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
   54   55   56   57   58   59   60   61   62   63   64