Page 73 - 《水资源与水工程学报》2022年第6期
P. 73

! 33 " ! 6 #                      & ' ( ) & * + , -                               Vol.33No.6
               2022 $ 12 %              JournalofWaterResources&WaterEngineering                 Dec.,2022

            DOI:10.11705/j.issn.1672-643X.2022.06.09


                       WPT-HPO-ELMé_pqŸrpq|}


                                                           1         2
                                                   wxv , yu*
                     (1. Q"#NdNRBv_9&Ji, Q" ·® 650021;2. Q"#OhbN/Ü, Q" Oh 663000)
                 F G: M¸.£"Mörÿ°4ÄÚPO, }~«MÉñ¡.(WPT)- éCéÃoé(HPO) G‘ - ‘W6eÅ
                 (ELM) R€ŒŸ£"Mörÿ°4Äڗç, ¤ƒuàQ"#"H!NOxó£"’ö£"Mörÿ°4ÄÚ。
                 i> HPOG‘†], Ï:L„O~ZäÞ 6 œÝç]±¥ HPO©ªê²†;; du 2 i WPT «£"Mr±í
                 £M 4 œ–rÿ–, Ÿ?nÑ£"rÿ±ííÁX’:²ÎXŸ1Ÿ; ”u HPOoé ELM„>i”-’Áò
                 iè-, }~ WPT-HPO-ELM—ç¥5ló£"’ö£"©ª°4ÄÚ。‹¬­®:HPOG‘âľ·ŸÙo
                 PO’µÜëv„…;WPT-HPO-ELM—ç¥Ä¯ÝM 1~3 œóŸó£"âÄ]şÄÚY¬, Äڟ²³2
                 ¥žDÇ¿≤2.43%, Œ¾ÿ≥99.2%, m•X[±≥0.999; ¥Ä¯ÝM 4~6 œóŸó£"âľ·ŸÄÚY
                 ¬, Äڟ²³2¥žDÇ¿≤15.0%, Œ¾ÿ≥73.3%, m•X[±≥0.991; ÌįÝ≥7 œóM, ÄÚY¬¾
                 ¿。¥Ä¯ÝM 1~3d Ÿö£"âÄ]şÄÚY¬, Äڟ²³2¥žDÇ¿≤1.23%, Œ¾ÿM 100%, m
                 •X[±≥0.999; ¥Ä¯ÝM 4~7d Ÿö£"âľ·ŸÄÚY¬, Äڟ²³2¥žDÇ¿≤15.3%, Œ¾
                 ÿ≥73.0%, m•X[±≥0.947; ÌįÝ≥8d M, ÄÚY¬¾¿。WPT-HPO-ELM—焏'û WPT、
                 HPO’ ELMŸo£, ­“¹¾.ŸÄÚPO’ΕX„, ÄÚÇ¿,-įݟ|@È|c, ×çPK‘;M
                 £"Mörÿ°4ÄÚ¸ë廣。
                 HIJ: £"ÄÚ; MÉñ¡.; éCéÃoéG‘; ‘W6eÅ; °4ÄÚ; ê²v:
                                +
                 KLDMN:TV121 .4;P333   OPQRS:A    OTUN:1672643X(2022)06006908
                             WPT-HPO-ELM multisteprunoffforecastmodel

                                                          1
                                               XUJianwei,CUIDongwen      2
                 (1.YunnanInstituteofWater&HydropowerEngineeringInvestigation,DesignandResearch,Kunming650021,China;
                              2.WenshanPrefectureWaterBureauofYunnanProvince,Wenshan663000,China)
                 Abstract:Toimprovethemultistepforecastaccuracyofrunofftimeseries,anovelmodelwasestab
                 lishedcombiningwaveletpackettransform(WPT),hunter-preyoptimization(HPO)algorithmandex
                 tremelearningmachine (ELM),whichwasthenappliedtothemulti-stepforecastofmonthlyanddaily
                 runofftimeseriesoftheNankangRiverHydrologicalStationinYunnanProvince.TheprincipleofHPO
                 algorithmisintroduced ,and6typicalfunctionsareselectedtosimulateandverifyHPOunderdifferent
                 dimensionalconditions.Thenthedataofrunofftimeseriesisdecomposedinto4subsequencecomponents
                 usingdoublelayerWPT,soastoreducethecomplexityandinstabilityoftherunoffsequencedata.The
                 ELMinputlayerweightsandhiddenlayerbiasesareoptimizedtoestablishaWPT-HPO-ELM model
                 forthepredictionofmonthlyanddailyrunoffinmultiplesteps.TheresultsshowthattheHPOalgorithm
                 hasgoodoptimizationaccuracyandglobalsearchability ;theWPT-HPO-ELMmodelperformsideally
                 attheforecastof1-3monthsmonthlyrunoff,withthemeanabsolutepercentageerror ≤2.43%,the
                 passrate ≥99.2% ,andthecoefficientofcertainty ≥0.999;itcanalsopresentasatisfactoryresultat
                 theforecastofthemonthlyrunoffwithaforecastperiodof4-6months ,withthemeanabsolutepercent

                !"#$:20220527; %&#$:20220725
                '()*: Q"#tå@A}_+,(2018HC024); Q"$%&'9:,1( -.>‹+,); 23çèŒE78,1
                        (2018-1177-02)
                +,-.:  }e(1984-), V, Q"!ªC, .ághI, $§}NdNRgh†:、 }_zYëg;*fgE。
                /0+,: ¦;O(1978-), V, Q"f~C, uwá.ághI, $§}NQRÊ]lmPb„G‘ÏNONQR[
                        \rŸƒu&JfgE。
   68   69   70   71   72   73   74   75   76   77   78