Conventional data processing methods record the entire amount of data generated, which often leads to highly redundant measurement signals that then have to be analyzed in a further step. This not only involves a great deal of effort, but usually only a small proportion of the data material is required for the respective work process. With the "SigMaSense" research group (signal processing for material data acquisition with intelligent sensor technology), TU Ilmenau has been working together with the Fraunhofer Institute for Nondestructive Testing IZFP in Saarbrücken on methods for acquiring intelligent data since the beginning of 2018.

The new approach: only record what is actually needed in the sensor. The scientific director of SigMaSense, Dr. Florian Römer, wants to create "intelligent" sensors: "Instead of simply recording data all the time, the sensors should learn to think and decide independently which data is relevant and which is not. That way, we only keep the part of the data that is actually needed."

Instead of large piles of data that have to be laboriously stored and searched for specific patterns, selected, condensed data could be obtained. A welcome side effect: the new process reduces the energy consumption of sensors, communication interfaces and data storage devices. The result would be not only intelligent, but also "green" sensors.

In order to achieve the research goals set, the TU Ilmenau is planning a new professorship in collaboration with the Fraunhofer-Gesellschaft, which will intensify the work of the existing SigMaSense research group. The Free State of Thuringia is supporting these activities this year with 150,000 euros. At the TU Ilmenau on October 21, 2019, Thuringia's Minister of Science Wolfgang Tiefensee also announced further support of up to 2.4 million euros from 2021 for the expansion of the research group and its establishment as a branch of the Fraunhofer Institute for Nondestructive Testing IZFP Saarbrücken.

Cognitive sensor systems - efficient processes
The Fraunhofer IZFP is an internationally networked research and development institute in the field of applied, industry-related research. Its activities focus on the development of cognitive sensor systems for the non-destructive monitoring of industrial processes and value chains. The institute's understanding of technical testing and sensor physics is complemented by technologies and concepts from AI research, which are used to develop sensor systems for the NDT of tomorrow. In addition to pure production processes, R&D activities focus equally on processes from materials and product development, maintenance, servicing and recycling of materials.

Current research topics concern the development of sensors that are able to record production-related microstructure patterns and combine them into a kind of product THEN in the sense of an individual fingerprint. In the future, these digital product files will open up new approaches for optimizing the individual areas of a material and product cycle.