ESTIMATION OF FREEWAY DYNAMIC ORIGIN-DESTINATION MATRICES: A NOVEL APPROACH

Abstract

This paper proposes a novel approach to estimate medium-to-long term freeway dynamic origin-destination (O-D) matrices. The proposed approach includes a two-stage prediction model with an integrated algorithm. The two-stage prediction model uses K-means algorithm to extract clusters of traffic patterns and then employs genetic programming to predict the traffic in each cluster. The integrated algorithm combines cell transmission model with extended Kalman filtering to estimate the arrival distributions and the O-D proportions. To demonstrate the applicability of the proposed approach, a field study of on-ramp traffic patterns on a freeway is examined. The results show that the proposed approach can accurately predict the traffic and satisfactorily estimate the O-D proportions along a freeway.