The expansion of renewable energies is also placing greater demands on the grid infrastructure. Fluctuating yields owing to the weather mean that ways need to be found to ensure a continuous electricity supply. Against this backdrop, a German-Finnish research project called FUSE – Future Smart Energy was launched in August this year. Researchers seek to intelligently connect the ever more complex energy grids. In order to raise the efficiency of the grids and to guarantee supply, information and communication technologies (ICT) based on AI are being developed.

By feeding in electricity generated, for instance, from solar power plants, the grids are no longer only being used one-dimensionally but bi-directional. That’s why constant surveillance and control is necessary to prevent breakdowns and other disruptions. The use of AI will help to predict potential failures long before they occur, so that appropriate maintenance measures can be taken (“predictive maintenance”). The structure of ICT-architecture for AI-supported data processing consists of many decentralised, intelligent sensors and processing nodes. The algorithms are to be run on small, energy-saving, embedded systems.

The integration in intelligent sensors or in machine control systems enables us to use complex and computationally-intensive machine-learning tools, such as deep neural networks, directly on-site. Methods for predicting the energy consumption, depending on grid usage and energy price, also play an important role; they can help to improve the stability of the grid long-term. The goal is to implement an integrated infrastructure in Germany and Finland. This will enable the connection between different processing nodes, for instance in the form of intelligent sensors, digital twins or backend systems, which will facilitate secure communication channels on all levels of the system.

The German project partners are German Research Centre for Artificial Intelligence (Bremen) as project coordinator, Easy Smart Grid GmbH (Karlsruhe), TU Berlin, ALL4IP GmbH & Co. KG (Darmstadt). The project is being funded by the Federal Ministry for Economic Affairs and Energy (BMWi) over a period of three years.