Evaluation of flow characteristics in signalised intersections is very important for a correct planning of traffic lights cycles and phase splits. A parameter involved in planning is the waiting time spent by vehicles at an intersection; different formulations of this quantity are present in technical papers, which derive however from analysis of field surveys. With the aim of evaluating the role of driver behaviour, field surveys are carried out in a research of more complex tasks and are used to work out a model of saturation flow. Data show the number of vehicles crossing an intersection during a green phase (which can be considered equal to the saturation flow) varying the length of the queue of waiting vehicles (for the following model only data concerning persistent queues are considered). The model of this phenomenon is carried out by using artificial neural network techniques and it furnishes very significant results (RMSE approx approx 6
展开▼