Page 83 - 《水资源与水工程学报》2022年第4期
P. 83

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

            DOI:10.11705/j.issn.1672-643X.2022.04.11


            '‡Þ§KdL؊艚TytŽˆ‰RS


                                                1          1         2         2
                                         (Š , ‹mŒ , LŽ , €
                (1. ]üHI 9|9èIX, !` ,_ 210098;2. ],#Ïp>BT‹eÔGW:¼ÓF., ], ºp 466623)
                 A B: ˜e9Ä(·^~k(r‘¢=:æã°oç?°»;çsîS, ÑÝt)0D§«—ß———.Gè
                 H—ß AEFAM LSTMñ{78‡Ÿop!‘§«, :t AEFA-LSTM(rB0, â‹ÏpH0T‹]‘19
                 ÄDTrc~kbT8n“‘¢op78§«}~?(r¨”, ½‚M²ô§«—ß( «²—ß GA?rì_—
                 ß PSO) :t GA-LSTM? PSO-LSTM(rB0opeÁ。‡ˆ‰Š:AEFA-LSTMB0(r™»°¼eµ
                 H¼1¯ GA-LSTMB0? PSO-LSTMB0¨öá¢Y 7.59%? 5.22%, è»°»eµH MAE、 °6µH MSE、
                 °6µH RMSE°8 3 tB0^<Œ, MŠÅ:t AEFA-LSTMB0¶‹Ã¥1v@(r~kb, 89Ä(
                 ·ŽÝt)0¥1v~k(r6ß。
                 CDE: ~k(r; .GèH—ß;AEFA-LSTMB0; !‘§«; T‹
                                     +
                 FGHIJ:P333;TV121 .2   KLMNO:A    KPQJ:1672643X(2022)04007906
                         Runoffpredictionmodelforlargeirrigationareasbasedon
                                  artificialelectricfieldalgorithm optimization

                                                                                         2
                                              1
                                                            1
                                                                         2
                               WANGXiaoxin,CENWeijun,LIZhaohui,WUGuanghua
                 (1.CollegeofWaterConservancyandHydropowerEngineering,HohaiUniversity,Nanjing210098,China;2.Construction
                  AdministrationBureauofPhase Ⅱ ProjectofZhaokouYellowRiverDiversionIrrigationArea,Zhoukou466623,China)
                 Abstract:Inviewofthenonlinearandnonstationarycharacteristicsofrunoffpredictiondataseriesinthe
                 hydrologicalprediction ,anewintelligentoptimizationalgorithm,artificialelectricfieldalgorithm(AE
                 FA),iscombinedwithLSTMneuralnetworktooptimizethenetworkparameters,thentheAEFA-LSTM
                 predictionmodelisestablished.ThemeasuredannualrunoffofXuanwuHydrologicalStationofWohe
                 RiverinZhaokouLargeIrrigationAreawasthenusedasthesampledataforthenetworkoptimization
                 trainingandpredictionanalysis,andtheresultswerecomparedwiththoseoftheGA-LSTM modeland
                 PSO-LSTM modelwhichwereestablishedusingconventionaloptimizationalgorithms (GAandPSO).
                 ThecomparisonanalysisshowsthatcomparedwithGA-LSTMmodelandPSO-LSTMmodel,theaver
                 agerelativeerrorofthepredictedvaluecalculatedbytheAEFA-LSTMmodelisreducedby7.59% and
                5.22% respectivelyanditsaverageabsoluteerror,meansquareerrorandrootmeansquareerrorarethe
                 smallestofthethreemodels ,indicatingthattheAEFA-LSTM modelcanpredicttherunoffmoreaccu
                 rately,andprovideanewhighprecisionrunoffpredictionmethodforthehydrologicalprediction.
                 Keywords:runoffprediction;artificialelectricfieldalgorithm(AEFA);AEFA-LSTMmodel;param
                 eteroptimization;irrigationarea

                                                               «– p ¼ e / 0, æ ã ° o ç ? ° » ; ç s î
            1 ./01
                                                               S  [3] , è~k=:ʼn5"£ytîçÐ÷ܚ
                 ~k(rT89Ä(·^Ý'R§N¿, È                         ¯GƒÈvœ。e[, P$%&'Ivx4Y~k
                                                                                  [4-6]
            (r‡ˆe(=JHIË、 ŸF(ƒ9KLæ㐠                           (r6‡¼"45              。(, ~k(r¦§ã
            bc    [1-2] 。Ùª, H0T‹~kbۂƒj                      : W^g(r?‘¢^g(r。È^, :tï

                !"#$:20220108; %&#$:20220516
                +,-.: ®­@(1998-), +, eŸÛ0., .045, 456789|GW¹º¨”。
                ÑÒ+,: •IŽ(1977-), +, ì!Ø., /0, 23, 456789|GW¹º¨”。
   78   79   80   81   82   83   84   85   86   87   88