Analysis of Rear-End Conflicts in Urban Networks using Bayesian Networks

Abstract

Crash analysis and modeling studies have provided insightful information on crash-contributing factors and the methodologies utilized provide evidence that they could also be beneficial for conflict analysis, as traffic conflict data share similar traits with crash data. In this study, a Bayesian network (BN) is estimated to comprehensively analyze rear-end conflict likelihood in an urban network, using disaggregate vehicle-by-vehicle data and the time-to-collision indicator to identify conflicts. The variables imported in the BN include (i) individual driver characteristics (e.g., speed); (ii) traffic operational characteristics (e.g., volume); and (iii) general characteristics (e.g., weather conditions). The inference analyses of the BN conducted to quantify the contributions of the variables affecting rear-end conflict likelihood in the urban network showed that rear-end conflict likelihood could be increased when the involved vehicles are of a different type, when the speed of the following vehicle is higher than the speed of the leading vehicle, when the individual speed is high when the individual headway is small, with a higher coefficient of variation of speed values, when the type of intersection nearest to the measuring point was a priority intersection, when the carriageway was of dual design, and when it was rainy. It was also shown that rear-end conflict likelihood increases during congestion and free-flow traffic. The findings of this study could be further developed to provide a good understanding of contributing factors to possible crashes in the urban network.