抄録
The model examined was the Gravity (GR) model combined with the Multi-Nomial-Logit (MNL) model. Non-Linear-Least- Squares (NLLS) estimation methods were used to calibrate the parameter of the combined model. Three of the stages of this process (four steps model), trip distribution, modal split and traffic assignment, combine to estimate expected O-D demands, and as such, are of relevance to this research. Iterative solution algorithms, that are modifications of the Newton Raphson and Elimination Gauss-Jourdan techniques, are proposed to solve each of the model formulations.
The procedure described here assumes that the observed network is at equilibrium assignment condition. In other words, the observed link flows and travel times represent the (observed) equilibrium conditions. The objective of the solution approach is to find an O-D matrix such that when this matrix is assigned to the network, the resulting O-D travel times will be equal to the observed O-D travel times.
