Journal of Korean Society of TransportationISSN:1229-1366(Print) 2234-4217(Online)대한교통학회

This study proposed a method for accurate calibration of the volume delay function (VDF), which has the effect on the result of traffic assignment. As it is the last step of the traditional four step model for traffic demand prediction. Through the VDF, macroscopic traffic flow is allocated to the traffic network in the virtual environment using the travel time calculated at each link. The traffic volume calculated based on the travel time brings a calibration task that reduces the error compared to the actual traffic volume. Many developed countries are performing calibration task to reflect the traffic characteristics of each region through various optimization techniques, but Korea has been using uniform coefficient values ​​provided by KTDB. The determined value of the VDF correlation coefficient does not properly reflect the different traffic behaviors by region, and this leads to inaccurate future traffic demand. Therefore, in order to overcome this point, this study developed a methodology to calculate the optimal value according to the road hierarchy in the region using GPS trajectory data. The developed algorithm was applied to interrupted flow and continuous flow, and Root Mean Square Error (RMSE) was used for effect evaluation. As a result, the RMSE between the GPS-based speed and the simulation-based speed for interrupted flow was higher than continuous flow. It was 38.38 and 9.67, respectively. This is because factors, such as traffic signals and localized driving behavior are not considered in the interrupted flow. In the future, it is necessary considered diverse factors at interrupted flow to derive the value of the VDF correlation coefficient to get accurate travel speed.