We are proud to announce the successful publication of our latest research paper, “A General Approach of Automated Environment Design for Learning the Optimal Power Flow”, at the prestigious ACM e-Energy 2025 conference. The authors are Thomas Wolgast and Astrid Nieße. 

The paper introduces the first general method for automated Reinforcement Learning (RL) environment design, by the example of solving Optimal Power Flow (OPF) problems. Using a multi-objective optimization approach within a hyperparameter optimization framework, the study shows that automatically designed environments outperform manually crafted ones across five benchmark problems. The work also identifies key environment factors that impact OPF performance and highlights potential overfitting risks.

This contribution marks a significant step toward more effective RL applications in energy systems but also in RL research in general, considering that RL environment design is an often neglected topic. 

The publication can be found at: https://dl.acm.org/doi/10.1145/3679240.3734626.