Abstract
The Logit models based on random utility theory are generally applied as a description model of travel behaviour. On the other hand, fuzzy reasoning and neural network are known as useful approaches on the basis of human knowledge. Furthermore, various types of fuzzy neural networks (FNs) are proposed to integrate the advantages of fuzzy reasoning as well as neural network. The travel behaviour of multi route choice on urban network is formulated as a model in the study. The result of the practical route choice survey is used to establish the route choice model. Fuzzy neural network can be easily applied to describe the route choice behaviour. Fuzzv-neuro model and neural fuzzy model are selected among many different types of models. These techniques many provide the better estimation with high accuracy in terms of absolute error comparing to logit models. Lastly, the change of traffic can be analyzed by the model with assuming the implementation of route guidance or traffic information service.
