Avtomatsko zaznavanje izrednih dogodkov : diplomska naloga

Marot, Jernej (author), Maher, Tomaž (mentor), Rijavec, Robert (co-mentor)

Promet ima v vsaki državi znaten vpliv na razvoj gospodarstva, zato je pomembno, da se odvija s čim manj motnjami. Ena izmed motenj v prometnem toku so izredni dogodki, ki se jih ne da napovedati niti po času niti po lokaciji. Njihova posledica so zastoji, ki povzročajo zamude vseh vozil v prometnem toku. Da bi zmanjšali njihov vpliv, je treba izredne dogodke zaznati kar se da hitro. S hitrim zaznavanjem je možno zmanjšati trajanje zastoja in znižati možnost za nastanek sekundardnih nesreč. V nalogi je sistem za zaznavanje izrednih dogodkov razdeljen na senzorje, ki pridobivajo prometne podatke in algoritme, ki te podatke obdelujejo. Glede na mesto vgradnje, v ali ob cestišče in v vozilu, senzorje različne vrste najprej predstavimo, nato pa še med seboj primerjamo. Za boljše razumevanje prometnih podatkov so na začetku naloge podane osnove teorije prometnega toka. V nadaljevanju so glede na učinkovitost in zahtevnost izvedbe raziskani različni algoritmi za zaznavanje izrednih dogodkov. V praktičnem delu teoretični čas zaznave algoritma primerjamo z dejanskim časom zaznave za dano nesrečo glede na skupno zamudo vseh vozil, ki je tudi ovrednotena. V sklepu podamo koristi vpeljave algoritmov, in sicer na podlagi prihrankov pri vrednosti izgubljenega časa in priporočila za vgradnjo in nadaljnje delo na področju avtomatskega zaznavanja izrednih dogodkov.

• Language: Slovenian
• Keywords: gradbeništvo, UNI, diplomska dela, vodenje prometa, izredni dogodki, videodetekcija
• Work type: Undergraduate thesis
• Typology: 2.11 - Undergraduate Thesis
• Organization: FGG - Faculty of Civil and Geodetic Engineering
• Place of publishing: Ljubljana
• Publisher: [J. Marot]
• Year: 2014
• Number of pages: XIV, 88 str.
• PID: 20.500.12556/RUL-32499
• UDC: 656.1(043.2)
• COBISS.SI-ID: 6834017
• Publication date in RUL: 10.07.2015

Secondary language

• Language: English
• Title: Automatic incident detection

Abstract

Traffic has a significant impact on national economy. In order to be effective, traffic flow has to run smoothly. One of the things that can cause disturbances in traffic flow and delays are traffic incidents. They cause non-recurrent congestions that cannot be predicted by time nor space. To minimize their impact an incident has to be detected as soon as possible. Whatever the method of detection is, it should be quick enough to reduce consequences like duration of congestion and possibility of secondary accidents. Incident detection systems are determined on two level: data collection technologies and data processing algorithms. Thesis presents various sensor technologies that are categorized into two major categories; roadway-based and probe-based sensors. They are broadly discussed and compared against each other. For a better understanding of measured traffic variables the basics of traffic flow are explained at the begining. A variety of algorithms for the purpose of incident detection are investigated in terms of their performance and ease of implementation which is later used for the case study. Theoretical detection time of implemented algorithm is compared against the one from given accident in terms of all vehicles delay. Finally, the delay is monetarized by valueof-time and the benefits are calculated for use of dedicated incident detection algorithms.

• Keywords: graduation thesis, civil engineering, traffic management system, incident detection, videodetection