Page 5 - 《水资源与水工程学报》2022年第5期
P. 5
! 33 " ! 5 # & ' ( ) & * + , - Vol.33No.5
2022 $ 10 % JournalofWaterResources&WaterEngineering Oct.,2022
DOI:10.11705/j.issn.1672-643X.2022.05.01
'1234 - 56789:;<=>?@AB
1 2 3 4
! " , # $ , %&' , ( )
(1. JKUVP& FG%&WUX&Y, JK 400067;2. Z[\%P& NO]^W_`ab8cdJefgh,
Z[ Z2 550025;3. 1ijk%lP& mnFG&Y, o1 pq 712100;
4. JKUVP& rstumn>?vw, JK 400067)
C D: xyzf{|C}~v
, ', -
{|C}~@¡。¢£G¤¥¦§{|C}¨© ª«¬, ®¯°±²³´µ¶·¸¹º
»¼½; $¾¿ÀÁÂúÄÅÆ, Ç®ÈÉ¿
ÊËÌÍÎ, fz{|C} ÌÍÏÐÑG; °
¾yÒÓs{|JKÔ 3 ÕC}aÖ×ØÂÃfÙST, gÚÛ@¡ Üݳ
Þ³。ÉßÎà: Û@¡á
¶âÂÃ{|C}~ÑG, ãäÈå@¡( 、 æç - 、 - èé), ÈêÜ
°ë ÞÍ(>0.97)、 ìÍ(>0.86)、 íîÍ(>0.86)
F1 ï(>0.86), B{|CNOFGðÜÝ l
ñòó。
EFG: C}~; G¤; ; ²³´µ; ³
HIJKL:X824 MNOPQ:A MRSL:1672643X(2022)05000110
Watershedwaterqualityassessmentmethodbasedon
roughsetandBayesiannetwork
1
3
4
2
BAIYun,LIYong,ZHANGWanjuan,HEJia
(1.SchoolofManagementScienceandEngineering,ChongqingTechnologyandBusinessUniversity,Chongqing400067,China;
2.KeyLaboratoryofEnvironmentalPollutionMonitoringandDiseaseControl,MinistryofEducation,GuizhouMedicalUniversity,
Guiyang550025,China;3.CollegeofEconomics&Management,NorthwestA&FUniversity,Yangling712100,China;4.Research
CenterforEconomyofUpperReachesoftheYangtzeRiver,ChongqingTechnologyandBusinessUniversity,Chongqing400067,China)
Abstract:Accordingtotheintelligentcomplementarystrategy,anewwatershedwaterqualityassessment
methodbasedonroughset(RS)andBayesiannetwork(BN)waspresentedforthewaterqualityassess
mentcontainingincompleteanduncertaininformation.Firstly ,RSwasusedtoextractthemainfactorsaf
fectingwatershedwaterquality ,soastoobtaintheminimumattributereductionset,whichcanbeusedto
reducethemodellingcomplexity.Then ,theBNwasconstructedandtrainedbasedontheattributereduc
tionset,anditsnetworkstructureandconditionalprobabilitytablewereobtainedtorealizetheprobabilistic
decisionreasoningofwatershedwaterquality.Finally,themodelevaluationindexeswereusedtoanalyse
threewaterqualitymonitoringsectionsintheChongqingsectionoftheJialingRivertoverifythecorrectness
andeffectivenessofthismethod.Theresultsshowthatthismethodisapplicabletothewatershedwater
qualityassessment ,andhasthehighestaccuracy(>0.97),precision(>0.86),recall(>0.86),and
F1-measure(>0.86)comparedwithothermethods(BN,GRA-BN,RS-NB).Thismethodcan
provideaneffectivetechnicalsupportforthewaterenvironmentmanagementinthewatershed.
Keywords:waterqualityassessment;roughset(RS);Bayesiannetwork(BN);attributereduction;
uncertainty
!"#$:20211026; %$:20220426
'()*: !"#$%&'()*(71801044); !"+,%&'()*(18CJY005)
+,-.: - .(1985-), /, 01234, 56, 789, :6;<=, >?@ABCDEFG。
/0+,: H I(1991-), /, JKLM4, 56, >?@ABNOPQRST。