Citation

Thomas K, Dia H. IEE Proc. Intell. Transp. Syst. 2006; 153(3): -.

Affiliation

ITS Research Laboratory, School of Engineering, University of Queensland, Brisbane, QLD 4072, Australia

Copyright

(Copyright © 2006, Institution of Electrical Engineers)

DOI

unavailable

PMID

unavailable

Abstract

The high cost of congestion caused by incidents such as accidents and other events that reduce the capacity of roads has prompted a growing worldwide interest in developing efficient and effective incident management programs. The success of these programs will depend to a large extent on the development of reliable and efficient automated incident detection (AID) models. These algorithms analyse real-time data collected from traffic detection devices to determine whether an incident has occurred. This paper reviews some of the most widely used AID algorithms and discusses the refinement and further development of an AID model based on fractal dimension analysis. The development of this algorithm was based on the notion that traffic parameters upstream of incidents and bottlenecks show substantial irregular behaviour when compared with downstream conditions. Fractal dimension analysis was used to provide a measure of the irregularity in traffic parameters. A comparative evaluation of the performance of a number of AID algorithms that have been developed from a variety of theoretical foundations is also presented. The results, which were based on the same dataset of 100 field incidents, confirmed the superior performance of the neural network and fractal dimension models over the Comparative or California-type models.

Language: en