Influence of variable message signs (VMS) on flow characteristics, and on drive behaviour, is an important topic till now not completely investigated. A method already applied to analyse flow in motorway is extended to insert also the effects of VMS on traffic flow. The models are worked out by means of using multi-layered feedforward neural networks. The first analyses the density flow relationship and the second the distribution of flow by lane. Input variables are density, total flow, speed, brightness, clearance, meteorological conditions, percentage of heavy vehicles and presence of messages on VMS. Output variables are flows by lane. Speed, flow and density values, necessary to ask the second models, are obtained from the first one. Different scenarios are prepared to ask the model varying percentages of heavy vehicles, meteorological conditions, brightness, clearance and presence of VMS messages. Results show a great influence of VMS on capacity curve but not very significant on lane occupation. (A)
Abstract