A Hybrid Discrete Choice Model With Heterogeneous Decision Rules On Attributes And Research On Stochastic User Equilibrium

Travel behavior research is an important subject in the field of transportation,which provides a solid theoretical basis for transportation system research.The traditional discrete choice model has made great contributions to the analysis of traffic demand,but it is usually assumed that the traveler is completely rational and maximizes the expected utility as the basis for decision-making.But the real traveler often can not be completely rational,not only pursuue the maximization utility or minimization impedance,and will try to avoid alternative options may be better and negative feelings of regret.The latest research shows that the traveler will have the problem of decision rules heterogeneity in the process of decision making,and usually can improve the fit degree of the model by considering multiple decision rules.So how to consider a variety of decision rules in a model is a question worthy of further study.This paper based on Regret Theory(RT)and Expected Utility Theory(EUT),rooted in the framework of Random Regret Minimization(RRM)and Random Utility Maximization(RUM).Allowing for heterogeneity in terms of decision-making rules on the attribute levels,presents a Generalized Random Modified Utility Maximization(G-RMUM)with better generality.The proposed model postulates that traveler may take two decision rules into consideration simultaneously,and different decision rules has different weight coefficient.In the case of selecting particular weight coefficients,RRM,RUM and other relevant existing models are special cases of G-RMUM.This paper first use the data of Swiss official which about travel mode choice,for the proposed model do parameter estimation and model validation by the discrete-choice software packages which namely Biogeme.By comparison the fit degree with other models,analysis results shows that G-RMUM has the best fit with data.In order to increase the trustworthiness of the model,we use the stated preference survey data about the travel route choice to test the multiple models again,the results show that compared with other models,the G-RMUM model has a greater improvement,thus confirming the model of general hybrid model can better characterize travelers’ choice behavior.In addition,this paper further introduces the G-RMUM model into traffic distribution research,redefines the stochastic user equilibrium condition,build a new assignment model,which gives the equivalent variational inequality model and corresponding algorithm.The influence of the relevant parameters on the distribution results was analyzed by numerical examples.The results show that the G-RMUM model can capture the heterogeneity of the traveler in the process of choice which at the attribute level.In the case of VOT take different values,G-RMUM model can also be a good description of link and path flow distribution in the road network.For the guidance of road network design,traffic demand forecasting provides a new way of thinking.