Page 88 - 《水资源与水工程学报》2023年第1期
P. 88
!34 " ! 1 # & ' ( ) & * + , - Vol.34No.1
2023 $ 2 % JournalofWaterResources&WaterEngineering Feb.,2023
DOI:10.11705/j.issn.1672-643X.2023.01.10
'«´H:H(H|Vµ:;
1,2 1,2 1,2
Zmq , crs , Ztu
(1. õðÑÁ*3ie 12Xè%&el, 9õ È 712100;2. õðÑÁ*3ie
pÑÒ1j%&IEn:;}~, 9õ È 712100)
A B: Ð=|S56Ì
Sqy¥Û、 ÑÒ®、 1rs¿]'):P^_。£Ì|S>X')ÍAö¿
´ösU´, È5ß±ª|S56éM VMD-(CNN-LSTM,ELMAN)。³·h VMD§|S>X»
¾«T»Ú, ¦6ÁëT»ÚÂU;á(SE), §`7»¾ÉKÎK»Ú; Æh CNN-LSTM
56ÉK»Ú, h ELMAN56ÎK»Ú; 8Ƨ56ÃÄ:ûk¨8®56ÃÄ。§
m
'Sqvó)êùëêí51Uª|S56, ¦X CNN-LSTM、ELMAN、VMD-CNN-LSTM56
ÃÄ·¸ÌÓX¡e。+^ÃijÅ: UU56ÃÄ NSEáÖi
0.99,
`ä, ³Å VMD-(CNN-
LSTM,ELMAN) ')ÍÉ56[¿, Þ
}"+^p|S56。
CDE: |S56; ê»T»; µBC; m'Sq
+
FGHIJ:TV121 .4 KLMNO:A KPQJ:1672643X(2023)01008407
Runoffpredictionmodeloffrequencydivisionbasedon
variationalmodedecomposition
1,2
1,2
ZHANGXiaoxuan ,SONGSongbai ,ZHANGBinglin 1,2
(1.CollegeofWaterResourcesandArchitecturalEngineering,NorthwestA&FUniversity,Yangling712100,China;
2.KeyLaboratoryofAgriculturalSoilandWaterEngineeringinAridandSemiaridAreas,MinistryofEducation,
NorthwestA&FUniversity,Yangling712100,China)
Abstract:Accuraterunoffpredictionisofgreatsignificanceforagriculturalirrigation,reservoirschedu
ling,floodcontrolanddisastermitigationinthebasin.Aimingatthestrongnonlinearityandnonstation
arityoftherunoffseries,ahybridmodelformonthlyrunoffprediction,VMD(CNN-LSTM,ELMAN),
isproposed.Firstly,VMDisusedtodecomposetherunoffsequenceintomultiplemodalcomponents,
andthesampleentropy(SE)ofeachmodalcomponentiscalculated,accordingtowhichthecompo
nentsaredividedintohighfrequencyandmediumlowfrequencycomponents.ThentheCNN-LSTM
modelisusedforthepredictionofhighfrequencycomponents ,theELMANmodelforthemediumlow
frequencycomponents.Finallythepredictionsresultsaresummedup.Themodelisthenappliedtothe
monthlyrunoffpredictionofBaimasiStationandHeishiguanStationinthemiddleandlowerreachesof
theYellowRiverBasin ,andthepredictionresultsareevaluatedcomparedwiththoseofCNN-LSTM,
ELMAN,VMD-CNN-LSTM models.ResearchresultsshowthattheNSEvaluesofthepredictionre
sultsofthismodelareallgreaterthan0.99 ,whichissuperiortoothermodels,indicatingthatthe
VMD-(CNN-LSTM,ELMAN)modelhashighpredictionaccuracyandcanbeappliedtoactual
monthlyrunoffpredictionofthebasin.
Keywords:runoffprediction;variationalmodedecomposition(VMD);neuralnetwork;theYellow
RiverBasin
!"#$:20220619; %$:20220826
'()*: 89*e2./(52079110)
+,-.: ½å(1999-), A, :q\hE, KG+^L, OPQR1U5ʵf+^。
/0+,: òøæ(1965-), b, 9õïçE, FG, IJ, OPQRSq1UNX1U5Ê]µeIeX*+%o。