Conceived as the German government's contribution to the energy revolution, this strategy is spearheading the move away from nuclear power and fossil fuels and toward renewable energies and improved energy efficiency. As part of this switch, the centralized power supply generated by a handful of large power plants is gradually being replaced by a decentralized network of many small plants feeding electricity into the grid. At present, around 40 percent of all the electricity generated in Germany comes from biomass or wind, solar or hydroelectric power. However, these renewable energy sources are of course subject to external factors, such as weather or time of day, that lead to fluctuations in the grid and thus compromise the reliability of the supply. That's why the research project "Sensors in the Grid 2.0" - a continuation of the three-year project of the same name that ended in 2017 - aims to contribute to the success of the digital revolution by looking into intelligent, simple and cost-effective ways of analyzing the status of distribution grids and the plants connected to them.

Led by PPC, a specialist in smart meter gateways and communication technology, and funded by the Federal Ministry of Education and Research, this project is making its first-ever venture into large-scale solutions for monitoring grid statuses with the aid of big data and artificial intelligence. The way it works is very simple. Broadband over power line (BPL) technology ensures all the points in the power grid can communicate with each other via the power cable. The "sensors in the grid" can then use this foundation to perform real-time grid monitoring and record the condition of cables and plants. AI algorithms based on machine and deep learning approaches from the German Research Center for Artificial Intelligence (DFKI) are used to recognize patterns and anomalies in the enormous volumes of data that accumulate within a very short time frame. The AI analyzes the data streams, identifies anomalies, learns from them and thus derives predictions and its own strategies, making it possible to predict the voltage curve and the asymmetry between phases. These findings could prove extremely valuable in the future, particularly when integrating e-mobility systems into distribution grids. Prof. Andreas Dengel from DFKI believes self-learning AI methods are potentially the key to a successful energy revolution: "Self-learning algorithms are ideal for analyzing both large volumes of data and the forecasting methods for energy generation and consumption developed from this data. They form the basis for an intelligent network monitoring and management system and are paving the way for the smart energy system of the future."