OASIcs.ATMOS.2023.6.pdf
- Filesize: 1.16 MB
- 17 pages
The European Train Control System (Hybrid) Level 3 (ETCS Hybrid Level 3) allows for introducing Virtual Subsections (VSS) into existing railway infrastructures. These VSS work similarly to blocks in conventional block signaling but do not require installation or maintenance of trackside train detection. This added flexibility can be used to adapt a given railway network’s (virtual) layout to the changing demands of new schedules. Automated methods are needed to properly use this flexibility and design such layouts on demand and avoid time-intensive manual labor. Recently, approaches inspired by design automation of electronic hardware have been proposed to address this need. But those methods - which are particularly well suited for inherently discrete problems in electronic design automation - have struggled with modeling continuous properties like train positions, time, and acceleration. This work proposes a Mixed Integer Linear Programming (MILP) formulation that, for the first time, can accurately model design problems for ETCS Hybrid Level 3 by including essential, continuous constraints, e.g., for train dynamics or braking curves. The formulation is designed to be flexible and extendable, allowing the user to include/exclude certain constraints or simplify the model as needed. By this, the user can decide whether he/she wants to quickly generate a less accurate solution or a more accurate one at the expense of higher runtimes - basically allowing him/her to trade-off accuracy and efficiency. A case study showcases the potential of the proposed approach and sketches examples to analyze which trade-offs are worthwhile and which simplifications can be safely made. The resulting tool and the benchmarks considered in this work are publicly available at https://github.com/cda-tum/mtct (as part of the Munich Train Control Toolkit, MTCT).
Feedback for Dagstuhl Publishing