A Two-Steps Dynamic Demand Estimation Approach Sequentially Adjusting Generations and Distributions

Performances of the traffic models in reproducing congestion phenomena are strongly related to quality of the information about the demand. The problem of calibrating the demand for traffic assignment models is known as Demand Estimation Problem. This paper focuses on improving the reliability of the demand matrix, dealing with the problem of having a robust solution with respect to the input parameters and the used real measurements. Authors introduce a Two Steps procedure, which separates the problem in two sub-optimization problems. Through this procedure, authors correct sequentially generations and distributions in the demand matrix, reducing solution space size and the variance in the solutions of the calibration process. The proposed approach is then applied to a real network, the ring of Antwerp, and results are presented.