Freeway traffic stream modeling based on principal curves

The paper first analyses the importance of freeway traffic steam model, then focus on the macroscopic traffic stream model which reflect the relationship between the aggregative traffic variables: traffic density or occupancy, speed, and flow. Different from the conventional method, principal curves are used to modeling the traffic stream without the assumption of certain function form. Furthermore, we presume that the traffic variables interact with each other and would like to summarize the joint behavior of the traffic variables. After giving a brief description of principal curves and the algorithm is given, experiments in two sets of traffic data were carried out for the comparison of the model accuracy between the proposed model and the classical model proposed by Greenshields. Results show that the accuracy of the proposed model is better than that of the classical model. What is more important, it gives a non-parameter modeling method that may be used as the uniform method for traffic stream modeling as long as enough and accurate traffic data are obtained. The conclusion and further work are outlined in the last section of this paper.