Combined Problem Of Modal Split And Traffic Assignment Based On Bounded Rationality

As the principal part of urban transportation system, travelers’behavior has a great influence on the operation effect of the whole network. Traditional research on travelers’ behavior is generally based on an assumption that travelers are perfect rational aiming at minimum traffic impedance or maximum utility, and rarely takes travelers’bounded rationality into account. This paper focused on travel behavior, identified the parameter system of prospect theory with a questionnaire survey, analyzed the mode choice and route choice behavior under the framework of bounded rationality, established a combined model of modal split and traffic assignment, and finally studied characteristics of the combined model, travelers’reference-dependent effect and sensitivity of parameters.At the beginning, the concept of bounded rationality was clarified, the basic ideas of prospect theory and TODIM method, as well as relevant research and applications of these two theories were introduced. Then the difference of different functional expressions in prospect theory was discussed, the meaning of parameters in prospect theory was analyzed. According to risk level, travelers or trip circumstance was divided into 3 groups. Meanwhile, parameters of each group under route choice problem were calibrated, and the validity of this parameter system was verified. Then a modal split model and a stochastic user equilibrium model under bounded rationality were put forward in accordance with TODIM method and prospect theory, respectively. Further, a combined model based on bounded rationality was brought up. At last, Nguyen & Dupuis network was taken as a numerical sample to verify the validity of the combined model.Results show that combined model could reflect the influence of total demand on private-car chosen probability, they are negatively correlated; actual demand of private car and travelers’ recognition to different routes reveal clear reference-dependent effect while travelers’route choice behavior doesn’t; actual demand of private car decreases with the growth of parameter 6, increases with the growth of parameter κ; in the combined model θ should be valued in (0,6) and κ in (0,1).