An Investigation on Computational Methods of Traffic Assignment in Road Networks

In this paper some computational methods of traffic assignment are proposed. At first solutions of traffic assignment by the equal travel times principle used nonlinear programmings are showed. They are a method to approximate by a quadratic programming, a method to use the SUMT transformation, a method to apply Rosen's gradient projection method and a method to apply the conjugate gradient projection method. Through a computational example the efficiency of them are compared. To decide route flows uniquely in traffic assignment by the equal travel times principle, a method requesting route flows such that the joint probability is a maximum under supposed a priori probabilities is proposed. In case of a traffic assignment in a large scale road network, it is significant to improve the efficiency of route searches. For this purpose a technique of minimum time route search by division of a network is proposed.