Hierarchical real-time train control in DC metro systems

Abstract

Metro systems or underground railways are the major means of mass transportation in metropolitan cities. To ensure the demand of passenger flow throughout a day, a reliable and regular metro service is highly desirable in life. Any delay or interruption of train service may bring a city to a standstill, which may induce a substantial economic loss and affect the public's daily activities. The main objective of this study is to develop a comprehensive train regulation and coordination system for metro systems, to optimise the train operation under the constraints of traffic conditions, tariff and computational demand. To achieve highly efficient and flexible train operation and control, a hierarchical train regulation and control system, HTRC, is introduced in this study. A top-down approach is adopted in the HTRC. Such a decentralised control allows manageable problem complexity and reasonable computational demand for real-time applications. The operational instructions are directed to trains through the three layers of control, that is, central train controller (CTC), regional train controller (RTC) and on-board train-based controller (ZBC). Run time and energy demand are allocated at different levels with respect to different operations constraints and requirements. Each level coordinates with the next level and enables the decision-making process within its own specific scope. Of the three levels of train control, TBC is adopted to control individual train operation with respect to given run-time at the bottom level. Classical and heuristic approaches are introduced to locate optimal coasting point(s), with the aid of a single train simulator, according to specified inter-station run times. Classical methods are preferred for the search of single coasting point in a typical inter-station run. Heuristic approach to locate multiple-coasting points, however, is more applicable for a long inter-station run with extreme track geometry for the sake of energy saving. RTC enables train movement coordination in a region to optimise train service in the second level of control. A region is defined as the section of track separating two successive substations, and the lengths of regions are fairly regular in a DC metro system. The optimal set of dwell times of trains at stations and run times for trains in successive inter-station runs (i.e., control actions) are devised by dynamic programming (DP) in this layer, under given traffic demand. Computational demand is further reduced with state grouping for real-time applications. An unexpected short period of large peak power demand on substations may be induced when a number of trains are accelerating at the same time. The power utility usually imposes an additional cost for the peak power demand when it exceeds a specified threshold. CTC thereby decides the appropriate sequence of headway changes to trains in successive regions to meet the passengers flow demand throughout a day. The attained transition in headway minimises the peak power demand on the supply system. Adjustments of dwell and run times of trains are not conducted in the CTC. Given CTC's relatively low complexity, exhaustive search and DP are employed in finding the optimal sequence of headway changes in this layer. From the simulation results in this study, the proposed HTRC is capable of providing the necessary operational instructions to trains in three layers for different constraints and requirements. The HTRC has shown the potential benefits of on-line train coordination and control.