Die Graduierten arbeiten in Self-Organized Study Groups (SOSGs), die jeweils aus 3-5 Mitgliedern bestehen, gemeinsam an Projekten zu speziellen Fragestellungen aus dem Kontext Energiezustandsdaten. Auf diesen Seiten möchten wir die Study Groups kurz vorstellen.
Members: Lukas Barth (ITI), Nicole Ludwig (IAI), Esther Mengelkamp (IISM), Philipp Staudt (IISM)
The increasing share of intermittent energy generation is a big challenge for the electricity system. Demand side management has been touted as one measure to tackle this challenge. Flexibility on the demand side is essential for its success. Extensive research exists that describes, models and optimizes various processes with flexible electrical demands. However, most of these approaches are very process-specific, and there is no unified notation. In this Self-organized Study Group, we develop a comprehensive modelling framework to formally describe demand side flexibility in smart grids, integrating various kinds of constraints from different existing models. We aim at providing a universally applicable modelling framework for demand side flexibility.
Storage and Privacy
Members: Jan Ullmer (IIWR), Rebecca Schwerdt (ITI), Shahab Karrari (ITEP)
“Energy Status Data” in the context of our same-named DFG Graduate School is meant to be understood in a broad manner, comprising all different types of energy-related data originating from a wide variety of possible sources. With the rapid increase in demand for (and availability of) high-quality energy data a hitherto negligible aspect of such data becomes ever more relevant: the protection of personal data and the privacy of energy consumers.
Data protection is a complex topic that calls for an interdisciplinary approach: Law and other regulations provide the legal framework and set the requirements, while analysis of privacy risks and development and implementation of data protection measures require mathematical and technical expertise. The SOSG “Data Protection” takes an interdisciplinary look into different aspects of data protection in future smart energy grids.
One of our topics is the privacy friendly and even privacy enhancing use of energy storage technologies. Providing sufficient energy storage capacities plays an important role in maintaining system stability in an energy system with an increasing amount of highly volatile electricity generation. But storage technologies may not only be used to store surplus energy or to optimize energy consumption in various ways. A very useful side benefit could be the possibility to employ energy storage devices to obfuscate actual energy consumption patterns and thus enhance protection of personal data.
Another topic are data protection issues arising from the integration of smart metering technologies in the highly regulated German energy market. The rollout of smart meters in Germany is rather slow-moving compared to other countries and associated innovations – like flexible tariffs or demand side management – are still rarely available. This is, inter alia, due to severe privacy and security concerns expressed by energy customers as well as in politics and expert circles. These concerns were already taken into account when technical and legal regulations were drafted, but there still remain data protection challenges that are not yet sufficiently thought-out. For the 2017 Summer School co-organized by our graduate school a workshop on smart meter privacy was held by our SOSG.
M.E.G.A. (Manufacturing Energy Data – Generation and Analysis)
Members: Simon Bischof (IPE), Holger Trittenbach (IPD), Michael Vollmer (IPD), Dominik Werle (IPD)
Maintaining a stable supply of energy requires constant monitoring. Tracking energy consumption for billing, forecasting and other analytic purposes often takes place in time intervals of 15 minutes. From a technical point of view however, measuring energy consumption is possible with a sample rate of seconds. But the benefits of collecting such high-resolution smart meter data for subsequent analysis purposes currently are not well understood. The expected loss of information introduced by reducing the sample rate or by relying on aggregated data depends on the specific analytics task performed. The research question of this self-organized study group is to identify and quantify the impact of aggregation on the quality of data-analysis results.
The object of our investigation is the production facility of the Institute of Power Electronics (IPE) at KIT Campus North. In this factory, each industrial processing machine is instrumented with a high-resolution smart meter. To realize the trade-off between data resolution and data-analysis quality, we compare the raw sensor data and aggregated values on different analytics tasks such as data stream clustering, classification and correlation analysis. Another use case is to identify the requirements on energy storage systems to run the production on renewable energy. It is important to identify the level of aggregation needed to perform such an analysis, because the requirements derived from the data might change with the level of aggregation.
In addition to this research, we envision publishing the data, and we plan to do in a privacy-preserving manner. Because many other open data sets in the energy domain come from private households, this data will allow researchers to work on data analysis challenges in an industrial context.
Time Series Reduction
Members: Thomas Dengiz (IIP), Edouard Fouché (IPD), Richard Jumar (IAI), Shahab Karrari (ITEP), Hasan Ümitcan Yilmaz (IIP)
Data Reduction (DR) is an area of Data Mining which consists in compressing large volumes of data to its most relevant segments. As Energy Status Data (ESD) often consists of very large time series, reducing the number of observations or variables is required for its analysis and further processing. However, current DR methods lead to unsatisfying results with ESD, because they do not account for energy-related characteristics. The focus of this Self-Organized Study Group (SOSG) is the development of novel methods for efficient Time Series Reduction (TSR). This SOSG currently focuses on two specific and distinct applications:
- Long-term system optimization models are useful to analyze political, technical, and economic questions regarding the development of energy systems. Especially w.r.t. the ongoing paradigm shift from conventional energy production to renewables, we will use them to analyze the impact of fluctuant electricity production sources on future energy mixes. However, the execution time of system optimization models increases rapidly with the temporal resolution and may range from several days to months if the underlying data is not adequate. To avoid this, we study whether one can use times series reduced via TSR methods. The expectation is that this will lead to acceptable runtimes without degrading the quality of the predictions significantly.
- High-resolution voltage and current measurements in the grid generate large time series that are not easy to deal with in various power system transient studies. Using TSR methods, it should become possible to describe the behavior of energy systems in a more concise manner, for example by producing representative waveforms for each of the measured variables. Once we can describe the normal behavior of the system, interesting knowledge such as abnormal events and current system status can be extracted in real-time.