Page 176 - 《水资源与水工程学报》2022年第2期
P. 176
!33 " ! 2 # & ' ( ) & * + , - Vol.33No.2
2022 $ 4 % JournalofWaterResources&WaterEngineering Apr.,2022
DOI:10.11705/j.issn.1672-643X.2022.02.23
']qr¢\õsãôOst_`auv'H²
1
2
2
1
1
x ^ , Ï ¢ , nÐ| , 1 , hº{ , JÑ{ 3
(1. î´aäå3% r¯%j, î´ ö6 123000;2. Ë$3% U`(%gU/UúaöwBCkl,
8 100084;3. £$Ç%¤IWobÜx-Õr, ^¶· }¤Ï] 021000)
C D: þTpDâY*`fpýGëSoBF», íçüost²#
DâYm4F。4~þÄpqª*`T¶, 4b%Êä(ELM) T+êm*`f, q
r²#Y¾éÚp¤ghmDâ4p¡。TT, 4f¸Ç¿T+
ê, è*`fh°Tª«-wQ-, ª«-f²-0ÿm, Q-f MATLABì|Ý
pb%Êäm, È-¹¼Û.*³´±Åp-;f¹。;Y ¡¢£: v@p ELM
IJKTýGÖSoTq0ÿÎp*`-;f, ¡pÇ¿TÍT 23.5mm, ì9DT 0.973。gÀ
[åYwOY¼, ¥åY(ACO) k;YrwDâp¡¨D, ¥åYDâb%Êäf
pTî±, Ç¿TÍT 10.1mm, ì9DT 0.998, *kæÖ*`pf4FmY。
EFG: *`; f; f¸Ç¿; ¥åY; b%Êä
+
HIAJK:TV698.2 3 LMNOP:A LQRK:1672643X(2022)02017207
Landslidedisplacementpredictionbasedonextremelearning
machineoptimizedbyantcolonyalgorithm
2
1
3
1
1
2
CAOBo,WANGShuai,SONGDanqing,DUHan,LIUGuangwei,ZHOUZhiwei
(1.SchoolofMining,LiaoningTechnicalUniversity,Fuxin123000,China;2.StateKeyLaboratoryof
HydroscienceandEngineering,TsinghuaUniversity,Beijing100084,China;3.OpenpitMineof
ShenhuaBaorixileEnergyCo.,Ltd.,Hulunbuir021000,China)
Abstract:Theapplicationofoptimizationalgorithmswithhighaccuracyisveryimportantforimproving
theaccuracyofthepredictionmodelforlandslidedisplacement ;however,theresearchonthecomparison
ofdifferentoptimizationalgorithmsisrarelyreported.Here ,theBazimenlandslideintheThreeGorges
Reservoirareawastakenastheexample ,andtheextremelearningmachine(ELM)modelwasusedto
predictthelandslidedisplacement.Meanwhile ,multiplealgorithmswereusedtooptimizetheparameters
inthemodellingprocesstoimprovethepredictionaccuracy.Inordertoimprovethepredictionaccuracy ,
basedonthemovingaveragemethod,thelandslidedisplacementwasdecomposedintotwophases,which
weretrendtermandperiodictermdisplacements.Thetrendtermdisplacementwaspredictedbyapoly
nomialfunction ,andtheELMmodelthatwascompletedbyMATLABcodewasusedtopredicttheperi
odictermdisplacement.Finally ,thetrendandperiodicdisplacementsweresummedupasthepredicted
totaldisplacement.TheresultsshowedthatELMmodelcouldaccuratelypredictthecumulativelandslide
displacementwithasteplikecurve ,theaverageerrorofthepredictionresultswas23.5mm andthe
goodnessoffitwas0.973.Comparedwithparticleswarmoptimizationandgeneticalgorithm ,theantcol
onyoptimization(ACO)performedbetteroncomputationaltimeandcalculationresult.Hence,theex
tremelearningmachinemodeloptimizedbyantcolonyalgorithmhadthebestaccuracy ,withtheaverage
!"#$:20210308; %$:20210702
'()*: öPQ7(%+,-.(2020M680583); PQ756OÔ9:;*-.(BX20200191);Ë$3%“ U%%
É” ;*-.(2019SM058); öwìí(%+,-.(51974144)
+,-.: ³ P(1984-), ], î´îNO, PQ, y<`, uQGab, 4FRST-Éþq`a。
/0+,: ù û(1989-), ], î´ö6O, PQ4FG, (vaÖÚÀqÖ¹aRp¼H4F。