From the stakeholder to the system: Grid operators need complex calculation models to accommodate a great number of power producers and flexible consumers. Current mathematical models for grid optimization, which ensure a stable and cost-efficient power grid, are increasingly reaching their limits.

The AI OPF project investigates whether the AI-based method of deep reinforcement learning (DRL) is suitable for optimization based on forecasting of grid bottlenecks and flexibility calls. The aim is to develop a DRL method for the uncertainty-prone optimal power flow (OPF) that can be trained independently using a high-performance computing grid simulator. The application fits well with the new requirements for the lower voltage level. Distribution grid operators can use it to support their system control in line with the Redispatch 2.0 specifications.