See TUMonline

Today's electric power grids and supplies are cyber-physical systems, where information and communication technology (ICT) play an important role in reliably operating all system components. Electrical data can be recorded everywhere, for example in households, office buildings and industrial plants. Analyzing such data enables a wide variety of possible applications. Furthermore, machine learning on electrical time-series data poses multiple challenges to both algorithms, platform design and data privacy. In this seminar, students will be able to make own research contributions to this research area. They can dive into the different areas of energy informatics, with focus on the various machine learning techniques applied to this research field and the applications for Smart Homes or Smart Factories which are based on the results of such analysis. Applications based on electrical data analysis range from detecting individual appliances in households to condition monitoring and predictive maintenance on industrial plants. The wide distribution of sensors in the context of “Industry 4.0 scenarios” further enables such applications and it is of particular interest to investigate the use of electrical data to facilitate the development of smart factories. The topics in this seminar are suited for both students with beginner or advanced knowledge in machine learning and ones interested in it from a higher-level application and architectural perspective. The topics consist of conducting literature research and some of them also require the development of a prototype.