Reinforcement Learning based Coordination of Virtual Inertia Provision from Inverter-dominated Distribution Grids

To achieve the goals of carbon neutrality in the energy sector, inverter-coupled renewable energy sources play an important role. In contrast to conventional generation, they are mostly connected on the distribution system level and do not provide inherent inertia when the system frequency changes. To address this problem, renewable energies in distribution systems are assumed to reserve and provide additional energy. This paper aims to develop a function that coordinates this provision to avoid congestion and optimize grid operation. Therefore, we propose a Reinforcement Learning (RL) based optimization utilizing operational parameters and energy costs. The functionality and performance are explored in a case study on a part of the Australian transmission and the CIGRE Benchmark MV grid.