Data assimilation based on a mesoscopic-LWR modeling framework and loop detector data : methodology and application on a large-scale network
Résumé
Traffic managers and operators need decision support systems able to provide online traffic flow monitoring and short-term traffic predictions on large-scale networks. Data assimilation techniques are employed to combine observed data and a traffic model. The paper proposes a comprehensive data assimilation framework based on a mesoscopic LWR model, which presents low computational times, is well-suited for network discontinuities, pro- vides individual vehicle tracking and can easily be coupled with any dynamic traffic assignment model. It also relies on state variables that require adjustments of the DA framework. Require- ments proposed by the paper are concerned with (i) the model numerical scheme, (ii) the traffic state transformation operators (iii) the update of the model. The proposed DA framework is first applied on simplified network. It validates the ability of the proposed framework to update and propagate traffic states accordingly. It is then applied to a real large-scale network. The results demonstrate its ability to monitor and forecast traffic conditions with online capabilities.