Day-to-Day Dynamics Of Traffic Flow Based On Prospect Theory

The increasing urbanization has prompted the population to move to the city,resulting in an increase in traffic demand,which in turn has caused imbalances in traffic supply and demand and caused traffic congestion.Therefore,it is necessary to find a strategy to alleviate the traffic congestion from the perspective of traffic supply and demand.However,relying only on accelerating the construction of transportation infrastructure cannot fundamentally solve the problem of traffic congestion.Thus,it is necessary to formulate corresponding management mechanisms to alleviate traffic congestion from the perspective of traffic demand.The traffic network will be non-equilibrium due to external factors.In this paper,the evolution process of the non-equilibrium state to the equilibrium state road network is taken as the research time domain,and the evolution of traffic flow caused by traveler's day-to-day travel behavior is analyzed to grasp the evolutionary laws of traffic flow.The laws provide theoretical basis and technical support for traffic demand forecasting and traffic guidance control in the long run,and then adjust the contradiction between traffic supply and demand and ease traffic congestion.Firstly,based on the existing related research,the prospect theory is introduced to describe traveler's day-to-day travel behavior.Combining the cognitive updating of travelers and the Markov learning model,the historical travel information is used to update the perception travel time,reference point and road network parameters.At the same time,the impact of memory rules is considered to study the day-to-day travel behavior.Combining the state transition equation and the transfer rule,a dispersed evolution model of path flow is established.The day-today variation process of road network traffic is simulated by a numerical example,and the influence of traffic transfer threshold and memory rules on traffic evolution is analyzed.Then,the day-to-day path traffic flow evolution is expanded to day-to-day departure time and path traffic flow evolution,and a spatiotemporal evolution model of the toll road network is established.In the modeling process,the traveler's homogeneity hypothesis is relaxed,the traveler is classified according to the value of time,and multiple reference points are set at the arrival time.The departure time flow transfer model is established with the maximum on-time arrival probability as the target,and the path flow transfer model is established with the maximum path prospect value as the target.The evolutions of the traffic flow of the heterogeneous traveler during the departure time,on the path and of each path during the departure time are analyzed.Finally,the impact of autonomous vehicle on the day-to-day evolution of road network traffic flow is considered.A hybrid flow day-to-day evolution model including traveler and autonomous vehicle is established,and the corresponding sub-models are the traveler flow daily transfer model and the automatic driving flow daily transfer model.With the maximum prospect value as the target,traveler flow is in the day-to-day evolution.With the minimum marginal impedance of the path as the target,automatic driving flow is in the day-to-day evolution.The day-to-day mixed traffic flow evolution of road network is simulated by a numerical example.The influence of the proportion of autonomous vehicles on the total time of the system,the impact of the traffic flow transfer threshold and the short-term events on the evolution of mixed traffic flow are analyzed.