| China has entered the rapid development stage of urbanization, the urbanization rate increase from 17.9%to 17.9%during the 35 years in 1978-2013. In the meantime, with the continuous development of national economy and the speeding up of urbanization, resident travel demand increase rapidly. Traffic congestion and other issues, such as traffic accidents and environmental pollution, has become the urgent problem in major cities. As an important part of urban transport, the development of urban public transport is getting faster and faster due to its special advantage. The problem of the development of public transport in cities was discussed in the state council executive meeting on October 10,2012. It is pointed out that the development of urban public transport in our country is too slow to adapt the development of social economy and people’s travel needs. The share rate of public transit mode is very low in most of the city. In order to ease traffic congestion, travel inconvenience, environmental pollution, etc., public transit must be in the first place of the development of urban transport. Thus, to analyze and understand the mechanism of transmission and public transit assignment has important practical significance to the priority development of public transport. Therefore, Chengdu conventional ground bus as the research object, the transit assignment problem under different emphases was analyzed. Based on this, the problem of day-to-day public transit system evolution considering learning behavior was studied. These problems are the focus of this paper.Firstly, the research background and research significance were introduced firstly, the weakness of the existing research and the problems needed to be studied further were pointed out based on the literature review. The research method, technical line and the main research content of this paper were concluded finally.And then, based on the basic description of public network, the generalized public traffic route which could accord with the route choice behavior of passenger was put out. By considering the section impedance and station impedance during the public traffic trip, the impedance function of public traffic route was set up, so as to determine the effective path. The concept associated with the transit assignment problem, such as common line and travel strategy, were explained and illustrated. The definition of transit assignment problem and the classification of the problem were expounded clearly. Due to the calculating example and simulation calculation in subsequent chapters of this paper are based on Chengdu public transit network or/and the actual operational data of Chengdu bus, thus, those data were introduced in the final section of this chapter.And then, the external factor of real-time information was considered to study the transit assignment problem. Chengdu bus electronic stop was taken as the research background, combined with the data of passenger route choice behavior and intention, using the method of qualitative and quantitative analysis, based on Logit model and Mixed Logit model, route choice models were established, using characteristic variables of various options and passengers’personal socio-economic attributes as explanatory variables. The results indicated that the differences of route choice behavior resulting from individual preferences can be reasonably interpreted by Mix-logit model. The most fundamental reason lies in the parameters need to be estimated of Mix-logit model can be subject to any form of distribution, which can effectively avoid the defects of other discrete choice model. Finally, the transit assignment problem was analyzed combining with the parameter estimation results, and the corresponding opinions and suggestions were puts forward.At the same time, the external factor of passengers’transfer behavior was considered to study the transit assignment problem. Based on the definition of the generalized public traffic route, the transit assignment problem can be considered as the process of passengers from the origin station and selecting the path at each transfer station according to a certain transfer probability and finally get to the terminal. Passengers’route choice process in accordance with "no aftereffect" of markov decision process was analyzed and proved. In addition, the state-action space and state transition probability involved in the process of transit travel was elaborated in detail. Passengers’route choice behavior based on Markov chain was analyzed, and the route selection probability resulted from transition probability and multinomial Logit model were proved to be consistent. Combined with the prevalence life-work mode of "live outside, work in downtown" in Chinese big cities, aiming at the public transit network of multi-source and single end, the corresponding algorithm is designed. For different bus network, regardless of whether passengers have to transfer, this method is applicable. Finally, part of public transit network of Chengdu as an example was used to verify the validity and the rationality of the theoretical analysis, and the possibility of application, the parameter sensitivity analysis was done based on the assignment result, and the corresponding opinions and suggestions were put forward.In addition, the internal factor of passengers’perception was considered to study the transit assignment problem. In detail, that is the passengers’decision based on the limited rational decision. The cumulative prospect theory analysis framework was introduced to analyze passengers’individual perception how to affect the route choice decisions, each route was subjective edited and evaluated, and a value function was established, combined with the decision weighting function to calculate the prospect value of each route, public transit user equilibrium under the perspective of prospect theory was put forward, and the equivalent mathematical programming model was established, and the solving algorithm was established. Finally, part of Chengdu public transit network was used to instance validation. Analysis result indicates that transit assignment considering passengers’perception can effectively integrated the average travel time and variance of the influence of passenger route choice decision, thus can be better able to describe and restore passenger route choice behavior, the results of the study has very strong theoretical guidance value on public transport planning and management, but also helps us understanding the complexity of urban public transport behavior, and can provide more scientific and more close to the actual decision support for planners and bus users. In addition, it can also be used as valid data support for traffic guidance.Further, the learning behavior caused by the excitation of the internal and external factors was considered to study the problem of day-to-day transit system evolution. Whether it’s real-time information, or transfer behavior, or individual perception, the public transit travel in real life is eventually a dynamic evolution process:in the time-varying traffic environment, passengers travel rely on past experience, reference the real-time information, to forecast future travel impedance, and adjust the time of departure and/or travel route, this process is known as passenger’s travel learning behavior. The experience weighted attraction learning model was used to model passengers’route choice based on learning behavior. Starting from the mechanisms for individual behavior, the evolution of the daily public transit system was studied. Combined with public transit network in Chengdu, the computer simulation method was used to interpret the evolution process of the public transit system to generalized user optimal equilibrium, and discusses the impact of external disturbance on system convergence or system equilibrium. Finally, comparative analysis result between the simulation results and the actual result indicates that passengers’travel in case of non-cooperation will eventually reach generalized user equilibrium affected by learning behavior, but the time required for equilibrium may vary due to the pace of learning. After reaching equilibrium, even suffered uncontrollable interference, the system will eventually reach a new equilibrium after a period of timey. Comparative analysis result indicates that the equilibrium can be effectively interpreted by transit assignment analysis based on EWAL model from the microscopic point of view. The real macro trend and distribution law of transit passenger were fully reproduced at the same time. Thus, passengers’decision-making behavior can be effectively explained and predicted.Finally, the main work of this paper was induced and summarized, the direction of the future may need to be researched was pointed out. |
PDF Full Text Request