Page 85 - 《水资源与水工程学报》2024年第5期
P. 85

! 5#       ÅÆÇ, :: s½ CNNLSTMAttention ¹ºÈhg¯°g¹ÉJÊ4ËÌ1ÍÄ                                  8 1

                Ð" >ð%<[J]. 9'r5,2021,43(6):8085+97.              š[J]. ›'[ÒØ,2022,38(6):59.
            [10]MAN Yuanyuan,YANG Qinli,SHAO Junming,etal.     [26]F ¥, à É, «ãã, å. wxe'5UVáäËA
                 EnhancedLSTM modelfordailyrunoffpredictioninthe   Ëü‘%< [J]. ARASŠŒ ( Q§B),2021,52
                 upperHuaiRiverBasin ,China[J].Engineering,2023,   (9):1631.
                 24:229238.                                   [27]ø 6, 4TÐ, ›8‘. /Á Delta >ð{Ç5UVáõ
            [11]BIAN Lekang,QIN Xueer,ZHANG Chenglong,etal.        ~" ¢£[J]. 9'cô,2012,43(7):1114+46.
                 Application ,interpretabilityandpredictionofmachine  [28]ò†), 1 ú, ` e, å. gVáõӜ Delta-DCSI Ë
                 learningmethodcombinedwithLSTM andLightGBM :a     ü‘>ð[J]. Œne..ø( §»-.ÿ),2009,41
                 casestudyforrunoffsimulationinanaridarea[J].Jour  (5):17.
                 nalofHydrology ,2023,625:130091.              [29]KLL, `ê.. ›'Ëü‘>ðä Delta >ðŒwr
            [12]`Lb, q F,  P, å.TVGM-LSTM¥èYVC,ç                  5EgáõӜ{š¢£[J]. á¬.ø,2007,65
                 UYZk±¢£[J]. ADEÚ¯,2023,39(6):104110.             (4):653662.
            [13]F_`, F þ, F&˜, å. /Á LSTM ^|îï{                [30]HOCHREITERS,SCHMIDHUBERJ.Longshorttermmem
                 WRF/WRF-Hydro¥ è ç U " ø [J]. A R . ø,            ory [J].NeuralComputation,1997,9(8):17351780.
                 2023,54(11):13341346.                        [31]CHAUDHARIS,MITHALV,POLATKANG,etal.An
            [14]GUJiuxiang,WANGZhenhua,KUENJ,etal.Recent           attentivesurveyofattentionmodels[J].ACM Transac
                 advancesinconvolutionalneuralnetworks [J].Pattern  tionsonIntelligentSystemsand Technology(TIST),
                 Recognition,2018,77:354377.                      2021,12(5):132.
            [15]r(5, ­JŠ, ½Äˆ, å. /Á LSTM{Ÿtà:ß                [32]º Š, +œ´. /ÁDÊ{ CNN-LSTM-Attention ’
                 sg&V D U V ç U Y Z C "   [J]. Ý n ß s,            ¾" YV%<[J]. Q!d(=ù-.ŠŒ,2023,19
                 2021,43(4):11441156.                             (9):6975.
            [16]1Bþ, 2úú, DT, å. /Á Attention ¸_ø            [33]KRIZHEVSKYA,SUTSKEVERI,HINTONGE.Ima
                 LSTM-Seq2seq YV{çUYZ%<[J/OL]. Ýnß                 geNetclassificationwithdeepconvolutionalneuralnet
                 s,2023,[20240415].http://kns.cnki.net/kcms/de  works [J].CommunicationsoftheACM,2017,60(6):
                 tail/62.1072.P.20231103.1335.002.html.            8490.
            [17]lZ´. ér5UV=jÚ¯äàK1&+cd$Ä{                      [34]SRIVASTAVAN,HINTONG,KRIZHEVSKYA,etal.
                 3‹[EB/OL].(20190918)[20191015].http://www.    Dropout :asimplewaytopreventneuralnetworksfrom
                 qstheory.cn/dukan/qs/201910/15/c_1125102357.htm.  overfitting [J].JournalofMachineLearningResearch,
            [18]“,†, 2ŸR, aMT, å. gVA´ïCbﻑ‘                      2014,15(1):19291958.
                 1>ð[J]. AR.ø,2024,55(1):112.                 [35]I„“, 1‚µ, 㠔, å. /Á LSTM-Adam{ú™
            [19]. Í, «Pî, “,†, å. /ÁgVA´ïžªä                       ¸Wá"ÚYV[J]. ¸S§»,2024,41(1):175182.
                 SWAT YV{V5UVAqr´ï˜~¢£[J]. AR                  [36]MORIASIDN,ARNOLDJG,VANLIEW MW,etal.
                 AS-Šu+,2024,44(1):915+22.                        Modelevaluationguidelinesforsystematicquantificationof
            [20]aMT, “,†, «Pî. /Á Tapio ä LMDI YV{V                accuracyinwatershedsimulations [J].Transactionsofthe
                 5UVö¤Ñ[ADER†³Œ¨¿¢£[J]. AD                         ASABE ,2007,50(3):885900.
                 EÚ¯,2023,39(4):94101.                        [37]LEGATESDR,MCCABEJRGJ.Evaluatingtheuseof
            [21]cùÈ, “,†,  ©, å. /Á SWAT{V5UVA                    “goodnessoffit”measuresinhydrologicandhydroclimat
                 EŸ–12¢£C" [J]. 6Aa\[AR-Š( Q                       icmodelvalidation [J]. WaterResourcesResearch,
                 §B),2023,21(5):862872.                           1999,35(1):233241.
            [22]x J, c Ž, «Pî, å. /Á SWAT{V5UVAD               [38]1 •,  [, F –, å. áõ[shR†~ixµA
                 E=j²J'ŽCã¡»‘¢£[J]. ADE[A§                         5UVçUYZ[J]. =j.ø,2017,37(4):12521260.
                 ».ø,2023,34(6):19.                           [39]p ¹, p…=.2021 vÆÿV5Zå‘ACAv\‘
            [23]^', ˆ º, “?, å. /Á SWATYV{V5U                    ¢£[J]. 9'r5,2022,44(S2):89+12.
                 V`QcÕçèzf”‰‹ŒC7†­¢£[J]. Q                     [40]c8ƒ, 1P<, c? , å. V5çU1~€C,
                 !xÆARAS,2022(6):7683+89.                         ‚ƒ„…¢£[J]. 9'r5,2023,45(11):4954.
            [24]+0ú, «¤Í, 2 ‘, å. /Á CMIP6 {()Czf              [41]ˆG„, ë <, ëµg, å. ¥è Encoder-Decoder {
                 ‹zf|}˜~[J]. ADEÚ¯,2023,39(2):4049.               LSTMçU"øYV%< [J]. jke..ø ( §.
            [25]àd[, 1Úu. ¼mY­ÛÔYVQ+1¿>ð                         ÿ),2022,55(8):755761.
   80   81   82   83   84   85   86   87   88   89   90