Real-time dynamic modelling of transit system evolution

This paper deals with dynamic transit models for the real-time short-term predictions of transit system evolution to be used by transit agencies to support operations control and traveller information. Such predictions are carried out by using a network modelling approach based on schedule-based dynamic models, which are the core of a transit simulation framework called DYBUS. The first part of the paper gives an overview of models and methods to predict link travel times and vehicle loads, which are the main state variables to represent the transit system evolution; the latter presents the architecture and the main modelling components of the transit simulation framework