Themen für studentische Arbeiten

Rail rescheduling is essential for maintaining efficient operations, particularly when disruptions occur due to delays or unforeseen events. Traditional rescheduling approaches often require manual adjustments, which can be slow and inefficient. This thesis proposes to investigate the use of machine learning (ML) techniques, specifically neural networks (NNs), to automate and optimize rail rescheduling. The focus will be on applying end-to-end learning, where a neural network is trained to directly map raw input data (such as train schedules, delays, and infrastructure constraints) to optimized rescheduled outputs without the need for manual intervention at intermediate steps. End-to-end learning allows the model to learn from historical data (or optimization models) and improve its rescheduling decisions autonomously, enhancing both efficiency and adaptability.

End-to-end learning refers to a machine learning approach where a model learns to map raw input data directly to the desired output through a single, unified process, without requiring intermediate features to be engineered by humans. In the context of rail rescheduling, this would mean inputting raw train schedule and delay data into a neural network, which would then output optimized rescheduled timetables. This contrasts with traditional methods that often rely on manually defined features and heuristics.

Further information: Thesis proposal - Rail Rescheduling Using End-to-End Learning

Contact:  Prof. Dr. Nikola Bešinović, Prof. Dr. rer. pol. Pascal Kerschke

Railway entities, including Infrastructure Managers (IMs) and Railway Undertakings (RUs), play a crucial role in ensuring continuous and reliable transport services. However, they face increasing challenges from extreme weather events, cyber threats, capacity bottlenecks, and supply chain disruptions, all of which can severely impact operations. The European Commission’s Critical Entities Resilience (CER) Directive underscores the need for enhanced resilience strategies for critical entities, including railways. Despite this, there is no standardized framework for systematically assessing and improving the resilience of these entities.

This thesis aims to develop a conceptual framework that allows railway entities to systematically evaluate their resilience, identify vulnerabilities, and implement effective improvement strategies. By bridging existing gaps in resilience assessment, the proposed framework will support railway entities in adapting to emerging threats and ensuring long-term operational stability.

Further information: Thesis proposal - Developing an Assessment Framework for Rail Entities’ Resilience

Contact:  Prof. Dr. Nikola Bešinović

Single Wagon Load networks play a crucial role in freight transportation by enabling flexible and cost-effective logistics solutions for small-to-medium shipments. However, their decentralized nature and interdependencies make them vulnerable to disruptions, whether caused by accidents, operational inefficiencies, or deliberate attacks. Understanding and mitigating these vulnerabilities is vital to ensure reliable service and maintain economic stability in industries dependent on rail freight.

This research explores critical infrastructure in Single Wagon Load (SWL) networks, focusing on identifying key nodes (yards) and/or links that are vital for operational efficiency. The research addresses system vulnerabilities to disruptions and develops strategies to mitigate their impacts. A generic SWL demand will be created to capture operational variability. By extending existing disruption management models, the study employs interdiction modelling to simulate targeted disruptions and design effective response strategies. Network performance will be evaluated over multi-day or weekly timeframes to reflect realistic operations and recovery dynamics.

Further information: Thesis proposal - Critical Infrastructure Analysis in Single Wagon Load (SWL) networks

Contact: Prof. Nikola Bešinović, Daniel Haalboom, M.Sc.

- in Bearbeitung -

Zur Gestaltung und Bemessung von Gleisgruppen können unterschiedliche eisenbahn­betriebs­wissen­schaftliche Ansätze und Verfahren genutzt werden. Neben bestehenden Verfahren auf Basis der Bedienungstheorie und der Simulation werden aktuell auch neuartige Modelle und Verfahren ent­wickelt, welche auf Grundlage der mathematischen Optimierung arbeiten. Mit den einzelnen Verfahren können unterschiedliche Kenngrößen bestimmt werden, die nur bedingt vergleichbar sind und für die nur teilweise bereits wissenschaftlich fundierte Qualitätsmaßstäbe existieren. Es ist daher zu untersuchen, inwieweit verallgemeinerbare Abhängigkeiten zwischen den verschiedenen Kenngrößen bestehen, welche Kenngrößen und Qualitätsmaßstäbe für die Einführung neuartiger Verfahren zweckmäßig erscheinen und wie man für diese belastbare Grenzwerte bestimmen kann.

