Study On Flex-route Transit Dynamic Dispatching Model Based On Travel Behavior Analysis

As a new type of urban public transit services, the flex-route transit system is resilient and variable, which has unique advantages compared to the traditional public transit system under specific transportation demand conditions. The flex-route transit system has been widely used in European and American countries, while in China it is rare and needs further studies. Although, it is an inevitable trend to promote the flex-route transit for the development of urban public transit system with the acceleration of urban-rural integration process.From the angle of psychology, this paper models the travel behavior in the flex-route transit system and then focuses on the construction of the dynamic dispatching model and the design of the intelligent transportation system. The major content of the research are following:1. In the literature review at home and abroad, this paper summarizes the basic concept and different types of the flex-route transit system, which is aimed at finding the influence factors and building the simulation scenario.2. The travel behavior model in the flex-route transit system is built based on the improved Theory of Planned Behavior (TPB) with experience as an additional latent variable. Descriptive indicators of variables which have effects on the behavior are obtained by means of psychological measurement. A Structural Equation Model (SEM) is constructed after the reliability analysis and validity test of the questionnaire data with SPSS. With the model calibration conducted by AMOS, the relationship between variables and their effects on traffic behaviors are quantified, which reveals the key points of the research on dynamic dispatching model afterwards.3. First, in consideration of the constraints of time, space and capacity, this paper establishes the dynamic reservation screening model of the flex-route transit system based on passengers’perception. Second, a new combined forecasting model is established to predict the driving time based on the theory of Exponential Smoothing and RBF neural network. While the dwell time at fixed and flexible stops are calculated by probability theory. At last, the integrated dynamic scheduling model is formed after the route optimization.4. The structural relationships among different modules and the workflows of the flex-route transit system are designed based on related technology of intelligent transportation systems.A case study of the flex-route transit system is analyzed as a verification of the dispatching model and algorithm mentioned above. Simulation is carried out under the environment of C++ and MFC, which gives the bus travel time, travel routes and the outcome of reservation screening. The results indicate that this model is feasible. In consequence, this case provides a practical example to the application for the dynamic scheduling system of the flex-route transit.