• Energy Research

Electrically powered buses reduce CO$_{2}$ and noise emissions in urban areas and thus promote the trend towards more livable cities. Upon this reason, more and more cities are introducing their first electrified lines as pilot projects. However, no standardized technology has yet emerged, which is why statements on interactions between vehicle, operation and infrastructure in public transport are proving to be difficult to make. In order to be able to make statistically significant statements in this respect, a simulation model was developed that depicts the three subsystems vehicle, operation and infrastructure. On the basis of measurement data from the PRIMOVE research project in Mannheim, in which an urban bus line is operated with two electrically powered buses, the simulation model was validated and a data basis was laid for further investigations. As a result, simulation studies with more than 700 simulated operating days could be carried out, the results of which represent the input for the following statistical analysis. Based on this analysis, the interactions described above will be demonstrated in the design of the main technical parameters, the battery lifespan and the energy demand of electrical bus lines. Through the findings of these simulations, an optimized version of the already electrified bus line in Mannheim will then be presented. Finally, a novel design methodology for electrification based on a multi-objective optimization is introduced. All parameters of the system are variable in order to apply the presented methodology to other projects and thus underline the general validity of the work.

Diese Arbeit untersucht, wie die Energieeffizienz und Pünktlichkeit von Straßenbahnen durch den Einsatz von KI gesteigert werden können. Auf zwei Szenarien wird die KI zu je drei Verkehrszeiten trainiert. Die ermittelten Fahrprofile werden mit denen von Fahrern aus dem regulären Fahrgastbetrieb sowie mit einem durch Dynamic Programming ermittelten theoretischen Optimum verglichen. Darüber hinaus werden Transferlernfähigkeiten der KI untersucht.