Fraunhofer ITWM at the E-world
Our experts of the departments »System Analysis, Prognosis and Control«, »High Performance Computing« and »Financial Mathematics« present our solutions in the field of energy in Hall 5, Booth 5-679:
- »Financial Mathematics«: We research on the future of energy networks from a financial mathematical perspective. The project team prepares forecast models of how the demand for electricity will develop over the course of the year in order to optimally design the electricity grid.
- »High Performance Computing«: Our group »Green by IT« is researching on the efficient use of renewable energies, one example is the myPowerGrid platform: The Amperix® networks producers and storage facilities of different manufacturers and carries out the sector coupling of electricity, e-heat and e-mobility in private households and industry.
- »System Analysis, Prognosis and Control«: The team researches the optimal maintenance planning of plants by predictive maintenance and the modelling, simulation and control of hierarchical energy networks in the MathEnergy project.
- »Optimization«: the team is on site with its expertise energy and the project FlexEuro »Economic optimization of flexible electricity-intensive industrial processes«.
Another major topic is Predictive Maintenance in the energy industry, i.e. optimizing plant efficiency through machine learning.
Ideally, a technical system is considered reliable and economical, if it is repaired promptly and available when required. This is only possible if the company can reliably predict the maintenance requirements of the systems, taking into account the current production plan and past load history, while guaranteeing the availability of the appropriate resources such as specialists, spare parts, logistics, etc.
Reliable prediction of future events is an integral part of any Predictive Maintenance (PM) system. An important key lies in the analysis of patterns in past events. In a joint modeling approach, we model not only the continuously measured sensor data, but also repetitive discrete event data and failure data. We develop machine learning methods to recognize and visualize complex high dimensional patterns as well as the dynamics and trends of production process states. Further-more, we use machine learning algorithms to predict and characterize the condition of technical systems.