Short-term Prediction of Run-occupancy in Transit Information System

Short-term Prediction of Run-occupancy in Transit Information System


A Nuzzolo, U Crisalli, P Coppola, and L Rosati University of Rome "Tor Vergata" II, IT



ITS (Intelligent Transport Systems) is the "umbrella" term which usually encompasses those innovative technologies from Telecommunications and Informatics (i.e. Telematics technologies) applied to transportation. Examples are given by the Advanced Traveller Information Systems (ATIS), i.e. advanced systems aiming at providing travellers with information on network performances in the attempt to facilitate their travel choices, and Advanced Traffic Management Systems (ATMS) aiming atmonitoring and managing real-time the components of the networks. In transit networks, typically an ITS can detect the position of vehicles on the network and, in some cases, also the number of passenger alighted and boarded at stops. This information allows to deploying real-time information on the waiting times and the degree of occupancy of upcoming rides at the stops. The latter information can affect to great extent traveller?s choices, especially in congested transit systems, where travellers may choose to skip overloaded rides and wait for less crowded ones by trading off between longer waiting-time and higher on-board comfort.

In this paper a system of models aiming to predict both arrival-times and ride-occupancy at terminals oftransit networks with ITS, is presented. Predictions are based on data gathered real-time by means of a monitoring system, which detects the position of a given number of vehicles on the network and counts the number of travellers boarded and alighted at a number of predefined stops. The overall modelling framework consists of:

* a diachronic transit network (Nuzzolo and Russo, 1994) whose temporal co-ordinates are updated real-time, based on the information on vehicle position;

* a time-varying simultaneous O-D matrices estimation procedure (Cascetta et al., 1993), based on observed number of passenger boarded and alighted from vehicles at stops;

* a path choice model based on Random Utility Theory, simulating how information can affect passenger choices at stops;

* a within-day assignment procedure (Nuzzolo et al., 2001), following a schedule-based dynamic approach, estimating the loads on each ride of the transit system at any time of the reference period.

The results of preliminary applications to realistic test network area are discussed in the paper.


Association for European Transport