Page 195 - 《水资源与水工程学报》2024年第5期
P. 195

!35 "!5 #                         & ' ( ) & * + , -                               Vol.35No.5
               2024 $ 10 %              JournalofWaterResources&WaterEngineering                 Oct.,2024

            DOI:10.11705/j.issn.1672-643X.2024.05.23


                 5”·¸¹·¸qº»¼€^g‚YZ_½¾¿


                                            1         2        1         2         1
                                     ./0 , 123 , ( 4 , (56 , 789
                       (1. /f0’0s+[<=>?, /f 1P 524000;2. !"[=[(&'0F…, .2 100048)
                 H I: §Ú×3n^Õ¹³Ð(TBM) é`öa!/ÏVQÀ’ì–¹³·§, ™f4{Š‹–¹³·§hš
                 *-·§Jì, Ž^Õ`a–5hȐ`E<H¨¬­。xo, G§ú6&7`a TBM¹³·§–q¢, *5À
                 :€– TBM¹³·§8ó–W‰, ð9Ý"¹³£*Q、 ˜y¹³£¾:“k¹³,·â;。o‡, «¬ü×8
                 P$1–a<Ã=>û?(LSTM) '‰•¨À@AB\•·(TPI) 8P¹N, “|§·§8ó–W‰–<kr。¼
                 ½Ò›: Ö*5–·§8ó–W‰hšnoC·§Jì, *-\æ'D¹N–8P(æ, |§:–º¡Žæ R                                  2
                 Z 0.503 *Eß 0.721, #—„· R′ Z 0.809 *Eß 0.900, *}#ŽF+ MREZ 3.107 âß 0.096。0F€½
                 Ž»*- TBM¹³·§–Ìœr¤õ}rE<H¨¬­, ‹õY#—GG–0F*+,-。
                 JKL: ^Õ¹³Ð(TBM); ·§8ó–; H&æŠ9; IqJKLM; a<Ã=>û?(LSTM)
                 MNOPQ:U455   RSTUV:A    RWXQ:1672643X(2024)05019110

              TBM excavationdatapreprocessingmethodandengineeringcaseverification

                                                                            2
                                                                                            1
                                                      2
                                      1
                                                               1
                             NIEQitan,XIAOHaohan,LIUFei,LIULipeng,NIURuiqiang
                               (1.GuangdongYuehaiYuexiWaterSupplyCo.,Ltd.,Zhanjiang524000,China;
                             2.ChinaInstituteofWaterResourcesandHydropowerResearch,Beijing100048,China)
                 Abstract:Thefullfacehardrocktunnelboringmachine(TBM)automaticallygeneratesmassiveexcava
                 tiondataduringthetunnelconstructionprocess.Properscreeningandcleansingofexcavationdataiscru
                 cialtodataquality,whichalsohasgreatguidingsignificancefortheintelligentconstructionoftunnelen
                 gineering.Therefore,basedonthecharacteristicsofTBMexcavationdataintheYinchuoProject,anin
                 tegratedTBMexcavationdatapreprocessingmethodisproposed,whichincludescompleteexcavationseg
                 mentextraction ,internalexcavationsegmentation,andexcavationparameternoisereduction.Toverify
                 theeffectivenessoftheproposeddatapreprocessingmethod ,atorquecutdepthindex(TPI)prediction
                 modelisdevelopedbythelongshorttermmemory(LSTM)algorithm,whichhasstrongtemporalpredic
                 tioncapabilities.Theresultsdemonstratethattheproposeddatapreprocessingmethodcansignificantly
                 improvedataqualityandenhancethepredictionaccuracyofdeeplearningmodels.Forthevalidation
                         2
                 dataset ,R increasesfrom0.503to0.721,R′ascendsfrom0.809to0.900,andMREplummetsfrom
                3.107to0.096.Theseresearchachievementsbearprofoundimplicationsforenhancingtheprecisionand
                 reliabilityofTBMtunnelingdata ,therebyofferinginvaluableinsightsforfurtherexplorationintherelated
                 domain.
                 Keywords:tunnelboringmachine(TBM);datapreprocessing;kerneldensityestimation;Butterworth
                 filtering;longshorttermmemory(LSTM)
                                                               ¬»ÍnBN、 5O、 ÒÓ、 ËÌÙ{°PR^ÕÓ¹
            1 ./01
                                                                                 [1]
                                                               O1–Ë¡ÐQ8R 。ŠS…^Õ¹³W‰#z,
                 ^Õ¹³Ð(tunnelboringmachine,TBM) ‘K»            TBME<`åæT、 AÓJì-、 ˆ&kŒU、 ¹³

                !"#$:20231126; %&#$:20240513
                '()*: !"[=[(&'0F…(/&01234**+(GE0145B012021); "E/n&'()*+(52179121); ž
                        G[E͹ºŠ4¯"EHz{|}/&0FCD(SKL2022ZD05)
                +,-.: )©*(1987—), M, ef&+T, '$, `aj, &¨S,[=`a•8ϖ`O。
                /0+,: ¸·%(1993—), M, -.N[T, #$, `aj, &¨S,[=`a&0`O。
   190   191   192   193   194   195   196   197   198   199   200