Astudy onincident detection modelapplying APIDmodel, fuzzy logic andtraffic pattern
Jin‐Woo Choi, Young-Kyu Yang · 2007
Thestudy hasdesigned anincident detection congestion thatcanbepredicted andeasedthrough early modelbyutilizing fuzzified APIDModelalong withtraffic traffic flowmanagement orbyproviding traffic information. pattern, forrealizing incident detection that ismoreefficient However, thenon-recurrent congestion causedbyroad andsuitable forroadenvironment withlamps. Traffic data works, incident, vehicle troubles ornatural calamity, occurs usedbythemodelaretraffic volume, occupancy andspeed collected for3monthsin5minute interval bytheloopdetectornon- uecificlocation and unspecifc tiTereore,cthe installed attheSeoul RingRoad. TheAPIDModeluses uppernon-recurrent congeston cannotpossibility bepredicted andlowerthreshold datadetected bytheloopdetector. andcanbeeased through accurate congestion detection and However, these datadonotreflect roadenvironment selected measures. Among these, a circumstance thatshows bythestudy. Therefore, thestudy hasfuzzified variables tosuitnon-recurrent congestion andisunpredictable iscalled an single siteloopdetector configuration, adopted MIN-MAX incident. Anincident isoneofamainfactor thatgives rise to Centroid Methodforinference methodandadopted Center of congestions onhighways orring roads andinterferes with Gravity Methodfordefuzzification method.Also, thestudy a . . hasdesigned anincident pattern byconsidering compressional norm wavetest, confirmation ofroadincident dissolution, dayofthe allowed capacity ofroads, causesairpollution, increases weekandvarying traffic occupancy shownbydifferent links.potential ofcausing secondary accidents, reduces road Inaddition, incident rate anddifferences inincident patternsafety andraises additional demands ofthesociety suchas values offinally modified APIDModelhadbeenusedtoderiveeconomiclosses(2). Therefore, the Intelligent theincident ratethreshold value forjudging occurrence ofan Transportation System requires an effective automatic incident. Theresult ofdetection rateandfalse alarmratetest, conducted tocomparatively analyze conventional APIDModel,incident detecton algorithm thatcanreduce effects ofan fuzzified APIDModelandproposed integration model, has incident whileresponding tosuchsituation quickly. shownimproved performance bythemodelsuggested bythe Diversified methods havebeendeveloped onincident study. detection algorithm andalgorithms suchasthecomparative algorithm, time-series algorithm, statistical algorithm, traffic