Powergrid Congestion Management with Model-based Reinforcement Learning

Matthias Dorfer & Marcel Wasser from enliteAI talk about how Reinforcement Learning can be utilized to implement non-costly congestion management measures.

Learning to Run a Power Network (L2RPN) is an international Machine Learning Competition organized by EPRI and RTE, the French Transmission System Operator. The annual competition attracts the attention of top international research teams with the goal to advance the state of the art in AI powered control of large scale power networks.

Our Reinforcement Learning Team won 3rd place among several international competitors and was now invited to present our approach at the Workshop: Learning to Run a Power Network (L2RPN) 2021 – Towards a Machine Learning Based Control Center Digital Assistant.

Link to enliteAI's RL Framework MazeRL: https://github.com/enlite-ai/maze