Calibration of the combined trip distribution and assignment model for multiple user classes

A combined trip distribution and assignment model for multiple user classes has been formulated by Lam and Huang. The authors discuss here its calibration and show how to choose a suitable set of model parameters by using supplementary information. They show that each of the quantities-the mean trip cost of the entropy or the integral network cost, in the general cases-can be used individually to determine the model parameters, but the entropy would be the best choice. An algorithm is developed to obtain the parameters by using the entropy and has been tested on an example network. The relationships between the three quantities with respect to the parameters are also investigated by analyzing results of the example. However, the observed data for calculating the three quantities are usually not available or are costly to obtain. Other techniques of estimating the parameters by using alternative travel information which may be available or easy to obtain, such as partial observed Origin-Destination flows and traffic counts, are described, and a random calibration method is presented in this paper.