F. Jin, G. Chen, J. Liu, T. Zhu

Pages: 123-132

Abstract
Aims: Vehicle flow estimation (VFE in short) is the base to make transportation plan and vehicle flow adjustment plan. It has been done by appointed dispatchers manually for a long time. Now the railway transportation information integration platform is under construction in China, which can provides information of trains, vehicles and locomotives almost at real time. It is urgent to study how to solve the VFE based on such a mass data. Scope and Methodology: The contents and artificial working process of vehicle flow estimation are analyzed, and a VFE model is built by gathering dispatchers’ experiences. The VFE model has a large number of 0-1 variables which are time-related. It is a kind of large scale optimal problem and is difficult to solve directly. The layered calculation process for railway network vehicle flow estimation is established based on decomposition coordination method. The calculation process divides the original estimation problem into three layers. They are: the coordination layer, the decomposition layer, and the execution layer. The railway network is divided into several subnets by decomposition method to lower the complexity. The coordination parameters are calculated first to coordinate the calculation results of lower layers. Decomposition layers can implement the parallel calculating which improves the efficiency of calculation. The calculation procedure is from top to bottom. The sub-problems corresponding to each layer are defined, and the calculation methods to solve sub-problems are recommended. Conclusions: Based on the layered calculation process method, a pilot VFE system is built and put into practice. The system can forecast changes of loaded vehicles and the time when vehicles go through demarcation station between railway bureaus. The estimation results are better than results made by dispatchers, especially in railway bureaus at the edge of the railway network. The practice shows that the layered calculation process is an efficient method to solve railway network vehicle flow estimation problem, and should be developed further.

Keywords: railway transport; vehicle flow estimation; model; layered calculation process