Estimation of a time dependent origin-destination matrix for congested highway networks

This paper presents a time-dependent origin-destination (O-D) matrix estimation procedure embedded with a dynamic traffic assignment model, in which the predictive dynamic user optimal conditions in congested networks are maintained. Two solution algorithms are proposed, namely: an iterative (ITR) scheme and a method of successive averages (MSA) scheme. It is found that the MSA scheme outperforms the ITR scheme. As a prior O-D matrix is an important input for the problem, its quality is essential for the reliability of the matrix estimation procedure. Empirical constraints are set in relation to the quality of the prior O-D matrix for the estimation procedure. Numerical examples are used to demonstrate the effectiveness of the proposed methodology.