Autonomous AI for cellular energy systems increasing flexibilities provided by sector coupling and distributed storage

Abstract

Given the growing demand for electrical energy and power due to the electrification of the heating, and mobility sector, the distribution grids are under stress. High peak power demands alternating with high energy generation challenge the secure operation of distribution grids. To cope with these challenges, this paper presents a hierarchical cellular energy management system developed to promote the integration of renewable energy and improve power quality in distribution grids. The cell manager is based on state-of-the-art reinforcement learning agents to efficiently control power flow and detect overload situations in low-voltage grids at an early stage. The combination of flexible power generation units and consumers within cellular energy structures can optimize the resources and increase the robustness of the grid. In addition, a sector-coupled simulation provides a deep insight into the interactions between the different forms of energy. The findings of a one-year simulation demonstrate a decrease in peak loads compared to conventional methods. This concept provides not only innovative solutions to address the challenges of the energy transition but also practical applications for enhancing efficiency in urban energy systems.