This study investigates the characteristics of incident-induced traffic disturbances with an aim to gaining some insights for formulating an expressway incident detection algorithm. A total of 160 incidents were collected with the support of a video-based vehicle detector system along the Central Expressway (CTE) in Singapore. The results of statistical analyses indicated that the magnitudes of changes in traffic parameters due to an incident varied depending on the traffic conditions immediately before an incident occurrence, and that classified traffic occupancy and flow conditions can explain the variations in incident-induced traffic disturbances. The relationships among the traffic disturbances revealed that not all incidents induced significant changes in traffic speed, occupancy and flow simultaneously at the start of an incident. In general, the findings suggested that an incident detection algorithm would yield improved performance if more variables as well as information on the pre-incident traffic conditions were incorporated into its detection framework. (a).
Abstract