计算机专业英语学术论文

《计算机专业英语学术论文》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、ComputerTechnologyandApplication3(2012)361-367⋯MonitoringFreewayIncidentDetectionUsingaHotellingControlChartJoonseLim’一．YoungSeonJeongandYoungsulJeong1．PolytechnicInstituteofNewYorkUniversity，SixMetroTechCenter,Brooklyn．NewYo11201．USA2．GeorgetownPreparator

2、ySchool，10900RockvillePike，NorthBethesda，MD20852，USA3．DepartmentofIndustrialandSystemsEngineering,KhalifaUniversityofSeienee，TechnologyandResearch，AbuDhabi，UAE4．DepartmentofNaturalSciences，WashingtonBaptistUniversity，Annandale，VA22003，USAReceived：March12，2

3、012／Accepted：March31，2012／Published：May25，2012Abstract：Inreal-lifefreewaytransportationsystem，afewnumberofincidentobservation(veryrareevent)isavailablewhiletherearelargenumbersofnormalconditiondataset．Mostofresearchesonfreewayincidentdetectionhaveconsidere

4、dtheincidentdetectionproblemasclassificationone．However，becauseofinsuficiencyofincidentevents，mostofpreviousresearcheshaveutilizedsimulatedincidenteventstodevelopfreewayincidentdetectionmodels．Inordertoovercomethisdrawback．thispaperproposesawavelet-basedHo

5、tellingcontrolchartforfreewayincidentdetection．whichintegratesawavelettransformintoanabnorma1detectionmethod．Firstly，wavelettransformextractsusefu1featuresfromnoisyoriginaltraficobservations．1eadingtoreducethedimensionalityofinputvectors．Then．aHotellingcon

6、trolchartdescribesadecisionboundarywithonlynormaItraficobservationswiththeselectedfeaturesinthewaveletdomain．Unliketheexistingincidentdetectionalgorithms．whichrequirelotsofincidentobservationstoconstructincidentdetectionmodels，theproposedapproachcandecidea

7、decisionboundarygivenonlynormaltrainingobservations．TheproposedmethodisevaluatedincomparisonwithCaliforniaalgorithm．Minnesotaalgorithmandconventionalneuralnetworks．Theexperimentalresultspresentthattheproposedalgorithminthispaperisapromisingaltemativeforfre

8、ewayautomaticincidentdetections．Keywords：Freewayincident，incidentdetectionalgorithms，Hotellingcontrolchart，wavelettransforms，featureselection1．Intr0ductiOnThesealgorithmsincludeCaliforniaalgorithm[1]，Minnesot