A mesoscopic whole link continuous vehicle bunch model for multiclass traffic flow simulation on motorway networks

摘要

Modeling of heterogeneous driver behaviour is vital to understanding of dynamic traffic phenomenon taking place on motorway networks. In this research, we present a mesoscopic whole link continuous vehicle bunch model for multiclass traffic flow simulation on motorway networks. Two main attributes of traffic flow classification have been used are: (i) vehicle type, specifying in turn a vehicle length and, together with type of a preceding vehicle, time headway; and, (ii) desired speed, defining together with the speeds of the neighbouring vehicles, the vehicle acceleration/deceleration mode. It is assumed that vehicles in uncongested to moderate congested flow move in bunches dividing the drivers into the two main groups: (i) independent “free” drivers which usually manifest themselves as leaders of bunches; and, (ii) followers, or drivers which adapt their speed to the leader’s speed and follow each other at constrained headways specified by predecessor/successor pairs. The model proposes a solution to arbitrary traffic queries involving a motion in bunches having various speed and size by assuming the rate of driver arrivals follows semi-Poisson distribution and proportion of free drivers is predefined. The solution, assuming limited overtaking possibilities for all drivers, involves formation of longer queue behind bunches moving with slower speed and transformation of some of the “leaders” into “followers” because of adjustment their speed to the speed of the preceding slow-moving bunches. The present solution considers both stochastic and deterministic features of traffic flow and, therefore, may be easily extended to a specific uncertainty level.