Research On The Extraction And Designing Method Of Potential Routes For Urban Customized Bus

With the development of economy and society,urban expansion and hyper-motorization have brought a series of traffic-related problems,such as traffic congestion and environmental pollution to the city.In order to alleviate the contradiction between the demand for urban traffic motorization and various resource shortages,many cities have gradually adopted a combination of "sparse" and "block" measures,and public transport priority is one of the most important measures.However,the quality of traditional public transportation services is low,and it has been unable to meet people’s increasingly diversified,refined,and personalized travel needs.In this situation,customized bus(CB),an innovative demand-responsive public transportation service,has come into being and is developing rapidly.The purpose of CB system is to create a direct,high-quality,ride-sharing service using information technology tailored for mid-and long-distance commuting passengers,thereby attracting more car commuters to use public transportation.However,as a developing transportation mode,there is a lack of a mature and complete theoretical system to support the sustainable development of CB system,especially researches on potential passenger mining and route planning.Under this background,this dissertation focuses on the attractiveness,potential route feasibility and route design based on the analysis of CB system.The main research content includes the following aspects:(1)Aiming at different types of commuters,ordered Logit models for CB attractiveness are constructed.Travel mode choices of passengers are transfermed into attitudes to customized bus comparing with the current commute mode.Then transfer,uncertain or not transfer decision is made.Through carrying out the RP/SP joint questionnaire survey,the passenger travel demand characteristics are obtained and then analyzed from five aspects.On this basis,validating the models comprehensively considering the personal attributes,current travel mode attributes,the CB service attributes and the travel mode attributes using STATA 15.Then exploring the influencial factors of the transfer decision and analyzing their marginal effects.The results show that one person one seat and the familiarity with CB play an important role in promoting the attractiveness of the service,which provides operators with good development ideas for directional improvement of the attractiveness of CB.(2)Dividing the OD areas using trip chains of passengers.In view of the uneven distribution characteristics of bus stops,the stop isolation(SI)concept is proposed and introduced into the DBSCAN algorithm.SI-DBSCAN algorithm,in which the neighborhood radius changes with the degree of SI,is designed to divide the potential OD areas of CB.This method can avoid the situation that the coverage of some OD areas or the number of OD areas is too large when clustering according to the uniform neighborhood radius.A balance between the coverage and numbers of OD areas is finally achieved,so as to prepare for the study of route feasibility.(3)By analyzing the operation characteristics of CB routes,the "Flow-Frequency-Length(FFL)" sequence constraints were formulated.Based on the trip chains and the OD areas,considering the spatial-temporal characteristics,trip chains are clustered vertically and horizontally to meet the flow and frequency constraints using the designed ST-DBSCAN algorithm.Then a length linear constraint is added to extract potential CB routes.This content is the core of the research on the feasibility of CB routes,which is helpful for the operators to mine the actual passenger flow of the directional routes.(4)According to the distribution of CB travel demands and the dynamic adjustment characteristics of CB route design,a bi-level planning model that reflects the interaction between the operator and the passengers is constructed.Among them,the upper model is constructed from the perspective of the operator,with the objective functions of maximizing corporate profits,minimizing the total travel cost of passengers and minimizing the penalty of low service rate.The lower model determines the travel mode with the smallest general travel cost between CB and private cars from the perspective of passengers.The model is solved by an improved genetic algorithm.(5)Using the smart card data(SCD)of Beijing conventional bus to identify the CB potential passenger demand.Matching between the SCD and the basic information database and dentifying the passenger transfer behavior by limiting the transfer time threshold.The passenger trip chains are finally obtained to extract and design the potential lines of CB in Beijing.The result shows that the routes extracted by the route feasibility model covers about 80% of the opened routes and two CB routes are designed for the direction from Jinding Street to Dinghui Temple in morning peak hours.