Digital Energy

AI is going to transform energy systems. The Fraunhofer Center Digital Energy is taking a comprehensive approach in supporting industry in this sector to go through this transition. With the project AI-EFFECT, the group is creating an innovative ecosystem supporting industry in developing and testing AI solutions. The project JAIF addresses the challenge of training large AI models and to access the appropriate infrastructure for this task. Finally, the team is in a leading position in the AI.Grids project, which has the aspiration to develop the first foundation model for European energy grids.

For the development of a generic AI tool, we can envision 4 steps: data collection, model development, model validation and testing, deployment.

Concerning data collection, the energy system has specific challenges related to the criticality of the infrastructure that can be addressed by use of synthetic data or provision of training services that are preserving data sovereignty, e.g., via data spaces. Concerning the first approach, the Fraunhofer Center Digital Energy is working at developing appropriate methods for realistic synthetic data generation as part of the project AI-EFFECT. Energy data spaces are under development within the project TwinEU and implemented in the project INSIEME, where Fraunhofer FIT acts as a coordinator and contributor, respectively.

The model development is a complex operation requiring significant computational capability. The Center is involved in the project JAIF, which is setting up one of the European AI Factories, aiming to provide the right support for the energy sector. Furthermore, Prof. Monti from Fraunhofer FIT has pioneered the launch of the AI.Grids initiative together with NTUA in Greece and INESC TEC in Portugal. With the direct involvement of DG Energy of the European Commission and the coordination of the association CRESYM, AI.Grids has the aspiration to develop the first foundation model of the European electricity grid - fully based on European technology with the purpose of creating technical sovereignty in a critical application such as AI for energy.

But model needs to be checked and verified. Within the project AI-EFFECT, FIT is developing one of the testing nodes that will be available to European actors in the future as part of the Testing and Experimentation Facilities initiative of the European Commission.

Finally, new solutions need to be integrated into the daily business of the grid operators, which is extremely difficult in relation to classical software architectures for control rooms. The Center is developing modular control systems (such as SOGNO under governance of LF Energy) that can provide the required flexibility in research projects such as Open Energy Twin or the FIT coordinated InterSCADA.