ESTIMATION OF TIME-DEPENDENT OD FROM TRAFFIC COUNTS

Keywords : transportation systems management, ATIS, ATMs
Syafi’i, Setiono, S.J. Legowo*)
LPPM UNS, Penelitian, DP2M, Hibah Kompetitif Penelitian sesuai Prioritas Nasional, 2009

Budget and space limitations have forced changes in transportation planning paradigm from infrastructure development to optimize the capacity of road networks and transportation systems management. One of the efforts associated with it is the application of Intelligent Transportation System (ITS). Application of ITS in various subsystems such as the Advanced Traveler Information Systems (ATIS) and Advanced Traffic Management Systems (ATMs), is believed to be also able to reduce the level of congestion. The model developed in the implementation of ITS should be able to accommodate the traffic changes dynamically due to the delay/congestion or a specific event. Dynamic traffic changes as a keyword in ITS applications can be accommodated by applying a dynamic traffic assignment (dynamic traffic assignment, DTA). One important input in the DTA model is Origin Destination Matrix (Dynamic OD). Many methods have been developed to obtain OD matrix from data traffic, but generally the model used does not depend on time (time independent) or static OD. Weaknesses applications based on static OD is not able to accommodate changes in the distribution of congestion, travel time. If we consider the time dimension (time dependent) estimation problem becomes more complex. In addition to the distribution problem needs to travel the route, changes in currents as a function of time and space must also be considered. This study aims to propose a model approach to generate dynamic OD, apply in Surakarta road network and test the model validation. To solve the problem of estimating dynamic OD, the method used is Bi-level formulations. At the upper level, done adjustment OD. Dynamic traffic flow and traffic data is input at this point. While at the lower level is completed by mesoscopic traffic simulation.. Mesoscopic simulator used is DynaMIT (Dynamic Network Assignment for the Management of Information to Travelers).
The results of this study show that our method proposed approach with bi-level formulation has been successfully produced dynamic OD with a high level validation. This is shown by the comparison of flow between observations and model indicated by the value of the coefficient of determination (R2) greater than 0.9. In addition, the assignment of dynamic OD onto road networks have been successfully demonstrated dynamic changes of traffic on the road network. Flow dynamics is shown by changes in road network performance through the visualization JRNE (Java Road Network Editor).