Development and application of incident detection techniques to improve incident management in freeway corridors

Abstract

Responding to the need for effective and reliable detection of freeway incidents, an essential element for improved traffic management and control in freeway corridors (Stephanedes and Chang, 1991), the authors initiated this research to investigate the performance limitations of conventional automatic incident detection systems and define the specifications for a new algorithmic logic that can lead to improved detection performance. The research initially focused on assessing the ultimate detection performance that can be accomplished with existing and new incident detection systems that use traffic data from presence detectors. A new algorithm was developed and tested against the major existing ones with promising results towards the development of a more-sophisticated detection structure. All tests employed a unified system of performance assessment (Stephanedes and Chassiakos, 1991), suitable for direct algorithm evaluation. The major accomplishments of this project are: * Review of current incident detection algorithms. * Testing major existing algorithm in the Twin Cities Freeway system. * Development of data preprocessing techniques to enhance the incident signal. * Development and testing of incident detection algorithms based on the data preprocessing.