The main objectives of the project are:
to develop and implement on the basis of the dynamic transport models developed in DRIVE 1 and elsewhere, practical dynamic traffic model system that can be used as an on-line decision support tool by traffic management; and
to apply and evaluate the model in a pilot project in an interurban motorway network, and to evaluate its performance.
Within DRIVE II, traffic forecasts have been recognised as a core issue in the area of interurban traffic management. DYNA aims to provide a practical model system to provide forecasts of traffic flows and travel times for interurban motorway networks. These forecasts are to be produced in real-time, for points in time up to 60 minutes ahead.
The DYNA system uses a combination of different forecasting models. A system architecture has been specified to define how these models should be integrated into a single forecasting system. The system consists of a prediction module, offers room for a control module, and contains two interfaces. One of the interfaces allows a Traffic Operator to access the system. Through the other interface the system is fed with measurements. The prediction module itself can be interfaced with a Control module, which is to be developed at a later stage.
The prediction system contains three different models: a model for on-line estimation of time-varying OD matrices (O/D), a time dynamic traffic assignment model (DTA) and a statistical traffic model (STM). The prediction module contains a separate module to merge the forecasts from the STM and the DTA. The STM module works in parallel with the OD and DTA modules; the system output consists of the merged results of the STM and DTA modules. Results from these modules will be merged where their prediction horizons overlap.
OD-estimation: Building on the results and recommendations of the DRIVE-1 ODIN project a sequential OD-estimation method was selected for use in the DYNA model. This approach combines a-priori information about OD patterns with the most recently observed traffic volumes on the entries and exits of the motorway network to produce up-to-date estimates of the time varying OD pattern. For application in a forecasting environment the estimated OD matrix needs to be extrapolated ahead in time to produce an expected near-future OD matrix. Although the final methodology to do this had not yet been established, an approach for OD-prediction has been developed.
Dynamic traffic assignment: The algorithm selected for use in DYNA was chosen following a review of available methods. The assignment works closely together with the OD estimation model, in order to guarantee consistency in operation. The assignment method uses a continuous packet approach, in combination with probabillstic path choice. It is anticipated that the dynamic assignment will have a prediction horizon of 15-60 minutes.
Statistical traffic model: Much more frequent forecasts will be generated by the statistical traffic model. It is hoped that this model will be able to operate at a 1 minute cycle, refreshing its forecasts every time that new measurement information becomes available. The methodology selected for the DYNA system is based upon original research and an expensive literature review. The proposed method uses a Kalman filter to continuously produce updated estimates of the state-space model and generate forecasts. It is suitable for use in real-time. The choice of its transition equations draws partly on the work done in DRIVE I project 'Christiane'. It is anticipated that the statistical traffic model will have a prediction horizon of 1 to 30 minutes.
Evaluation: pilot application
The real-time model system will be applied and tested on a motorway network in a pilot area round Rotterdam comprising some 200 kilometres of highway. The model will be fed with traffic speed and flow information measured using loop detectors from the Rijkswaterstaat monitoring system for the Dutch motorway network. The average inter-detector spacing will vary from 1 to 5 kilometres. Measurements from these detectors will be concentrated, in a control centre in real-time. The model system will first be calibrated, and then be applied to obtain short term prediction of speeds and flows for periods of 1 to 30 or 60 minutes ahead. This information, if sufficiently reliable, will be used, after completion of the project, to provide drivers with information about expected congestion and possible delays.
The testing schedule is as follows: first off-line tests will be performed involving the use of synthetic and recorded loop information to simulate on-line conditions; after this the models will be tested in an on-line setting. The results of the model predictions can be compared against actual realisations (speed and flows). The monitoring system (loops and processors) is expected to be operational in the first half of 1994.
Further Technical Specification
For the components of the proposed model system further technical documentation is available from the members of the Consortium specifying in more detail the technical approaches that are planned.
Specifically the DYNA project has set out to do the following:
to design and specify a model system to provide the required very short term interurban network traffic forecasts in real-time.
to provide an efficient implementation of the model system suitable for testing,
to test the model system using simulated data, and
to apply the model system using real data in a pilot project, and to evaluate its performance.
DYNA will provide forecasts of traffic flows and travel times for inter-urban motorway networks in real-time, with a prediction horizon of 1- 30 (or 60) minutes.
Since such forecasts form an essential part of any on-line traffic management system, the impact of the project will be in providing it's users with a practical decision support tool for traffic management.
Topic(s)Data not available
Call for proposalData not available
Funding SchemeData not available
2585 GJ Den Haag
See on map
See on map
See on map
3000 BA Rotterdam
See on map
See on map
LA1 4YF Lancaster
See on map
See on map