搜索
:100124098(2010)0520075206X,(,100044):,2min4min6min8min10min12min14min16min,RöSHurst,Hurst,,Hurst,,,:;;RöS;;:U491:A12070,,[1-3],[4][5][6]Kolmogorov[7]Lyapunov[8-10],,,,,,,,,,,RöS(RescaledRangeAnalysis,)(ApproximateEntropy,ApEn),2672,(1)20063339(45),2,2,2min4min6min8min10min12min14min16min3(16min)3,(14÷0016÷00),4,285(197)Vol.28,No.520105SystemsEngineeringMay,2010X:2010203224:(NCET20820718);(60874078);(2006BAG01A01);(20070004020):(19862):6720104,x1,x2,,xn,Sn=ni=1(Xi-Xn)2ön(1):Xi=xiöxn(2)Xn=ni=1Xiön(3)xn=ni=1xiön(4)5,2min16min,,,56,,656,2min4min,,36,,,3RöSRöS,Hurst1951,Hurst,[11-15]311{xi},AN775,:Xt,n=tu=1(xu-Mn)(5),Mnnxu,Xt,nnR=max(Xt,n)-min(Xt,n)(6)SXu,RöS,Hurst,:RöS=K(n)H(7)(7),log(RöS)n=Hlogn+logK(8)312HurstHurst:0H015,,,,H=015,,015H1,,,H1,C(t)1,313HurstRöSH,,,77Hurst7,Hurst[015,1],,,,,Hurst,,,,2min,H015434,,1Hurst,5Hurst,H,6,,Hurst,,Hurst,4,,,[16-19]411,mm+1,,,:n{xi},m,Y(i)=(xi,xi+1,,xi+m-1),i=1,2,,n-m+1(9)Y(i)Y(j)dij,dij=maxúxi+k-1-xj+k-1ú,k=1,2,,m(10){Y(i)}Cmi(r)=n-m+1j=1H(r-dij)n-m+1(11):H(õ)01,00;Cmi(r)Y(i),mr,Y(i)Y(j)r,Y(j)(ji)Y(i),{Y(i)}Um(r)=(n-m+1)-1n-m+1i=1ln[Cmi(r)](12):Um(r){Y(i)}872010Um(r)()m,CApEn=Um(r)-Um+1(r)(13)nmrm2,r01100125,ApEn4125,m=2,r=011582min,16min,2min,16min2min4min6min16min,5,,,,9,,,,,6,,Hurst89,,,2min3595,16min629,5,RöS,,:2i16iH,Hurst[015,1],,,Hurst,Hurst,,0195,Hurst,Hurst,,Hurst,,16i629975,:,2min3595,,-,,:[1],.[J].,2004,21(3):8285.[2],.[J].,2008,8(5):6872.[3].[M].:,2008:210,99100.[4],.[J].,2003,36(1):6874.[5].[J].,2006,15(5):445450.[6].[J].,2004,20(2):7376.[7],,.[J].,2004,37(2):101104.[8]ShangPJ,LiXW,SantiK.Nonlinearanalysisoftraffictmeseriesatdifferenttemporalscales[J].PhysicsLettersA,2006,357:314318.[9]ShangPJ,LiXW,SantiK.Chaoticanalysisofuraffictimeseries[J].Chaos,SolitonsandFractals,2005:121128.[10],,.[J].ITS,2005,7(2):1114.[11]WangJ,etal.Analysisonpredictabilityoftrafficflow[J].ITSCommunication,2005,7(2):1114.[12].[J].,2009,26(10):105110.[13],.RöS[J].,2004,19(2):166169.[14]ShangPJ,LvYB,SantiK.Detectinglong2rangecorrelationsoftraffictimeserieswithmultifractaldetrendedfluctuationanalysis[J].Chaos,SolitonsandFractals,2008,36(1):8290.[15].[J].,2006,23(2):115119,127.[16]SnchezGraneroMA,TrinidadSegoviaJE,GarcaPrezJ.SomecommentsonHurstexponentandthelongmemoryprocessesoncapitalmarkets[J].PhysicaA:StatisticalMechanicsandItsApplications,2008,387(22):55435551.[17],,.[J].,2009,58(9):60456049.[18].[J].,2007,25(7):7882.[19].[J].,2009,7(3):102106.[20],.[J].,2009,9(2):8992,115.ThePredictabilityofShort-termTrafficFlowinDifferentTemporalScalesHUANGA2qiong,GUANWei(SchoolofTrafficandTransportation,BeijingJiaotongUniversity,Beijing100044,China)Abstract:Thepredictableintervaloftrafficflowisthekeytoachievereal2timetrafficcontrolandguidance.Inordertoinvestigatethepredictabilityoftrafficflowtimeseriesindifferentintervals,thispaperfocusesontrafficvolumetimeserieswiththeintervalsof2min,4min,6min,8min,10min,12min,14minand16minfromBeijing2nd2RingRoad.UsingRöSanalysismethodtocalculatetheHurstexponentofdifferentobservationscales,discoveringthatHurstexponentincreasesastheobservedscalesgetlargerinthesameday,anditincreasedasthesamplesizedecreasesinthesameobservationscales.Finally,usingapproximateentropymethod(ApEn)tocalculatethecomplexityofsequence,discoveringthecomplexityindexhadasimilarvariation,whichconfirmsthatthereisanegativecorrelationbetweentheobservationscalesandthestrengthofrandomnessofseries.Keywords:TrafficFlow;Predictability;RöSAnalysis;ApEn;ObservationScales082010