Persons involved:

Ira Diethelm, Martin Fränzle, Sebastian Lehnhoff, Jorge Marx Gómez, Wolfgang Nebel, Jürgen Sauer

We research and develop innovative processes, IT systems and models for an environmentally friendly, economical and secure power supply of the future. To this end, we work closely with the OFFIS Institute's energy department.

The electrical energy supply is facing major changes, triggered on the generation side by the increasing integration of decentralised and above all renewable energies - especially those with fluctuating feed-in - and on the consumer side by the possible and necessary control of consumption systems. For example, photovoltaic and wind energy plants deliver their power depending on meteorological influences, which are usually not related to the load profiles of consumers: while solar energy is produced particularly at midday, consumption peaks in households typically occur in the evening. Shifting the operating cycles of individual consumers and appliances to such oversupply situations is just one way of levelling out this imbalance. The trend towards electromobility is also expanding the grid to include options for decentralised electricity storage, which will require new methods for charging batteries and the associated settlement.

Innovative approaches from Computing Science are needed to coordinate generation and consumption. In the Internet of Energy, producers, consumers, electrical grids and new types of service providers are cooperating to create an environmentally friendly, economical and secure power supply for the future.

In Oldenburg, research is being conducted in the following areas in particular:

• ICT integration of decentralised systems: Future intelligent energy supply systems, known as smart grids, will be characterised by an increasing number of active components that monitor the consumption and generation of electrical energy and coordinate them during operation. Starting with the digital electricity meter in the household, through new decentralised generators and controllable consumers, to forecasting and monitoring systems: "smart" IT-supported and fully networked components everywhere will exchange standard-compliant information with each other and independently coordinate and optimise their processes.
• Multi-agent systems and self-organisation: As the number of active components and players in smart grids increases, so does the complexity of the overall system to be optimised. Operational optimisation, which could previously be integrated and carried out centrally, is becoming increasingly difficult and is already no longer manageable during operation in many areas. Self-organisation in natural distributed systems should serve here as a model for decentralised energy management, in which autonomous software agents coordinate with each other and in this way achieve optimum operation of the overall system.
• Machine learning methods for an intelligent power supply: Fluctuations in consumption and generation in the millisecond range make it necessary to continuously re-evaluate deployment plans on the generator and consumer side based on current supply situations. Due to a lack of precise information about the overall system status, machine learning methods are an option, in which software agents - deputising for consumers and generators - evaluate the success of their own actions on the basis of locally available information and adapt them if necessary. By systematically testing new strategies, possibly in conjunction with pattern recognition methods, even unpredictable situations can be reliably learnt and suitable reactions derived.

• Data stream management systems for the integration of distributed, continuous measurement data under tough latency requirements: The approaches from decentralised energy management can only be implemented if a Computing Science infrastructure for the distributed monitoring, control and regulation of energy systems is available that is also suitable for highly dynamic supply situations. This requires the scalable and efficient handling of a massive volume of data, typically in the form of data streams, and any resulting reaction in compliance with real-time requirements. We are researching how data stream management systems can be used to process distributed, continuous measurement data under tough latency requirements.
• Real-time capable analysis methods and protection systems for use in highly dynamic supply situations: As the number of decentralised generators increases, the number of active protection and control technology components must inevitably grow, both on the system side and within the existing grid infrastructure, in order to guarantee the necessary protection and control functions for security of supply in the grids. On the one hand, reliable detection and timely reaction to faults of varying degrees of urgency (short circuits, supply instabilities, etc.) are required. On the other hand, continuous reconfiguration of the installed protection systems to current supply configurations is required in order to avoid false tripping due to unforeseeable situations.

In order to make the relevance of ICT for energy supply more transparent in society and thus also contribute to more responsible, efficient use, these topics are also part of didactic research and development. Based on specific systems in the schools themselves, for example, which often contribute to energy generation as operators of a solar system, the above-mentioned ICT challenges in energy supply can be incorporated into teaching and teacher training concepts for practice-oriented, real-life Computing Science lessons and thus contribute to anchoring the topic of ICT for energy supply in school curricula.