The Day-to-Day Dynamic Traffic Assignment Models With Considering Network Characteristics And Traveler Choice Behaviors

In order to relieve urban congestion problems, lots of traffic assignment models have been proposed to study the transport operation rules. Travelers’ daily route choice behavior can be described by day-to-day dynamic traffic assignment model, which can be divided into two stages. The first stage is travel cost estimating process, where travelers evaluate travel cost according to travel experiences and traffic information from government or other ways. The second stage is routing choice and flow updating, where travelers make routing decisions based on the estimation and path flow changes with the decisions. This dissertation studies the day-to-day dynamic traffic assignment models with considering network characteristics and traveler choice behaviors, which are traffic network with elastic demand, link-nonseparable cost function, dynamic route choice behaviors and the influence of cost prediction deviation. The main contents of the dissertation are as follows:First of all, elastic demand and link-nonseparable cost function are considered into the day-to-day dynamic model respectively, based on the work by Cantarella et al.(2003). Stability and bifurcations of equilibrium are analyzed by using nonlinear theory. Comparisons are made to study the effects of elastic demand and link-nonseparable cost function on the stability and dynamical patterns. It is found that the stability domain of equilibrium changes greatly and the sensitivity of parameters decreases when considering elastic demand. However, in the case of link-nonseparable cost function, stability domain decreases and the sensitivity of parameters increases. Dynamical evolution has been studied. The flow jumps at two different values (i.e. two-period behavior) when near flip bifurcation boundary and multi-period oscillation appears when near neimark-sacker curve.Furthermore, a day-to-day traffic assignment model with dynamic route choice behaviors is presented by considering the difference between forcasted travel cost and the expected value of the minimum cost of the OD pair. In the model, the weight of traveler, who reconsider the route choice, changes dynamically. Stability and bifurcations of equilibrium are analyzed by using nonlinear theory. Compared to the model with static route choice behaviors, it increases the stability domain and decreases the sensitivity of parameters in the presented model. Besides, it improves the convergence speed and decreases the oscillation amplitude in the evolution process. Finaly, the difference between forecasted travel cost and real travel cost, which is called as cost prediction deviation, is considered into the day-to-day dynamic traffic assignment model. It is proved that the greater the cost prediction deviation, the smaller the route choice probability. It increases the stability domain when the proportion of cost prediction deviation in a certain region and the stability domain of equilibrium of the system with considering only cost prediction deviation enlarges. Though it decreases the convergence speed, it reduces the cost prediction deviation and the oscillation amplitude of route choice probability greatly when unexpected traffic incidents occur.