Page 130 - 2023年第54卷第8期
P. 130
1355 - 1362.
[16] 李荣,李义天,曹志芳.河网水情预报的神经网络模型及应用[J].应用基础与工程科学学报,2000(2):
179 - 186.
[17] 朱承山,潘英杰.欧阳海水库径流量预报的模糊神经网络方法[J].水电能源科学,2000(3):16 - 18.
[18] ABADIM,BARHAM P,CHEN J,etal.TensorFlow:A system forlarge - scalemachinelearning[C]??Proc
th
ccdingsofthe12 USENIXConferenceonOperatingSystemsDesignandImplementation.2016.
[19] 谷歌推出 TensorFlow机器学习系统[J].电信工程技术与标准化,2015,28(11):92.
[20] ABADIM,AGARWALA,BARHAM P,etal.TensorFlow:Large - ScaleMachineLearningonHeterogeneous
DistributedSystems[J].arXiv,2016.arxiv:1603.04467v2.
[21] 符淙斌,王强.气候突变的定义和检验方法[J].大气科学,1992,16(4):482 - 493.
[22] 胡春宏,王延贵,张燕菁,等.中国江河水沙变化趋势与主要影响因素[J].水科学进展,2010,22(4):
524 - 532.
[23] 林海宁.应用 GoogleTensorflow的推荐系统[J].电脑知识与技术,2018,14(28):195 - 196.
[24] NELSONM M,ILLINGWORTH W T.APracticalGuidetoNeuralNets[M].MA:Addison - WesleyPublishing
Corp ,1991.
[25] 王攀杰,郭邵忠,侯明,等.激活函数的对比测试与分析[J].信息工程大学学报,2021,22(5):551 - 557.
[26] 卢金友,姚世明.水库群联合作用下长江中下游江湖关系响应机制[J].水利学报,2018,49(1):36 - 46.
[27] 方春明,胡春宏,陈绪坚.三峡水库运用对荆江三口分流及洞庭湖的影响[J].水利学报,2014,45(1):36 - 41.
[28] 曹文洪,毛继新.三峡水库运用对荆江河道及三口分流影响研究[J].水利水电技术,2015,46(6):67 -
71 ,78.
EvolutionlawanddrivingfactorsofthreeoutletsalongJingjiangRiverafter
operationofThreeGorgesReservoir
ZHAOWei,MAOJixin,GUANJianzhao,WANGDayu
(ChinaInstituteofWaterResourcesandHydropowerResearch,StateKeyLaboratoryofSimulationand
RegulationofWaterCycleinRiverBasin,Beijing 100048)
Abstract:ThethreeoutletsalongJingjiangriverarethecoreofthechangeofriver - lakerelationship,andthe
changeofdiversionlawwillaffecttheriver - lakerelationshipinthedownstream oftheThreeGorgesDam.There
fore ,itisnecessarytoquantitativelystudythechanginglawanddrivingfactorsofthethreeoutletsalongJingjiang
river.Basedonthemeasureddatafrom 1991to2020,throughregressionanalysisandartificialneuralnetwork
(ANN)modelbasedonTensorflow,thispaperexplorestheinfluenceofdrivingfactorssuchastheregulationand
storageoftheThreeGorgesReservoirandtheupstreamreservoirgroup ,andthemismatchscourbetweenthethree
channelsofJingnanoutletsandthemainstreamonthevariationlawofthethreeoutlets.Theresultsshowthatbe
foreandaftertheimpoundmentoftheThreeGorgesReservoirin2003 ,thediversionvolumeofthethreeoutletsa
longJingjiangriverhaschangedgreatly ,andtheaverageannualdischargefrom2003to2020isreducedbyabout
3
20% comparedwiththatfrom1991to2002 ,withadifferenceof12.44billionm .Furtherexplorethedrivingfac
torsoftheevolutionofthethreeoutletsdiversionsalongJingjiangriver ,theinfluenceofmismatchscourbetween
3
thethreechannelsofJingnanoutletsandthemainstreamisthemainfactor ,whichisabout7.57billionm ?a;The
3
secondistheinfluenceoftheThreeGorgesReservoirregulation ,about2.86billionm ?a;theinfluenceofrunoff
changeandstorageofcascadereservoirsintheupperreachesoftheThreeGorgesonthediversionofthethreeout
3
letsisabout2.01billionm ?a.AftertheimpoundmentoftheThreeGorgesReservoirin2003,theregulatingrunoff
andunequalerosionofthereservoirdominatedthedrivingfactorsofthediversioncapacityofthethreeoutlets.The
jointoperationoftheThreeGorgesReservoiranditsupstreamcascadereservoirsshouldbecontinuouslyoptimized ,
andmeasuressuchasdredgingandexcavationshouldbetakentoincreasethediversioncapacityofthethreeoutlets
tomaintaintherelationshipbetweenriversandlakes.
Keywords:thethreeoutletsalongJingjiangRiver;flowdiversion;artificialneuralnetwork;Tensorflow
(责任编辑:韩 昆)
— 1 0 4 —
1