Chen Li from the research group of FIAS Senior Fellow Horst Stöcker successfully defended his doctoral thesis on September 22, 2025. Using machine learning algorithms, he optimized decision-making problems in power grid models that arise due to fluctuations in renewable energies.

Chen Li's work also attracted considerable public attention because the technology developed by the team stabilizes the use of renewable energies in the grid, thus preventing weather-dependent fluctuations in the power system (see News / FIAS, News / FIAS).

“To address the challenges by the increasing proportion of renewable energy in the process of energy system transformation, we explored the possibility of using machine learning to assist in solving optimal power dispatch problems,” explains Li. The central innovation of his work is that the system becomes scalable by replacing costly optimization steps with neural network inference, projection techniques, and user-defined terms, analyses, and visualizations. This makes the system adaptable to different network solutions and energy mixes.

Using attention-based graph neural networks, Li showed that power flow and energy storage dispatch  in highly renewable systems can be made more efficient while ensuring the interpretability of the algorithms. This not only reduces computational costs but also enables faster real-time decisions in modern smart grids.

Li hopes to continue his research in the field of AI for science and technology in order to advance the development of energy systems toward renewable, cleaner, and smarter systems. FIAS congratulates him and wishes him all the best for the future!