THE DEVELOPMENT OF MAXIMUM-ENTROPY ( ME ) ESTIMATION FOR CALIBRATING TRANSPORT DEMAND MODELS BASED ON LINK VOLUMES

The paper introduced that many problems in transport planning and management tasks require an Origin-Destination ( O-D ) matrix to represent the travel pattern, however the O-D matrix usually obtained from a large survey such as home or roadside interview tends to be costly, labour intensive and time disruptive to the trip makers. Therefore, more use should be made of easily available and low-cost traffic data. One possible way to represent the trip making behaviour within the study area is by applying a transport demand model,described as a function of one or more parameters, that estimates the number of trips made by each trip purpose during a period of time. The paper developed three estimation methods to calibrate such a model from traffic counts, namely: Non-Linear-Least-Squares ( NLLS ), Maximum-Likelihood ( ML ), and inference-Bayes ( IB ). The objective of the work is to develop an estimation method based on a Maximum-Entropy (ME) approach. These four estimation methods will be used as methods to estimate the parameters of various forms of transport demand model. The work so far concentrated on the estimation of two types of models. They are : Gravity ( GR ) and Gravity-Opportunity ( GO ) .