Page 54 - 《水资源与水工程学报》2022年第4期
P. 54
!33 " ! 4 # & ' ( ) & * + , - Vol.33No.4
2022 $ 8 % JournalofWaterResources&WaterEngineering Aug.,2022
DOI:10.11705/j.issn.1672-643X.2022.04.07
'IJKdLØ BP M»NO
PQRYWRS
1,2,3 1,2,3 1,2,3 1,2,3 1,2,3
\]2 , L^_ , # ` , % a , Vbc
(1. §D#@wHjxf. ¯dÝ9Ä@GW@H+, §D e, 250014;2. §D#@A9NO,-M9/GW
%ð45^U, §D e, 250014;3. §D#@wHjxf. @A9KLMNORSTUV, §D e, 250014)
A B: WP6à±Hz±q¤Ù、 Ä«?Çfs't`çR§¤ÙKL, e6à±@{eÚÌf4ãÕ
R§BoT。8Y1/(rà±zgij«½, 8à±zKL,-j, ï¯ 2016-2018 c²&z
z<Há9b、 à±9xmb、 .Gi¡Lbs¢, ¨ö:Y 6 t BP ñ{78*m«²ß§«
BP ñ{78(rB0, wxYR½(rB0e²&z9Ð(r3。45: M BP ñ{78¼Á, Ñ GA
ßþÅ?{T8 BP ñ{78@¶ï2@¥ñ{78(r;ç, ½¶HHbµñ{
78²Ô, ̪¿#Hb$Â; m Levenberg-Marquardt }~6ß GA-BP(LM) 78B0æã;
ç¥、 $¢、 (rµHî, ïà±z9Ð(r。
CDE: à±z9Ð(r; «²ß;BP ñ{78; }~6ß; ²&z
FGHIJ:P641.134 KLMNO:A KPQJ:1672643X(2022)04005008
PredictionofkarstspringwaterlevelbasedonBPneural
networkoptimizedbygeneticalgorithm
1,2,3
1,2,3
1,2,3
1,2,3
CHENHuanliang ,LIChangsuo ,GAOShuai ,SUNBin 1,2,3 ,LINGuangqi
(1.801HydrogeologicalandGeoengineeringBrigade,ShandongProvincialBureauofGeology&MineralResources,
Jinan250014,China;2.ShandongEngineeringResearchCenterforEnvironmentalProtectionandRemediationon
Groundwater ,Jinan250014,China;3.KeyLaboratoryofGroundwaterResourcesandEnvironment,Shandong
ProvincialBureauofGeology&MineralResources,Jinan250014,China)
Abstract:KarstspringsinnorthChinaareimportantnaturalresourceswithmultipleattributesofland
scape ,cultureandtourism,whichplayimportantrolesinthedevelopmentoflocaleconomyandsociety.
Monitoringdataofprecipitation ,groundwaterwithdrawalandartificialecologicalrechargeofBaotuSpring
from2016to2018arecollectedformodellingpurposesinordertoachievemoreaccuratepredictionre
sultsofdynamicchangesofthekarstspring.Meanwhile ,6typesofBPneuralnetworkandgeneticalgo
rithmoptimizedBPneuralnetworkareestablishedtopredictthewaterleveloftheBaotuSpring,andthen
thepredictionresultsarecomparedandevaluated.TheresultsshowthattheBPneuralnetworkoptimized
bygeneticalgorithmcanimprovethepredictionstability ,reducethemaximumiterationofneuralnetwork
andsaveagreatdealofcalculationcost ,comparedtoBPneuralnetwork.TheGA-BP(LM)neuralnet
workwhichadoptsLevenbergMarquardttrainingmethodismoresuitableforthepredictionofkarstspring
waterlevelduetoitsadvantagesofstableperformance,lowcalculationcostandsmallpredictionerror.
Thisresearchcanprovidereferencesforthedesignandimplementationofprotectionmeasuresforkarst
springsinnorthChina.
!"#$:20210729; %$:20211226
'()*: PQ¤Ù$Iï_'((41772257、41472216);§D#¤Ù$Iï_'((ZR2021QD084);§D#@wH
jxf.$4'((KY2018003、KY201933、KY202108); §D#@w.¯dÝ9Ä@GW@H+( °§
D#@A9NO,-M9/GW%ð45^U) ï_'((801KY202001)
+,-.: &±>(1974-), +, $cd´"., I0, ¥GW., 45678@A9KLMNO、 à±z9ÂÍ*,-。
ÑÒ+,: ¥ F(1990-), +, §DR¹., /0, GW., 45678@A9³NMz«、 à±z9ÂÍ*,-。