sada

Bachelor/Masterarbeit

Unten aufgeführt finden Sie Themenvoschläge zu Bachelor-/Masterarbeiten. Nach Absprache können auch andere Themen vergeben werden.

All projects can be carried out either in English or in German.

Profitoptimierung in Smart Grids

Smart Grid (http://www.renewableenergymexico.com)
Smart Grid (http://www.renewableenergymexico.com)

Das Intelligente Stromnetz (Smart Grid) dient der „kommunikativen Vernetzung und Steuerung von Stromerzeugern, Speichern, elektrischen Verbrauchern und Netzbetriebsmitteln in Energieübertragungs- und -verteilungsnetzen der Elektrizitätsversorgung“ [http://www.nist.gov/smartgrid/].

Für solche Systeme müssen mathematische Methoden angewendet werden, die einen effizienten, nachhaltigen und robusten Funktionsablauf der Smart Grids garantieren.

Es gibt viele Möglichkeiten für Studenten solche Methode für unterschiedliche Szenarien in Smart Grids auszuprobieren und einige neue Ideen zu entwickeln, um mit einem von folgenden aktuellen Problemen in Gebiet der Smart Grids voranzukommen:

1) Praxisrelevante Zielfunktionen und Verhaltensmodel für Teilnehmer des Grids;

2) Vorhersagen und Onlineoptimierung in Smart Grids;

3) Verteilte Regelung von Verbrauchern und Energieerzeugungen in kleinmaßstäblichen Energiesystemen (Microgrids);

4) Nachfragemanagement in Smart Grids.

Alle theoretische Ergebnisse sollen auch in MatLab/C++ simuliert werden.

Distributed optimization in multi-agent systems

Distributed optimization is a rapidly developing sub-area of distributed computation that aims to design algorithms solving decomposable multiagent optimization problems efficiently. There many examples of such optimization problems in real world applications: optimal wind farm control, stability of power grids, model predictive control in engineering processes, data analysis in machine learning.

In this project students will choose an application of distributed optimization, investigate the properties of the corresponding environment, and develop an optimization algorithm that can be applied to this environment. The efficiency of the algorithm needs to be evoluated according to the standard criteria such as convergence to a local/global solution as well as the convergence rate. The theoretical analysis should be supported by the simulation of the optimization algorithm by means of a technical computing language (C++, Matlab, Python).

Efficient Approach to Machine Learning

Picture originates from https://www.healthcatalyst.com/clinical-applications-of-machine-learning-in-healthcare
Picture originates from https://www.healthcatalyst.com/clinical-applications-of-machine-learning-in-healthcare

Machine learning aims to achieve competitive advantages in different spheres of our everyday life. Machine learning provides us with self-driving cars, such practical skills as speech and image recognition, fast and effective web search, reliable prognoses for energy consumption, and improved understanding of the human genome. To achieve such ambitious goals, it uses mathematical techniques to analyze the masses of relevant data. Due to large amount of these data computational complexity in machine learning applications becomes the limiting factor.

This project offers students an opportunity to investigate different ways to overcome the computational complexity in machine learning and to develop methods that would guarantee a trade-off between efficient data analysis and fast learning ability of the corresponding algorithms. All theoretical results should be supported by simulations (C++, Matlab, Python). The project can be carried out either in English or in German.

zurück zur Übersicht der am Fachgebiet angebotenen Arbeiten