Kontakt: Dr.-Ing. Jan Eisold Tel: 0351 463-42390   

- in Bearbeitung -

Die Spedition Helrom betreibt mit ihrem innovativen Umschlagsystem erfolgreich verschiedene Linien aus Deutschland in Richtung Österreich, Ungarn und Italien. Nun sollen Helromverkehre in Richtung Polen ausgeweitet werden. Um eine hohe Auslastung zu generieren, sollen die Züge mit einem Zwischenstopp im Bereich Halle / Leipzig / Sachsen aufgebaut werden. Bei diesem Zwischenstopp soll ein Teil der LKW-Trailer getauscht (umgeschlagen) werden, während der übrige Teil weiter bis zum Zielbahnhof verkehrt. Ziel der Arbeit ist es, anhand der vorhandenen Randbedingungen zunächst eine Empfehlung für optimale Standorte für einen solchen Zwischenstopp zu erarbeiten. Für eine Vorzugsvariante sind zudem die bauliche Gestaltung und die Anbindung des Umschlagbereichs an das Streckennetz der DB InfraGO AG abzuleiten sowie die Betriebs- und Umschlagprozesse zu gestalten.

Kontakt: Dr.-Ing. Jan Eisold Tel: 0351 463-42390   

The concept of “flying cars” originated in the early 1900s and has recently re-emerged in public consciousness as urban air mobility (UAM). UAM involves the use of electric vertical take-off and landing (eVTOL) vehicles to provide air transportation services for passengers and goods. These services aim to reduce commuting times for long-distance travel in highly congested areas by utilizing eVTOLs.

The emergence of UAM could significantly and complexly impact existing railway passenger flows. On one hand, UAM could compete with long-distance railway routes due to its high speed and comfort. On the other hand, UAM’s accessibility is limited by infrastructure constraints. Integrating railway stations with UAM vertiports and using railway systems for access and egress to UAM services have been widely discussed. Understanding the interaction between UAM and railway systems can provide valuable insights for decision-makers regarding infrastructure planning, operational scheduling, and policy regulation.

This project aims to develop an integration model of UAM and railway systems to help decision-makers understand their interactions within an integrated system. The model includes supply and demand components:

• Supply Modeling: Develop a new network representation for the integrated system and analyze passenger costs within the network. The model is capable of modelling the service supply of UAM and railway separately and the connections between them. 
• Demand Modeling: Propose a passenger assignment model for the integrated network that considers different passenger archetypes. The model takes the origin-destination matrix as input and assigns the passengers to various UAM and railway services.

More information: Thesis proposal - How does urban air mobility affect railway systems? An integration model and policy analysis

Train delays and cancellations occur when disruptive events create an imbalance between system capacity and demand. For example, in the Netherlands, internal factors causing failures (e.g. infrastructure, vehicle) take up to 70% of all disruptions in the network. On average, about 14 of such disruptions occur every day. Some of these disruptions remain even unnoticed by passengers; however, others generate problems spreading all over the network causing many cancelled and heavily delayed trains leading to great dissatisfaction of passengers.

© BSR

Therefore, the question is: how can we predict performance of the system during disruptions? The aim of this project is to study resilience curves (as depicted above) in railway networks and capture the magnitude and spatial impact of delays and cancellations. By using historical railway traffic data, we want to identify representative resilience curves and uncover the interaction between disruption and recovery. This will enable us to predict the future behavior of the system once a disruption occurs. Finally, such prediction system can help railway operators in better estimating and anticipating impacts of future disruptions and thus define best mitigation measures.

More Information: Thesis proposal - Modelling and predicting dynamics of disruption and recovery in railway networks

Contact: Prof. Nikola Bešinović

Power peaks are an undesirable phenomenon occurring in railway networks when multiple electric trains require large amount of power simultaneously, for instance, during acceleration. This phenomenon puts too much pressure on the power grid, which in the worst cases it can result into a blackout, and hence it represents a relevant concern for operators (Regueiro Sánchez, 2021). Furthermore, the high fluctuations in power consumption over time have a significant direct impact on operation costs, even though power peaks are generally very short in time (Albrecht, 2014). Reducing energy consumption is anyways a top priority in sustainability policies in many countries.

One solution for this is fine-tuning timetables to minimize power peaks. Nevertheless, the benefits of adjusted timetables can be lost in situations with train delays in the network. In this work, the goal is to develop a new approach to mitigate anticipated power peaks in real-time by means of train control measures, i.e. traction power limitation and departure time shift, combined with real-time rescheduling.

The goal of this project is to develop a pure optimization approach for the problem of mitigating power peaks in railway networks using train control measures in real-time, possibly including train delays. This optimization approach will be made in the form of a mixed-integer linear program. We consider power limitation and departure time shift as possible train control measures. The problem minimizes the total induced delay while capturing all relevant constraints that model feasible railway traffic (block sections, single and double track sections, technical headways between trains, train conflicts, etc.) (Pachl, 2014), for which we need to use detailed infrastructure models (Radtke, 2014). The approach will be tested in a case study consisting of the line between Giubiasco and Locarno (Canton Ticino) of the Swiss Federal Railways (SBB).

Further information: Thesis proposal - Optimization Approaches for Real-time Mitigation of Power Peaks in Railway Networks using Train Control Measures

Contact: Prof. Nikola Bešinović