As the electricity mix evolves from predominantly centralized thermal power generation to decentralized renewable generation, grid operations become increasingly complex. As an expert in AI, Optimization and energy, Artelys is supporting major energy players that have launched initiatives to devise innovative ways of operating the power grid based on optimization and machine learning. Among them, RTE, French TSO, is animating the L2RPN competition (Learning to Run a Power Network) to encourage the development of solutions based on Reinforcement Learning approach. The competition has fostered the development of the open-source framework grid2op, which Artelys is a contributor for years. In the 2023 edition co-organized by the Ile de France region, Artelys successfully participated as a contender with its solution that was ranked first on the leaderboard! Our solution is based on a multi-agent cooperation framework where specialized agents based on different technologies (Reinforcement Learning, Optimization and Expert System) can take advantage of the different options (topological actions and flexibility mechanisms) when they are best suited. This success paves the way to the next steps to safely bring these AI assistants in operator’s hands.
Recent Comments