Research On Optimization Method Of Public Transportation Scheduling Based On Demand And Traffic State

Public transportation scheduling is an important component in public transportation operation and management; meanwhile it is one of the key techniques to provide convenient and comfortable public transportation services. Focusing on the current situations that the public transportation is usually scheduled manually and that the scheduling optimization is lacking of valid data support in China, the methods to extract the passenger demands and the traffic conditions along the bus route are presented using the public transportation AFC (Automatic Fare Collection) data. Moreover, the techniques to optimize the bus timetable and to dynamically scheduling are mainly studied.In terms of studying the methods to extract the passenger demands and the traffic conditions along the bus route, the AFC data characteristics are analyzed. Then, the applicability of the historical average method and the multiple linear regression method in extracting the passenger demand and the traffic conditions along the bus route is discussed. On this basis, an effective algorithm is proposed to extract passenger arrival rate and the link travel time to represent the public passenger demands and the traffic conditions along the bus route, respectively. The proposed algorithm is verified using real-world data. In terms of studying the method to optimize bus timetable, the problem of bus timetable optimization is abstracted as a nonlinear optimization problem. On this basis, the applicability of several intelligence algorithms in solving the nonlinear optimization problem is discussed. Thus, a genetic algorithm is presented to optimize the bus timetable based on the historical public transportation demands and the traffic conditions along bus route. The proposed algorithm optimizes the original bus timetable using the average passenger waiting time as the optimization objective and the bus frequency as the constraint condition. In terms of studying the method to dynamic bus scheduling, the corresponding functional requirements are firstly analyzed. Moreover, the applicability and detailed algorithms of two types of dynamic bus scheduling strategies, i.e. the bus short turning in peak periods and the deadheading, are studied.Both the proposed algorithms of the bus timetable optimization and dynamic bus scheduling are evaluated by simulation. The simulation results show that the proposed algorithm for optimizing the bus timetable can effectively reduce the mean passenger waiting time on the baseline of original bus timetable. The simulation results also show that the two types of dynamic bus scheduling strategies, i.e. the bus short turning in peak periods and the deadheading, can obvious improve the capacity and quality of bus services during the optimizing periods, while has little impact on the overall average passenger waiting time for the whole day.