Disaggregate Urban Residents Travel Destination Choice Behavior Model

Travel demand forecasting is one of the key techniques of urban transportationplanning. Due to the remarkable aggregate characteristic of TAZ (Traffic Analysis Zone),the most widely used traditional four-step model still has many limitations. Around theworld, as an alternative or supplementary of aggregate model, disaggregate behavior modelhas been rapidly shifting from research to application phase in the past two decades.Among previous disaggregate model researches, most the practice are used in tripproduction and mode choice. Only a few disaggregate model studies dealing with tripdistribution. This paper studies the destination choice behavior of travelers and tries to usedisaggregate model to provide a more reasonable and precisely analysis.The paper begins with a review of disaggregate behavior model based on stochasticutility maximization, mainly about the most widely used logit model, including thederivation, the limitation, the expand form and the estimation and validation of logit model.These discusses are very helpful to understand the theory and provide a good base forfurther study.Based on the previous studies, a new destination choice model based on disaggregatebehavior theory is built. In this model, the utility function is improved and optimized. Theauthor also discusses the interpretation and quantification methods of main explanatoryvariables based on trip survey data. Further more, the filtration method for explanatoryvariables is provide in this paper.As an application example of new model, a destination choice model for NHBW (NonHome Based Work Trip) of a TAZ in Anyang, a city in north china, is built. The parametersof this model is estimated and calibrated by trip survey data. The result indicates that, thegravitation factor and dynamic travel impedance based on activity, two variables newlyintroduced in this paper, remarkably improve the performance of model. At the same time, the case study also proves that socio-demographic characters of individual have impact onthe accuracy and explanatory ability of the model.The work which has been done in this paper will supply some reference for therelevant research and application in the future.