Urban Traffic Network Analysis Based On Travel Time Reliability

People require more and more high travel quality than ever before, however, because of the influences of all kinds of factors on actual traffic network, traffic supply and demand are both uncertain. If travelers want to arrive at their destinations fast, safely, and smoothly, they need choose an appropriate route. Travelers have different risk attitudes, and it’s necessary to study their travel time reliability and route choice behavior. On the basis of the existing results, some studies have been carried out in the following:(1) Study on travel time reliability evaluation methods of road network① The backward cloud algorithm without certain degree information is amended. A three-time estimation method based on cloud theory is presented. Expected value, variance, and hyper entropy of link travel time are then computed through the proposed method. In addition, digital characters of route travel time are obtained, and according to them, the routes are ranked, and route, OD and road network travel time reliabilities are provided.② Dynamic travel time reliability is proposed from the view of road users and travel time distribution. A travel time reliability index is then proposed. Link and route travel time reliabilities are computed with the index, and the dynamic travel time reliability of urban road network is calculated and investigated.③ Link travel time reliability is divided into three states: reliable, intermediate, and unreliable. And then travel time reliability of failure-related link can be obtained. In the light of road structure, route and OD travel time reliability vector are computed. Finally, travel time reliabilities of a simple road network and a complex road network are studied through examples.(2) Traffic assignment and route choice model based on travel time reliability① Firstly, the definitions and the characteristics of optimistic and pessimistic values of a benefit function and a cost function are compared. Then, a min min model and a min max model of chance-constrained programming of travel time budget are established. The α-optimistic and the β-pessimistic values of link and route travel time are obtained. And according to the Hurwicz optimistic coefficient criterion, a comprehensive balance model is established. Finally, several travel time reliability indexes are proposed.② The dynamic generalized travel cost is defined as the weighted sum of travel time, travel time reliability and monetary cost. A stochastic user equilibrium model based on dynamic generalized travel cost is developed. The method of successive average(MSA) algorithm for the model is also proposed and sensitivity analyses are carried on some parameters. Then, a multiclass stochastic user equlibrium model is established and a solution algorithm is designed.③ The total demand is supposed to follow a normal distribution and the link capacity is assumed to follow a uniform distribution. On this basis, link and route travel time random distribution characteristics are derived. Three models including Stochastic User Equilibrium(SUE) model, Travel Time Reliability based Stochastic User Equilibrium(RSUE) model and Mean-Excess Stochastic User Equilibrium(MESUE) model are established under uncertain supply and uncertain demand, and the corresponding solving algorithms are designed.④ Cumulative Prospect Theory(CPT) with two reference points is generalized. Then, a unify method for estimating commuters’ reference points is presented. And a route choice model for commuters is established. At last, relationship between reliability and reference point and relationship between reliability and cumulative prospect value are studied on a test network.