International Research Atelier on AI Applications in Energy Systems

German Version

Research and industry together for a better future: More than 40 participants from business and universities from four countries discussed the use of AI in energy supply with experts.

Bielefeld (fhb): From May 3 to 5, the Working Group Grids and Energy Systems (AGNES) of the Bielefeld University of Applied Sciences (UAS) organized an international workshop on the use of Artificial Intelligence (AI) in the electrical grid. Experts from industry and research institutions from different countries gave an insight into their AI applications in the power system. The research atelier was sponsored by the Franco-German University of Applied Sciences and was part of the International Week of the Bielefeld UAS.

"The international research atelier was a great success!" said Prof. Dr. Jens Haubrock, head of AGNES. "It became clear that AI is a promising and future-oriented topic for the energy sector. However, there is still a need for further research. It is therefore even more important to communicate this important technology to young scientists in a way they can understand." This is a goal that the participating universities from Bielefeld, Grenoble, Athens and Morocco have set themselves.

Presentations from research and industry and grid operators

Prof. Dr. Ulrich Schäfermeier, Vice President for International Affairs and Digitalization, pointed out the importance of research in the field of AI and renewable energy sources in his welcome address. This was also highlighted in sessions on "AI in energy systems" by representatives from research and industry in the. Guests of honor and experts Prof. Nikos Hatziargyriou from the National Technical University of Athens and Prof. Nouredine Hadj-Sajd from the University of Grenoble Alpes drew attention to the challenges ahead for our energy system and highlighted the importance of AI in addressing them. Among the key challenges, in addition

to reducing CO2 emissions and slowing down climate change as well as the growth of volatile renewable energy systems, is the fact that more and more sectors are becoming electrified. This is leading to an increase in large-scale electrical consumers. "There is a need to adapt the power grid. It is important to think ahead in terms of the challenges," said Nouredine Hadj-Said.

The transmission system operator 50Hertz and the local distribution system operators Westfalen Weser Netz, Netzgesellschaft Gütersloh and Bielefelder Netze presented concrete applications of AI in energy systems and the associated challenges. Polina Sokolnikova from 50Hertz presented the approach of using AI algorithms to localize and classify potential defects based on high-resolution static images of overhead power lines. Maintenance, repair and dismantling can also be improved by using AI, as Timo Busse of Westfalen Weser Netze explained. Ansgar Ottensmann from Netzgesellschaft Gütersloh and Stefan Junghändel from Bielefelder Netze underlined the importance of AI to support employees in operational tasks and the need for well-trained engineers.

Close cooperation between network operators and industry

In meeting the challenges for energy systems, grid operators are receiving support from the industry. PSI Software AG offers software for transmission and distribution network operators as well as for industry. Dr. Michael Wolf from PSI presented AI solutions for relevant problems of transmission system operators, e.g. for the management of reactive power and its availability/reserves for current and future periods. Hélène Schricke from Atos Worldgrid in France addressed the need for flexibilities in the power system. To this end, Atos Wordgrid developed AI solutions to optimize battery storage systems. "Energy storage systems can be used efficiently in the context of large-scale deployment of renewable energy through the use of AI," said Hélène Schricke.

More research needed

Although industry has already presented several AI solutions, further research is important. Five research institutions from three countries presented their current research topics in AI for the energy sector at the Research Atelier. Dr. Marc Hesse and his research group from Bielefeld University are currently developing resource-efficient cognitive edge computing to run AI algorithms. This will allow data for AI to be processed directly in the field, reducing data flows and ensuring data security. Michael Kelker from the Department of Engineering and Mathematics at Bielefeld University of Applied Sciences gave a brief overview of the current research work on AI in the Networks and Energy Systems group. Nadja Isabelle Hirsemann from Ilmenau University of Technology is investigating data analysis of voltage dips using AI to find the cause of the dips. In Morocco at the Research Institute for Solar Energy and New Energies, Aboubakr Benazzouz and his team are using AI approaches for load forecasting. Dr. Rémy Rigo-Mariani at the University of Grenoble Alpes presented an unconventional approach: they incorporate Twitter feeds as additional features to improve forecasting tools.

Young researchers present their current work

The last day was dedicated to the young researchers who presented their current work and results on AI and digitalization. The working group Grids and Energy Systems of Bielefeld University of Applied Sciences presented their AI-based prediction approaches for locally generated PV power (Katrin Schulte) and for heat demand for planning the use of combined heat and power plants (Julian Hövelmann) as well as Deep Reinforcement Learning for autonomous control of low voltage grids, presented (Lars Quakernack). The University of Bielefeld presented requirements for the use of AI in the energy sector by Bastian Steinhagen as well as an adapted approach for state estimation of the low-voltage grid based on the methods of the University of Applied Sciences Bielefeld, presented by Timon Jungh. Another current research work at the Bielefeld University of Applied Sciences concerns education. Fynn Liegmann is developing a grid control station in virtual reality for learning how to operate real power grids. The project in which this elaboration is being created is funded by the Digi Fellowships program and aroused great interest especially among grid operators. After the presentations, the participants had the opportunity to take part in a lab tour, where they could also try out the virtual reality grid control room. (mgu/she)