GridAssist

The GridAssist collaborative project aims to develop assistance systems for optimized automated system management in distribution grids. This is intended to enable greater integration of renewable energy generators into the power grid. Solutions tailored to specific voltage levels are being developed. The transferability of these solutions and the field testing of the assistance systems are central components of the project.
Sufficient observability forms the basis of any automation in distribution grids. The lower the voltage level, the less information is available about the current grid status. GridAssist will therefore first establish the necessary observability. In addition to installing measurement technology, appropriate concepts must be developed to determine how the measurement data can be most effectively integrated into the respective grid operators’ systems and processes. Once sufficient measurement and communication technology has been installed—i.e., once observability is established—the respective grid state can be analyzed. In addition to analytical methods, data-driven methods will also be employed. Based on the analysis of the grid state, decisions must then be made by the assistance systems. These can range from recommendations for action for the system operator to fully automated flexibility calls.

One of the chair’s main focuses is on developing embedded AI for substation control systems to enable forecast-based congestion management in local distribution substations. Other topics include automated congestion and fault management based on machine learning methods in distribution networks, as well as the development of an assistance system for grid restoration. In addition, the Chair of Pattern Recognition (Pattern Recognition Lab) is participating in the development of a graph database to standardize heterogeneous data.