SUMO-Spurplan-5 – Deadlock processing for large-scale scenarios
SUMO-Spurplan-5 – Deadlock processing for large-scale scenarios
SUMO is used to evaluate operational processes and is being further developed in the Track Diagram Data 5 project in order to improve the realism of train trajectories and to simulate and manage network disruptions more effectively.
SUMO is DLR's microscopic traffic flow simulation, which can be used to model motorised private transport as well as other modes of transport such as rail transport. This software is used by DB AG to train and evaluate scheduling interventions in operational processes using AI methods. The high-dimensional solution space of this scheduling task with a large number of constraints and intervention options, which is modelled by SUMO, must be mapped as realistically as possible. As part of theSpurplandaten 5 project, SUMO is being expanded and improved in various areas for modelling rail traffic. The work includes improving the realism of braking and acceleration trajectories of the trains simulated in SUMO. The aim is to achieve greater fidelity to reality with regard to the simulated travelling speeds and therefore the journey times. To this end, simulated and measured (historical) train trajectories are analysed in depth. Furthermore, typical speeds are analysed according to timetable position (with/without delay). On this basis, future delays can be simulated more realistically and taken into account in the initialisation of subsequent trains as part of an operational simulation.
In addition, the necessary model extensions are being analysed and implemented to improve the performance of simulations of large-scale network disruptions. Up to now, the SUMO models have been able to analyse and subsequently resolve individual conflicts resulting from the concentration of trains on reduced infrastructure (e.g. bidirectional track usage, running on sight, shunting). By extending the model and upgrading the simulator interfaces, it should also be possible to dynamically handle network disruptions and changes in the future. This will enable significantly improved assistance in the management of large-scale network disruptions in the future.
