Train Timetable On-line Performance And Simulation System On Railway Passenger Dedicated Lines

By 2020, more than 16000km high-speed railway passenger dedicated lines (PDLs) will be operational in China to serve the great territory. It could be expected that PDLs will significantly enhance the railway capacity. Meanwhile, with the operation of PDLs, passenger trains and freight trains will be operated separately on the railway backbone lines. All the above changes provide beneficial conditions for improving train service quality, especially train service punctuality, so as to meet the continuously increasing requirement of passengers. It is well known that on-line performance of train timetable plays a vital role for train service punctuality. Based on the exisiting studies, this research mainly focuses on "on-line performance"-based train timetabling issues, and the main work in this thesis could be summarized as follows.(1) By analyzing the process of train timetabling and the execution of train timetable, we presented the concept of train timetable on-line performance, as well as related indices. In addition, we put forward the key factors that affect the on-line performance from both objective and subjective perspectives. Then we provided a conceptual model to describe the problem addressed in this thesis based on robust stochastic-project scheduling theory and methodology.(2) Aiming at slack time allocation which is one of the most vital factors that affect on-line performance of train timetable, we provided the definition and types of slack time, and analyzed the effect of slack time on train timetable on-line performance. After that, we analyzed the traditional methods used to allocate slack time in train timetable. Then we proposed a stochastic expected value model to improve the on-line performance of timetable and reduce total travel time on the basis of considering interactions among various slack times. We put forward the conceptual framework and detailed mathematical formulations of this model. At the same time, we provided a hybrid solving algorithm combining improved genetic algorithm and branch and bound method. Furthermore, we took allocating train slack time in a timetable on WuHan-GuangZhou PDL as an example to test the model and algorithm. We tested convergence of the algorithm and time complexity as well as space complexity. Finally, we presented the optimizing process, as well as the results of timetable execution from both train and station perspectives.(3) Aiming at train path layout (TPL) of timetable which is also one of the most vital factors that affect on-line performance of train timetable, we provided the concept of train timetable structure and corresponding index-heterogeneity of train timetable to assess TPL, by assuming that the train types and numbers are the same. Then we put forward computational formulas of heterogeneity, relationship between heterogeneity and train speed difference, as well as slack time. Meanwhile, we put forward a description model of the relationship between heterogeneity of train timetable and delay propagation. Furthermore, we designed three typical timetable cases corresponding to average-layout, partial aggregation-layout and aggregation-layout modes of train paths, also we designed experimental schemes, then we got some conclusions on the quantative relationship between on-line performance and heterogeneity based on a number of "train dispatching in the loop" simulation experiments. At last, we discussed three measures which can be used to optimize the TPL of train timetable and reduce the heterogeneity to improve the on-line performance of timetable. We also presented a method to find the key train(s) from a cluster of trains both from train timetabling and train rescheduling perspectives based on simulation experiments.(4) Aiming at capacity utilization of train timetable which is another one of the most vital factors that affect on-line performance of train timetable, we proposed a train timetabling model "feasible solution"-based method to assess section capacity and put forward a method to assess section capacity by simulating human behavior based on consideration of characteristics of PDLs. In addition, we presented a method to calculate capacity utilization level. Meanwhile, we put forward a description model of the relationship between on-line performance of train timetable and capacity utilization. After that, we got the section capacity of three typical timetables corresponding to average-layout, partial aggregation-layout and aggregation-layout modes of train paths by using the above method. Then we designed an experimental scheme, and got the quantative relationship between on-line performance and capacity utilization of the three different timetables based on a number of "train dispatching in the loop" simulation experiments.(5) "Train dispatching in the loop" simulation system is a very important platform for the studies in this thesis. Based on the analysis of train running and dispatching process, we proposed a simulation modeling method for train running. The modeling method includes formalized description and an agent-based train running simulation model for simulating train running under different pattern signaling systems. We also put forward the methodology, function structure and module structure of this system, as well as the fundamental rules for designing disturbance scheme and methods for generating scenarios. It is generally accepted that simulation initialization is one of the key factors that affect train running simulation precision. For this issue, we proposed a new accurate simulation initialization method based on train performance calculation and rules-reasoning. Finally, we illustrated the main graphical user interfaces of this system.