforhighrockfilldam constructionsimulationinputmodelingdrivenbymultimodalperception.
First ,acloudplatformforreal - timeacquisitionofkinematicsandacousticdatabasedonmobilesmartphonesen
sorswasdevelopedundertheSpringBootframework ,relyingonthis,themulti - modaldataofrockfilldam con
structionmachineryiscollectedinreal - timeandpre - processedusinglow - passfiltersandmelspectrograms.Sec
ondly ,anensembledeeplearningmodelforrockfilldamconstructionmachineryactivityrecognitionisproposedto
automaticallyextractmultimodaldatafeaturesandaccuratelyidentifymechanicalfineactivity.Themodelin
tegratesanimproveddeepconvolutionlongshort - termmemoryneuralnetwork(ImprovedDeepConvLSTM,IDeep
ConvLSTM)accuratelyperceivesthemotiondirectionofconstructionmachineryandtheadvantagesofdeepconvo
lutionalneuralnetworktoperceivethevibrationstateofconstructionmachineryfromsoundmodes.Amongthem ,
IDeepConvLSTM addsabatchnormalizationlayerinthemiddleoftheconvolutionlayertoimproveconvergence
speed,andgradientscalingandclippingareusedtoavoidtheproblem ofgradientexplosion;further,alarge
windowmovingfilterisusedtoprocesstherecognitionresultsofmechanicalactivityonline ,andreal - timemodel
ingofrockfilldam constructionsimulationparametersisrealized.Theengineeringcaseshowsthat,compared
withthesinglekinematicoracousticmodemechanicalactivityrecognitionmethod ,therecognitionaccuracyofthe
proposedmethodisimprovedby9.22% and23.62%,respectively.Theresearchresultsprovidenewideasand
technicalmeansforimprovingtheaccuracyandreal - timeperformanceofrockfilldam constructionsimulation ,
whichhavesignificantapplicationandpopularizationvalue.
Keywords:high rockfilldam; real -time modeling ofconstruction simulation parameters; multimodal
perception ;ensembledeeplearning;mobilesmartsensors
(责任编辑:李福田)
— 1 0 2 —
7