Page 52 - 2025年第56卷第2期
P. 52
报(中英文),2023,21(1):10 - 22.
[18] 白玉川,黄本胜.河网非恒定流数值模拟的研究进展[J].水利学报,2000(12):43 - 47.
[19] ELIRN,WIGGERTJM,CONTRACTORDN.Reverseflowroutingbytheimplicitmethod[J].WaterResources
Research,1974,10(3),597 - 600.
[20] 王家彪,赵建世,雷晓辉,等.基于旋转 x - t平面的河渠水流反向演算[J].清华大学学报(自然科学版),
2020,60(10):855 - 863.
[21] 曹若馨,张可欣,曾维华,等.基于 BP神经网络的水环境承载力预警研究—以北运河为例[J].环境科学
学报,2021,41(5):2005 - 2017.
[22] 朱振国,田松禄.基于权值变化的 BP神经网络自适应学习率改进研究 [J].计算机系统应用,2018,27
(7):205 - 210.
[23] 程伟平,廖锡健,陈一帆,等.基于逆向概率密度函数的污染云团空间重构[J].四川大学学报(工程科学
版),2012,44(4):32 - 38.
[24] 王家彪,雷晓辉,廖卫,等.基于 耦 合 概 率 密 度 方 法 的 河 渠 突 发 水 污 染 溯 源 [J].水 利 学 报,2015,46
(11):1280 - 1289.
[25] 胡鹏,崔小红,周祖昊,等.流域水文模型中河道断面概化的原理和方法[J].水文,2010,30(5):38 -
41 ,79.
Methodologyforidentifyingthesourceofsuddenwaterpollutionin
typicalloopedrivernetwork
1
3
1
2
1
2
WANGJiabiao,ZHAOTongtiegang,CAISiyu,ZHAOJianshi,CHENXiaohong,WANGHao
(1.SchoolofCivilEngineering,SunYat - SenUniversity,Zhuhai 519082,China;
2.ChinaInstituteofWaterResourcesandHydropowerResearch,Beijing 100038,China;
3.TsinghuaUniversity,Beijing 100084,China)
Abstract:Withtherapidurbanizationprocess,theproblem ofsuddenwaterpollutioninChinaisbecomingin
creasinglyprominent.Afterpollutionoccurs,theprimarytaskistoidentifythesourceofpollutionforemergency
disposal.Nowadays,therearemanystudiesonthesuddenwaterpollutionsourceidentification,buttheyaregen
erallynotapplicabletotheloopedrivernetworkcases.Thereasonforthisissueisthattheproblem ofidentifying
pollutionsourceinaloopedrivernetworkisseverelyill - posed ,involvingcomplexcalculationsandbeingratherdif
ficulttosolve.Therefore,thisresearchproposesacreativemethodforidentifyingthesourceofsuddenwaterpollu
tioninaloopedrivernetwork,whichincludes:(1)thetaskofidentifyingpollutionsourceisdividedintotwo
steps:thefirststepinvolvesidentifyingtheriverwithapollutionsource,andthesecondstepinvolvesidentifying
detailedinformationaboutthepollution ;(2)atthefirststep,theBPneuralnetworkalgorithmisusedtocalculate
thespecificriverwhereapollutionsourceislocated ,andtheflowandwaterqualitymodelsarefurtherusedtocal
culatetheconcentrationofpollutantsattheoutletoftheidentifiedriverwhenobservationsarenotavailableinthis
river ;(3)atthesecondstep,aftertheriveridentified,thedetailedinformationofpollutionsourceincludingthe
releaseposition ,thereleasetime,andthereleasemass,canbefurtherestimatedusingthebackwardlocation
probabilitydensitymethod ;(4)theproposedmethodhasbeenappliedtoatypicalloopedrivernetworkcase,and
theresultsshowthatthismethodcaneffectivelyidentifythesourceofsuddenwaterpollutionfromaloopedrivernet
work.Thisresearchachievementcanservefortheemergencyresponse,responsibilitymanagement,andpre - e
mergencyplanningofsuddenwaterpollutionintherivernetworkareas ,withgreatpracticalsignificanceandbroad
applicationprospects.
Keywords:loopedrivernetwork; pollution sourceidentification; inverseproblem; sudden waterpollution;
mechanism - data
(责任编辑:耿庆斋)
— 1 9 —
2