5G and Machine Learning

There is still much to be done before the announced launch of the fifth generation of mobile communications. The »5Gain« project focuses on 5G infrastructures for cellular energy systems. Development is carried out using machine learning methods. Because 5G is more than a new standard.

5Gain – 5G Infrastructures for Cellular Energy Systems With AI

The BMWi-funded project »5Gain – 5G infrastructures for cellular energy systems using artificial intelligence« started in December 2019 with the aim of developing and evaluating 5G-based communication for the distributed control of cellular energy systems.

»5Gain« focuses on the advantages of a 5G-based infrastructure for the management of cellular energy systems. In combination with AI methods, new possibilities for intelligent grid control emerge.
 

Cellular Partitioning and Distributed Control of the Energy Network

Due to the decentralized expansion of renewable energy sources with controllable loads and storage (e.g. electric mobility), the control of energy systems is becoming increasingly complex. At the same time, the expansion requirement of the power grid should be kept low. Our approach to solving this challenge is to divide the energy grid into regional cells. Each cell has different participants and characteristics and carries out decentralized load, feed-in management and marketing. We develop adaptive AI procedures (e.g. reinforcement learning), which learn to control the individual energy network.

5G-supported application scenarios of cyber physical production systems
© Fraunhofer ITWM
5G-supported application scenarios of cyber physical production systems

Communication via 5G Regional Network Slices

The regulation of distributed generators and consumers requires a communication infrastructure that provides the required data rates, response times and resources for different numbers of participants in any situation (e.g. traffic jams, old town festival). The 5G standard provides a dynamic and location-based assurance of quality of service guarantees through »5G Network Slicing«.

We develop forecasting methods to identify communication requirements of events at an early stage. This allows »5G Network Slicing« to be selected regionally so that required communication resources are provided locally.

Billing Through Smart Contracts

Billing between generators and consumers should be automated and secure. In 5Gain we are testing this on the basis of smart contracts in a block chain. Furthermore, the feasibility of a data-intensive, automated remote maintenance of distributed infrastructures using drones is being analyzed.
 

Our Competences in the Project

Based on our methodical competences in machine learning and our broad project experience in monitoring and control of energy networks, we develop in 5Gain:

  1. Adaptive AI methods for the distributed control of cellular energy systems.
  2. Forecasting models to predict communication requirements based on demand.

Project Partner

  • adesso AG
  • Dortmunder Energie- und Wasserversorgung GmbH
  • urban ENERGY
  • PHYSEC
  • City of Dortmund
  • RWTH Aachen
  • Technische Universität Dortmund (TUDo), Faculty of Electrical Engineering & Information Technology
  • Innogy SE
BMWi gefördert
© BMWi
Applied non-nuclear research funding in the 7th Energy Research Programme Innovations for Energy System Transformation of 01.10.2018.