Firstly, this study presents a brief summary of the findings of the DRIVE II MARGOT project concerning journey time prediction in normal traffic conditions. Important findings are that: (1) good journey time prediction is essential for good routeing; and (2) that low penetration dynamic route guidance (DRG) scenarios need particular attention, so as to optimally combine historic data with scarce real-time data. Secondly, the study gives a more detailed description of a particular methodology for predicting link journey times in real-time incident conditions, for potential use in low penetration DRG scenarios. The methodology was developed within the DRIVE 11 MARGOT project. A ew modelling approach has been considered in which an `incident database' was compiled. This was done by using the CONTRAMI simulation tool applied to a range of network, traffic, and incident scenarios. The study develops generalised statistical models from the database. The aim is to predict the spread of congestion effects following an incident, and the required journey time modifications on incident link and on affected links. The `goodness' of fit of the models were evaluated by comparing the results of the developed models with that of the CONTRAMI simulation results. These models have demonstrated a reasonable predictive quality.
Abstract