| Improving the development of an urban travel service system with public transportation as the main body,thereby better meeting people's needs for travel is one of the important tasks to build China's strength in transportation.Under the innovative concept of Mobility as a Service(Maa S),transportation planners and managers have an urgent need for on-demand traffic guidance,especially for refined guidance theory and practical tools,gradually making refined route guidance a new research hotspot.Maa S describes a shift away from owning personal modes of transportation to using existing public transportations that are consumed as a service.Through integrated mobility solutions and one-stop service,the experience of public transport travelers can be improved.Especially,with the increasing travel distance of urban residents,the intelligent travel guidance of metro-bikeshare integration has become an important part of the realization of the Maa S concept in many cities.The integrated metro and bikeshare mode can not only meet long-distance travel needs of urban residents,but also improve the efficiency and quality of travel service dominated by public transport.However,the provision of separated travel information and connecting services between the two systems are still a problem in many cities.In order to meet the diversified and personalized travel needs of urban residents and ensure high-quality travel environment,an integrated metro-bikeshare service system needs to be established.This demands a further understanding of the whole process of individuals' metro-bikeshare trips and a grasping of user's route choice mechanism.However,current research on travel route optimization methods mostly focuses on a single mode of transportation and lacks of consideration of the network capacity problem which may lead to the overload of the whole system.Therefore,using multi-source data including metro-and bikeshare-smart card data,metro operation timetable and electronic map data collected from Nanjing,China,the integrated travel route of metro-bikeshare was optimized from the perspective of supply and demand.First,the concept of metro-bikeshare composite network was defined,and bikeshare network was integrated into the metro network by establishing a metro-bikeshare composite topology network.In the combined network,shared bike stations as the transfer nodes connecting to the metro network were linked to the nearby metro stations,so that travelers' origin/destination can be matched on the integrated network.On the metro network layer,a depth-first algorithm was used to search the effective route set between metro OD pairs.Using three-week metro smart card,metro passenger's routes were obtained from the effective route set by using a time-based GMM clustering algorithm.On the bikeshare network layer,shared bike OD pairs nearby the metro stations were selected using Arc GIS.Using the Dijkstra algorithm,the route reconstruction model of shared bike users was established based on bikeshare smart card data during the same time period.Second,since the aim of this study is to analyze the route characteristics of metrobikeshare trips,it is necessary to identify the metro-bikeshare trips from all travel records.Considering the difference between special cards and universal cards of public bikes,this thesis constructed separately(a)a universal bikeshare and metro smart card database and(b)a special bikeshare and metro smart card database,and divided the transfer behavior between metro and bikeshare into two modes: “Access-to-metro” and “Egress-from-metro”.The former refers to the user returning a public bike and entering the metro station,while the latter refers to the user exiting from the metro station and renting a public bike.By establishing the association rules between the two modes and database(a)and database(b),respectively,we proposed a metrobikeshare transfer behavior recognition method based on metro-and bikeshare-smart card data,and then used the confusion matrix to verify the effectiveness of the method.On this basis,we extracted the entire combined travel chain of users in the combined network,and analyzed the spatial and temporal characteristics of the metro routes,cycling routes and the combined metrobikeshare routes visually.Third,in order to investigate the combined travel route choice mechanism from the perspective of supply and demand,we introduced the concept of load status of the integrated metro and bikeshare network to reflect the relationship between the network capacity and the network load,and proposed the corresponding measurement indices of the load status.Consider that a large number of alternative combined travel routes for each user is not conducive to the study of route choice mechanism,we combined shared bike stations around metro station into a new "transfer node",and proposed a user-based search method to extract alternative combined travel routes for each user.As possible factors influencing the route choice of the combined mode,individual attribute,path scheme attribute and other attributes were extracted from multisource data.A mixed Logit model was used to examine the effects of these factors on the route choice for the transfer users of “Access-to-metro” and “Egress-from-metro”.The results show that the perception of train congestion and transfer perception of "Egress-from-metro" transfer users is much higher than that of "Access-to-metro" transfer users,and both of the transfer users are more likely to choose the combined route with sufficient bike facilities and high inventory rate.In addition,the user's gender,travel time and whether the user is a regular user also affect the choice of combined travel route.Fourth,in order to realize the real-time optimization of travel path for metro-bikeshare users,it is also necessary to master the real-time load status of the composite network.With analyzing the characteristics of travel demand in metro and bike sharing system,we employed a Kalman filter(KF)model and a hybrid model based on combining wavelet transform technique with support vector regression(WT-SVR)model to respectively estimate the dynamic load status of metro and bikeshare system.The proposed dynamic load status estimation models were examined by the data of in Nanjing,China.The performance of the models was evaluated by mean absolute error(MAE),mean absolute percent error(MAPE)and relative mean square error(RMSE),and also compared to other models.On the one hand,the results show that the KM model is more accurate than the traditional historical value estimation method.Especially in the off-peak period,the accuracy is improved by more than 15%.On the other hand,the results show that the WT-SVR model has better tracking ability and dynamic behavior than regular SVR and ANN models.The forecasting performance is remarkably improved,which proves that the suggested approach is feasible and applicable in load status estimation.Finally,the real-time optimization method of metro-bikeshare travel route was explored from the perspective of supply and demand balance.Taking "transfer nodes" as the optimization object,the maximum total utility of the combined traffic network as the optimization goal,a travel route optimization model based on comprehensive consideration of the system load status and optimization cost was constructed,and the branch and bound method was used to solve the optimal solution of user combined travel route.Taking the actual data of a working day of Nanjing composite traffic system as an example,the combined travel routes of users at every15-minute interval from 5:30 to 24:00 were optimized,and then the combined route optimization scheme based on the balance of supply and demand was proposed.The results show that by scheduling an average of 1.18 times for each "transfer node" in 74 time periods and optimizing 12.51% of the user group,the total travel utility of users can be increased by8.89%.After optimization,the number of times that the "transfer node" triggers the shared bike dispatching is significantly less than that before optimization,and the triggering period is more centralized,which not only reduces the dispatching cost,but also helps to improve the operation efficiency of the system.In addition,the optimized "transfer nodes" of users are relatively dispersed,which not only ensures the cycling reach range of users,but also contributes to the load balance of the passenger flow of the composite network.Further,according to the user attributes,the combined route optimization strategy considering the heterogeneity of the population was proposed. |
PDF Full Text Request