Automatic Incident Detection for Urban Expressways Based on Segment Traffic Flow Density

Abstract

Urban expressways play an important role in urban road networks. Although automatic incident detection (AID) methods have been studied for a long time, most of the existing AID algorithms are designed for freeways and do not explicitly consider the detection of incidents near ramps with frequent in and out weaving flows. Urban expressways usually have short spaced ramps, which makes the traffic flow characterizations quite different from the ones of freeways. In addition, expressways usually have heavy traffic flow, which makes it more difficult to distinguish incidents from congestion. This article presents an AID method for urban expressways, using loop detector data. Based on the geometric conditions and detector locations, the expressway is divided into short segments. The equivalent upstream and downstream traffic flow density difference is defined and calculated using the loop detector data. A detection logic based on the pattern of the density difference fluctuation is then proposed. The performance of the proposed algorithm was tested using real incident data from a corridor of the Shanghai expressway and compared with two classic AID algorithms. The results indicate that this method performs very well and is suitable for urban expressways.