Page 12 - 《水资源与水工程学报》2022年第4期
P. 12

!33 " ! 4 #                       & ' ( ) & * + , -                               Vol.33No.4
               2022 $ 8 %               JournalofWaterResources&WaterEngineering                 Aug.,2022

            DOI:10.11705/j.issn.1672-643X.2022.04.02


                     '‡ WD-COA-LSTMˆ‰Š‹ŒY7Ž



                                                ()*, +,-, ./0
                                        ( EŸ9|9èHI 9KLIX, ], /0 450046)
                 A B: 8oÝA¥dá9b(r1v, "Y﯌2¨Z(WD) ?34§«(COA) —߁r5Ôã6ñ{
                 78(LSTM) á9b(rB0(WD-COA-LSTM)。9¨’Œ2¨Ze‚ƒ=:op(µF, ò=:°»;
                 ç, þÅ 1 l¢<=:? 3 l¥<=:; Ù×K34§«—ßeñ{78(LSTM) B0op!‘§«; <×ÑØ
                 ì=:(r™=4þÅdá9b(r™。э"B0'’¯>?¶@jABCt?>ÄAýAt lÐbD
                 dá9b(r^, âM LSTM、COA-LSTM、WD-LSTMB0(r‡ˆopeÁ。‡ˆ‰Š: " WD-COA-
                 LSTMB0(r1v<¥, MŠŒ2¨Z?34§«—ßDã34ÿ LSTMB0(r1v?Û«DE, 8dá
                 9b(rŽYÝt)}~。
                 CDE: dá9b(r; Œ2¨Z; 34§«—ß; r5Ôã6ñ{78
                 FGHIJ:TV125   KLMNO:A    KPQJ:1672643X(2022)04000806

                   MonthlyprecipitationpredictionbasedonWD-COA-LSTM model

                                     WANGWenchuan,YANGJingxin,ZANGHongfei
                 (CollegeofWaterResources,NorthChinaUniversityofWaterResourcesandElectricPower,Zhengzhou450046,China)
                 Abstract:Inordertoimprovethepredictionprecisionofmonthlyprecipitation,theprecipitationpredic
                 tionmodelofWD-COA-LSTMisproposedbasedonwaveletdecomposition (WD),coyoteoptimization
                 algorithm(COA)andlongshorttermmemory(LSTM)neuralnetwork.Firstly,thetimeseriesispre
                 processedbyWDtoeliminateitsnonstationarity ,andalowfrequencysequenceandthreehighfrequen
                 cysequencesareobtainedastheresult.ThentheparametersoftheLSTMmodelareoptimizedbyCOA.
                 Finally ,thepredictedmonthlyprecipitationisobtainedbysuperimposingthepredictedvaluesofeach
                 subsequence.TheproposedmodelwasappliedtothemonthlyprecipitationpredictionofBaituTownin
                 LuanchuanCountyandGuxianTowninLuoningCounty ,LuoyangCity,andtheresultswerethencom
                 paredwiththoseoftheLSTM,COA-LSTM andWD-LSTM models.Itisfoundthattheproposed
                 WD-COA-LSTM modelproducedthehighestpredictionaccuracy ,indicatingthatWDandCOAcan
                 improvetheprecisionandgeneralizationabilityofLSTMmodel.Thismodelprovidesanewapproachfor
                 thepredictionofmonthlyprecipitation.
                 Keywords:monthlyprecipitationprediction;waveletdecomposition(WD);coyoteoptimizationalgo
                 rithm(COA);longshorttermmemory(LSTM)neuralnetwork

                                                               8  [6-8] s。È^r5Ôã6(longshortterm memo
            1 ./01
                                                               ry,LSTM) ñ{78B0D1•@(r‚ƒ=:, ã
                 á9(r±9Ä(·F<Ý'R§45$                           3ZQYÝzñ{78^"5RvSTÖ×, Â
            ð, á9b?L±GfHI、 JKIˁÀ·Íy›                          8mc£45vS。ô Kumar s             [9] m’UOñ
            Ý, Í[, eȕ/(ræãR§bc。mc£&'                          {78(recurrentneuralnetworks,RNN) B0?r5
            &QIveá9b(ropYHbLM, ¦§4                            Ôã6(LSTM) ñ{78B0eVvR½‹<á9
            56ßã¤NOB0          [1-2] 、 jë7bP   [3-5] 、 ñ{7     bopY(r, ‡ˆ‰Š LSTMB0Á RNNB0(


                !"#$:20211109; %&#$:20220429
                '()*: ],#RS4fM‚Ú&'(202102310259、202102310588); ],#¥'$%()*+(18IRTSTHN009)
                +,-.: ®Äj(1976-), +, ],,-., /0, 23, /0., 456789ÄI*9KL。
   7   8   9   10   11   12   13   14   15   16   17