Page 74 - 《水资源与水工程学报》2022年第6期
P. 74

7 0                      & ' ( ) & * + , -                 2022 $

                 ageerror ≤15.0%,thepassrate≥73.3%,andthecoefficientofcertainty ≥0.99;however,itper
                 formedpoorlywhentheforecastperiodis ≥7months.Ithasanidealforecastingeffectforthedailyrunoff
                 withaforecastperiodof1-3d,withthemeanabsolutepercentageerror ≤1.23%,thepassrate=
                100%,andthecoefficientofcertainty ≥0.999;italsoperformssatisfactorilywhenforecastperiodis4-
                7d,withthemeanabsolutepercentageerror ≤ 15.3%,thepassrate ≥ 73.0%,andthecoefficientof
                 certainty ≥ 0.94;whereaswhentheforecastperiodis ≥ 8d,theforecasteffectispoor.TheWPT-
                 HPO-ELMmodelcangivefullplaytotheadvantagesofWPT ,HPOandELM,showinghighforecast
                 accuracyandstability ;however,theforecasterrorincreaseswiththeincreaseoftheforecastperiod.The
                 proposedmodelandmethodcanprovideanewapproachforthemultistepforecastingofrunofftimese
                 ries.
                 Keywords:runoffforecast;waveletpackettransform(WPT);hunter-preyoptimization(HPO);ex
                 tremelearningmachine(ELM);multistepforecast;simulationtest
                                                               WD-PSO-LSSVM`Œ—祀!"o}NNO
            1 ./01
                                                               xó£"©ª«Äv, ‹¬­®Ã`Œ—çâľ
                 £"MörÿÄÚ, ™(Ü£"Mörÿ°4                         .ŸÄÚPO; =ÊÊf          [9] }~«ò£Ð(sin
            ÄÚKðÜNOÄÚ&JŸº%’3%。¸.£"                             gularspectrumanalysis ,SSA)- 8óoéG‘ - Øé
            Mörÿ°4ÄÚPO¥àNOÄÚ、 NQRa'                            L–Å2„(supportvectorregression,SVR) —ç, {
            dufâÄ$$Z[。^à!ò£"íŽ|hÅ]                             |õ!m[:ó£"Äv5l¥—穪«†;,
            íÁ,BP(backpropagation) ËÖÖh       [1] 、 ØéL–       ‹¬­®Ã—ç„Ž­+>YôÇNO£"Ÿ¯Ï
                                          [2]
            Å(supportvectormachines,SVM) 、 ,Å^8(ran           ¼X, âÄ­·ŸÄvY¬; ûH¶f               [10] i>¡
                          [3]
            domforest ,RF) 、 }ÏM€`ËÖÖh(longshort              —k£(VMD)、 ¢£Éʏ(detrendedfluc
                                       [4]
            termmemorynetworks,LSTM) fÑ\ÄvK‘3                  tuationanalysis ,DFA)’ } Ï M € ` Ë Ö Ö h
            ÀÇCcZŸÄÚY¬。1×, 7à“ £ - Äv                         (LSTM) K‘, ¸¹ DFA-VMD-LSTMö£"ÄÚ
             - ûŽ” Ÿb„ÄvԌK‘·*üƒuà£"M                          —ç¥9:Njö£"©ª«ÄÚ, ÞC¾·ÄÚ
            örÿÄv&J, •ñìQf             [5] 7ࡏ—k£            Y¬; j‚#f     [11] 7àMɏ£K‘’ØéL–Å
            (variationalmodedecomposition,VMD) K‘’í*           (SVM) }~«`Œö£"Äڗç¥ÕÆxö£
             - }Ï݀`ËÖÖh(convolutionalneuralnetwork            "©ªÄÚ, †;«Ã—çâÄ­·ŸÄÚΕX;
            andlongshortterm memorynetworks ,CNN-LSTM)        j'*f    [12] «MÉ¡.(ØéL–ÅR‹Œ, ¸¹
            }~ VMD-CNN-LSTM—ç, {|!"oÐ3                        duMÉØéL–Å2„—祫!"oÐ3hN
            hx’Ó3îx 1960-2005 ™5vó£"±í¥                        Oxö£"©ªÄÚ, ‹¬Q˜Ã—ç;ÄY—¹
                                                                                     [13]
            —穪†;, ‹¬­® VMD-CNN-LSTM—ç                        ’ÄÚö£"; {éVf             €ŒMɏK‘、 J
            âÄ­oŸÄvPOÀP­MŸÇ¿; â;ïf                      [6]    –èoéG‘’ØéL–Å, h}«MÉ - J–è
            7àֆ—k£(empiricalmodedecomposition,               - ØéL–2„¨Œö£"Äڗ礥8}@r
            EMD) ( Elman ËÖÖh}~« EMD-Elman ó£                  gùõNOxö£"©ª«ÄÚ, ‹¬­®Ã—ç
            "`ŒÄv—ç, ¤«(ƒuàj!ËgKùðN                            Ïö£"ÄÚrâľõŸ‚ƒX; ÐÏWf                    [14] }
            Ox 1979-2009 ™Ÿó£"MörÿÄvr; ñG                      ~«Ö†—k£ (EMD)-‘/öe (forensic
            tf   [7] }~«S—ûŒÖ†—k£(complemen                  basedinvestigation ,FBI) G‘ - ‘W6eÅ(extreme
            taryensembleempiricalmodedecomposition,CEEMD)      learningmachine ,ELM) `Œ£"Äv—ç, ¤«Ã
             -BP ËÖÖh—ç, ¤{|8!ó£"Äv5l¥                         —çƒuàQ"#÷Î!NOx™£"ŸÄv, Ç
            —穪«†;, ;®«Ã`Œ—çÄvY¬¾·;                           C«¾·ŸÄvY¬。
            sÝ>f     [8] 7à/ÞûŽÖ†—k£(complete                     5È, ˈ—çÃK‘÷ÏÀZE<Ã:j:
            ensembleempiricalmodedecompositionwithadaptive     (1) c°±£"ÄÚ¸`¥Ä¯ÝM 1œóà 1d
            noise ,CEEMDAN) ’Mɏ£ (waveletdecomposi           Ÿø4ÄÚ, ÈÏ5Tƒur, ø4Äڒ’À‘c
            tion ,WD) `ŒŸ¸Ô£K‘, ¸¹ CEEMDAN-                   jó£"Ãö£"Äڟ5T÷K, {[í÷$ì
   69   70   71   72   73   74   75   76   77   78   79