Development of real-time traffic state and travel time estimation for travel time predicton

Abstract

In the state of practice, traffic operators require dynamic traffic information in order to make decision to choose the suitable traffic operational plan that could mobilize traffic safer and with less travel delay. Data from conventional point detection system is widely available for determining travel time. Nonetheless, the penetration of detector stations is directly correlated to the accuracy of estimated link travel time. The theoretical as well as practical challenge is the improvement of accuracy of estimated travel time using available or lower level of data from point detection system. This study aimed to propose on-line microsimulation integrated Unscented Kalman Filter for traffic state and travel time estimation to be an alternated method for estimating link travel time on expressway which the developed microsimulation model was calibrated using genetic algorithm. Estimated travel time was further conducted to predict short-term travel time which uncomplicated and easily to implement were evaluated. Predicted travel time is benefit for traveler and traffic operator. The result from the study was shown that the proposed method is able to represent travel time on both uncongested and congested traffic conditions.