Citation

Khan AM. IET Intell. Transp. Syst. 2007; 1(2): 138-143.

Affiliation

Civil and Environmental Engineering, Carleton University, Ottawa, Ont. K1S 5B6, Canada.

Copyright

(Copyright © 2007, Institution of Engineering and Technology)

DOI

10.1049/iet-its:20060086

PMID

unavailable

Abstract

Accident avoidance is a very important part of enhancing road safety. The development of a queue-end warning system for highway work zones which automatically predicts queue-end location and alerts drivers so that rear-end collisions can be avoided has been described. In the absence of the widespread use of in-vehicle collision warning devices, dynamic messages about queue-end, displayed on portable variable message sign (PVMS) boards, are necessary for improving road safety. Although queues at selected locations along the road can be detected by the simple use of sensors, the changing nature of queue length would require numerous sensors to find the end of traffic queue. The reported queue-end warning system is based on a combination of sensors for detecting traffic and an artificial neural network (ANN) model-based algorithm for predicting queue-end location and issuing warning messages displayed on PVMS. Following the characterisation of work zones in terms of functional areas and geometrics, a microsimulator was calibrated and validated. Simulations were carried out next with traffic sensors and queue counter looped-in and the resulting data were used for training and validation of ANN models for queue length prediction. An automated information system was synthesised that integrates traffic sensors, ANN models, PVMS and potential links with other media. Selected results of ANN models illustrate their application in the queue-end warning system.