Merging behaviors are viewed as a key trigger in early-onset breakdown at urban expressway on-ramp bottlenecks. This paper proposes a dynamic Bayesian network (DBN) model to predict the various behaviors, which considers the impact of drivers’ history behavior choices on the current behavior decision. The two sites of U.S. Highway 101 (Hollywood Freeway) in Los Angeles, CA, (for short US101) and HongXu on-ramp bottleneck of Yan-an Expressway in Shanghai, China, (for short SHHX) are studied. 411 non-merging and 306 merging events at US101 and 525 non-merging and 328 merging events at SHHX are obtained. For the purpose of comparison, this paper investigates static Bayesian network (SBN) models only considering the current merging conditions. The results indicate that DBN models show better prediction accuracy than the SBN ones. The precision has respectively improved 3.56% and 3.66% for the two study sites.
Exploring Freeway Merging Behavior Using Dynamic Bayesian Network Models
International Conference on Transportation and Development 2018 ; 2018 ; Pittsburgh, Pennsylvania
2018-07-12
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Driver Behavior at Freeway-Ramp Merging Areas
Transportation Research Record | 2009
|Age-Based Simulation of Merging Behavior at Freeway Merging Ramps
Transportation Research Record | 2017
|On the Consistency of Freeway Macroscopic Merging Models
Transportation Research Record | 2014
|Gap stability and freeway merging
Tema Archiv | 1981
|Empirical Analysis of Merging Behavior at Freeway On-Ramp
Online Contents | 2010
|