This paper proposes and tests a framework for modelling travel choice behaviour, that is based on the concepts of fuzzy logic. The proposed framework uses fuzzy sets to represent perceptions and preferences, and assumes that travellers base their decisions on simple rules, rather than trying to maximise a complicated utility function. The fuzzy rule base that is used contains a set of fuzzy logic rules, relating an individual's preferences to his perceptions about the system attributes. An approximate reasoning process is used, which deduces a set of imprecise conclusions from imprecise premises. Travel choices are calculated, assuming knowledge of: (1) the structure of the rules in the rule base; (2) the membership functions of the fuzzy sets in the rule base; and (3) the rule weights. The model is calibrated, using a `neural network' approach to train the model, and an optimisation procedure that defines the travel choice variables. The proposed modelling framework is tested, by examining its ability to replicate simulated travel choices in specific simulation experiments; the results are encouraging.
Abstract