Institute for Automation and Applied Informatics (IAI)
Area of research:
Diploma & Master Thesis
In recent years, the dynamics of our electricity networks has changed dramatically by integrating renewable energy production and more dynamic loads – e.g. smartly managed heating systems or car-loading stations - into the German power network. Thus, an accurate load forecasting becomes a very important prerequisite for balancing generation and consumption to not overload current electricity lines.
In this thesis, a variety of deep learning methods e.g. Convolutional Neural Network (CNN) or Recurrent Neural Network (RNN) will be scientifically analyzed and instrumented for achieving a more flexible and accurate prediction of short- and long-term load forecasting for different energy consumer settings. A focus lies on exploring unsupervised learning strategies for solving the problem..
State-of-the-art deep learning frameworks such as TensorFlow or Keras will be used to build the forecasting models. The prototypical implementation of those models can foster existing technologies, e.g. Jupyter notebook running on an underlying Big Data cluster. Additional, a powerful big data environment including Hadoop and Apache Spark will be available to achieve the best performance.General interest in application of data analysis, deep learning and Big DataGood programming skills in python and Java Thesis can be written in English or German languageStudent of computer science, electrical engineering, mechanical engineering or related disciplines
Shadi Shahoud, M.Sc.
Institut für Automation und angewandte Informatik, Campus Nord 0721 608 24123