Page 50 - 《水资源与水工程学报》2023年第4期
P. 50

4 6                      & ' ( ) & * + , -                 2023 $

               lished.Finally,theestablishedmodelsareusedtopredictandreconstructthecomponentsofpH,DO,
               COD ,NH—N,meanwhile,WPT-GRU,WPT-LSTM andWPT-CSA-SVM,WPT-CO-SVM,
                    Mn    3
               WPT-MGO-SVM modelsareestablishedforcomparativeanalysis.Thenthemodelsareverifiedbythe
               predictionofpH ,DO,COD ,NH—NofGuanyinshansectioninKunming,YunnanProvince.There
                                              3
                                        Mn
               sultsshowthatthepredictionaccuracyofWPT-CSA-GRU,WPT-CO-GRU,WPT-MGO-GRU,
               WPT-CSA-LSTM,WPT-CO-LSTM,WPT-MGO-LSTM modelsforpH,DO,COD ,NH—Nof
                                                                                             Mn    3
               thecasesectionarebetterthanthecomparisonmodelswithbetterpredictionperformance ,ofwhichWPT-
               CSA-GRU ,WPT-CO-GRU,WPT-MGO-GRUmodelshavethehighestpredictionaccuracy;CSA,
               COandMGOcaneffectivelytunethehyperparametersofGRUandLSTM,andsignificantlyimprovethe
               predictionperformanceofGRUandLSTM;thesixproposedmodelshavehighpredictionaccuracyandsmall
               calculationscale,whichcanberecognizedaseffectivepredictionmodelsforwaterqualitytimeseries.This
               studycanprovidesomereferenceforrelevantwaterqualitypredictionresearch.
               Keywords:waterqualityprediction;gatedrecurrentunit(GRU);longshortterm memorynetworks
               (LSTM);waveletpackettransform (WPT);chameleonswarm algorithm (CSA);cheetahoptimization
               (CO)algorithm;mountaingazelleoptimization(MGO)algorithm

                                                               ‘  [12] uh^?jY·1 ¶"8pBPæ‹nµ
            1 ./01                                                                               [13]
                                                               (SSA-GA-BP) [”qr; /Ë@‘               uh^U
                 uX@¡âfg[”ÁZox@•–áW                          ·qB·z UÛ-qV8p 8ñNÉÊæ‹nµ
            @8BFÕþÿ&‚)“             [1] 。£t, * MIKE21 ‘        (VMD-PSO-LSTM) [”qr; ÃLA‘             [14] uh
            @¡q™qrz, ÿ|@¡[”^=pµ·? 3 [。                         ^,°‹ÛqB·z ùò>Z23³(EEMD-
            Ë 1 [´ÎÅ723?Ýï^"#[”=p; Ë 2                         SVR) [”qr; OÈA‘        [15] uh^U·qB·z
            [´Î³I&ù?Ýï^[”=p, Y BP(back                         BTæ‹nµ 8ñNÉÊæ‹nµ(VMD-CNN-
            propagation) æ‹nµ   [2-3] 、 ùò>Z23³      [4] 、 8   LSTM ) x°@¡[”qr。
                                                                   j•Û, ²DIù×W3N´Q^'~,LSTM、
            ñN É Ê æ ‹ n µ (longshortterm memorynet
            works,LSTM) [5-6] 、 ˜‡l#23æ‹nµ         [7] ‘, +    /Ôá•a3(gatedrecurrentunit,GRU) £*¬­
                                                               x@¡[”     [15-16] ^;<¢。LSTM ´á•æ‹n
            [=puXoo˜‡lƒ„^Qi, !´Æ<
                                                               µ(recurrentneuralnetwork,RNN) ^U½, /º?o
            Æð|Æûù^°…¤Ã‡o/[”ÁZ^u
                                                               RNNÎV^@ZBHW@ZFCƒ„ , €ƒ„·
            à, *tÞßg­-V´Q8p}¶­q, Ä9¥û
                                                               [  [17] 、 [”;< [18-21] ‘’e¬­R~。GRU´
            ¥^[”.—。Ë 3 [´x°[”p, x°[”p
                                                               LSTM^U½, K LSTM{º,GRU\ÿk¤^æç
            ·è½=¹, ¼½´'xè†ðè†ÎL[”I^
                                                               ZÎ B k ¥ ^ [ ” l Q, * ¬ ­ x v [ [ ” ;
            x°[”, Y TimeGAN-CNN(convolutionalneural              [22-24]
                                                               <      &¢。€¾‘¬­¢,GRU/LSTM õûù
            networks)-LSTMqr     [8] 、GRNN(generalregression
                                                               ( ¨©ëù、 &ù 、 ÜÝëEÑù、dropout、Learn
                                       [9]
            neuralnetwork)-Markov qr     、STG(spatial-tem
                                                               RateDropPeriod ‘) o GRU/LSTM[”lQþÿû
            poralgraph )-LSTMqr    [10] ‘, x°[”=p^œ
                                                               Iab, R‚‹î3NêëÓ_óL¥å GRU/
            Ñ´J­vqr^q<Ûº?a¼[”qr^ê
                                                               LSTMõûù, êM‚X«ý3óþÿš›^Óû
            P, ![”¯—$$€ûI^êÂ=l; Š¼½´                          ‹Û, ž Ÿ ˆ % ! Ø Ü Í ˆ Æ, I Î O P GRU/
            'xŽ·zH«^[”=p, »=pJ­·zH                            LSTM^[”RX。*t, ³÷e‰ý8p                 [18] 、 UÛ
            «Ÿâfù×·z?bÏkþÇÈ^Ûfg·Z,                             -qV8p‘       [25-28] -V´Q8p (swarm intelli
            ‹îoÛfg·Z}¶[”、 &¶ô¥å±®[”                            gencealgorithms ,SIA) £€­q GRU/LSTMõûù
            ¯—。Üx'xŽ·zH«^[”=pþÿû¥                             ¢¥å¬­, Ĥ¥oû¥^Óq.—。
            ^[”ÁZ, »=p£€@¡âfg[”;<¢                                ?uX pHÆB COD 、DO、NH—NeZâ
                                                                                               3
                                                                                      Mn
                                   [11]
            ¥åR~¬­, Yø=‘              uh^(›·z 8ñ              fg^[”ÁZ, a} GRU/LSTM^[”lQ, š
            NÉÊnµ(W -LSTM) âfg[”qr; (õ>                       ~ SIA€ GRU/LSTMõûùÓq¢^¬­Ö|, u
   45   46   47   48   49   50   51   52   53   54   55