The junior team led by Dr. Christoph Scholz, which emerged from the Competence Center Cognitive Energy Systems at Fraunhofer IEE, is pursuing the goal of developing the potential of deep reinforcement learning in the energy system and making it safer, more effective and more cost-effective for the energy industry. "In recent years, the development of the research field of deep reinforcement learning has made great progress. However, central questions remain unanswered in the context of the use cases under consideration, which will be investigated in more detail in RL4CES," explains junior research group leader Dr. Christoph Scholz. "The solutions we develop will then be integrated into the application."

The focus of the junior research team will be on the two use cases of automated grid control and automated energy trading, while the IT infrastructure, simulation environment and AI expertise developed by the Fraunhofer IEE competence center will serve as the basis for highly specialized advanced research.

The collaboration between Fraunhofer IEE and the University of Kassel plays a central role within the junior research group. The Department of Intelligent Embedded Systems (IES) at the University of Kassel is involved in basic and applied research in the field of artificial intelligence, especially deep learning. In addition to application areas such as transport and materials science, further use cases have been realized in the field of renewable energies and future energy systems.

"At the IES department, the theoretical foundations for problems within the energy industry can be developed in a targeted manner," says Prof. Dr. Bernhard Sick, head of the department at the University of Kassel. This should ensure continuity from basic research to application-oriented utilization and the use of reinforcement learning in the energy industry. This will not only enable research and application to go hand in hand within the project, but will also support and promote a new generation of experts whose personal and scientific development will benefit from the expertise and mentoring of the Fraunhofer IEE and the IES department at the University of Kassel as part of the doctorates they are aiming for.

Background Deep Reinforcement Learning

Deep reinforcement learning (DRL) is a class of autonomous learners (artificial intelligences) in which autonomous neural networks independently find and try out possible solutions. The strategies learned in this way do not necessarily have to be monitored or limited by experts. In contrast to classic optimization methods, they are very quick to make decisions after training and are also able to deal with the increasing complexity of dynamic systems such as the energy grid, which gradually pushes classic optimization and rule-based methods to their limits.

Contact:

Dr. Christoph Scholz
Fraunhofer IEE
E-mail: christoph.scholz@iee.fraunhofer.de

Prof. Bernhard Sick
University of Kassel
Department of Intelligent Embedded Systems (IES)
E-mail: bsick@uni-kassel.de