Loading...

Design of an Algorithm for Automatic Incident Detection by Using Loop Detector Data Compatible to Traffic Conditions in Iran

Asadabadi, Ali | 2012

627 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 43787 (09)
  4. University: Sharif University of Technology
  5. Department: Civil Engineering
  6. Advisor(s): Vaziri, Manouchehr
  7. Abstract:
  8. Incidents are one of the main sources of delays and they have significant influences on road safety and congestions, air pollution and trip costs. In the time of an incident occurrence, minimizing the response time is essential for several reasons. The most important issue is providing aid for the injured. The faster the persons who are severely injured receive aid, the higher their chances are to be rescued. Also, if the rescue crews get to the incident scene faster the clearance procedure would be accomplished faster and this would result in fewer disturbances in traffic stream and lowering the probability of secondary incidents occurrence. Automatic Incident Detection, AID, has been noticed in recent decades as a mean for faster detection of incidents. Despite the progresses that have been achieved in the both technology and algorithms of Automatic Incident Detection in recent decades around the world, because of the lacking infrastructures in Iran, there has been no attempts for development and implementation of such systems in Iran road control and management centers. By the advancements in recent years in implementing loop detectors and online transcending of their information to road control centers, the required infrastructures for Automatic Incident Detection is being completed. The goal of this study is gathering information about successful algorithms which were tested and developed in other parts of the world, choosing the algorithms compatible to the conditions of available data and traffic conditions of Iran roads, then calibrating the algorithms using the gathered local data, testing the developed algorithms and finally introducing algorithms which are compatible to traffic conditions in Iran. Gathered data which belongs to Shiraz-Marvdasht freeway is consisted of two main categories which are occurred incident information and output data of loop detectors. Considering the characteristics of the gathered data for this study, algorithms which use the information of an isolated single station were designed. Algorithms were designed in two different categories; first group for detecting incidents in a station upstream of an incident and the second group of algorithms for detecting incidents in a station downstream of an incident. Finally, the chosen algorithms were tested on a part of the information which were kept aside for final validation of models. Neural Network algorithms, in both upstream and downstream categories, showed good results. Also Logit and Discriminant algorithms operated well in upstream stations. Logit algorithms did well in downstream stations as well
  9. Keywords:
  10. Automatic Incident Detection ; Traffic Condition in Iran ; Loop Detector

 Digital Object List

 Bookmark

No TOC