Models And Algorithms For Urban Transit Dispatching Coordination Based On Time-Varying Dual Sources Of Data

The modern socio-economic development requires an effective coordination between all public transit modes to maximize the enhancement of its attractiveness. In comparison with the static dispatching coordination of the urban public transit system, the dynamic dispatching coordination can incorporate the consideration of the time-varying nature of the passenger demand as well as the uncertainty of public transit operations. As such, it has become a hot research issue in the field of the public transit. However, due to the complexity of the dispatching, the existing dynamic dispatching coordination of the public transit has mostly focused on the study of vehicle control strategies at the emergency. Moreover, in spite of being the key techniques for the dynamic dispatching coordination, the dynamic passenger origin-destination (OD) demand estimation and bus travel time prediction have been studied independently. Thus, there is a lack of effective application-oriented dispatching coordination models incorporating time-varying multiple sources of data.In this context, the research in this dissertation develops dispatching coordination models and algorithms for urban public transit based on the time-varying dual sources of data. First, the dissertation synthesizes state-of-the-art on both static and dynamic bus dispatching models, and the relevant problems. Then, it conducts the following research work:First, it conducts the study on the data quality control for the public transit dual sources of data. It develops public transit IC card data quality control models, the bus route OD sampling expansion models based on the distribution of riding distance, and the public transit GPS data quality control models towards the application to the dispatching.Second, with respect to the key techniques of the public transit dispatching coordination, it develops the method for identifying boarding and alighting stations for flat fare lines, as well as a short-term transit ridership forecasting model based on the state-spacial model. It further develops the bus travel time prediction models based on an identification of the state of the bus travel time.Third, it develops the public transit dispatching coordination model based on time-varying dual sources of data, as well as the model solution procedure based on the NSGA-Πalgorithm. The model is developed by incorporating the consideration of three objective functions, the supplier cost, the proportion of congested mileage, and the average waiting time for transferring passengers. In the modeling of the average waiting time for passengers to transfer from rail transit to buses, the arrival rate distributions of direct transfer passengers and non-direct transfer passengers are first analyzed. Then, an average waiting time model for transferring passengers is developed based on the lognormal and gamma distributions.Fourth, the proposed models and algorithms are applied to the case study of the dispatching coordination of Lishuiqiao station in Beijing. The results calculated using the proposed models are compared with the results obtained from the traditional method. It is demonstrated that, comparing with the results calculated from the traditional method, the average supplier-cost based on the proposed models increases by 10.50% while the the proportion of congested mileage decreases by 75.63%, and the average waiting time decreases by 20.05%. Therefore, the proposed model in this dissertation performs better in improving the overall quality of the transit service.Finally, it provides recommendations on the time intervals of the public transit dynamic OD, and the development of software and engineering applications of the proposed dispatching coordination models.