Knowledge graphs for Industry 4.0

Using connected data to move toward smarter process analysis and more trustworthy electronics

Tomorrow’s digital factory will be able to organize itself more and more independently and will also be increasingly automated or able to respond on its own to malfunctions. However, to become reality, this vision requires a wide range of machine-readable data in large quantities. To this end, our Center for Applied Research on Supply Chain Services is developing “knowledge graphs.” These knowledge graphs use artificial intelligence (Al) to link data from different sources and pool it all together. With this Al-based technology, it is possible not only to represent data, but to conduct predictive analytics as well. For example, in such a data space, Al can analyze in advance what time is good for a repair so that upcoming orders can be processed in line with demand.


Al-Nalyze – intelligent analysis of production processes in real time 

Disruptions often occur at inopportune times. The Al-Nalyze project aims to create an intelligent process analysis solution that, using AI, can learn from previous failures and predict future ones in real time. Together with Trevisto AG, we are working to find opportunities for controlling and improving transparently mapped processes for the Siemens AG plant in Amberg, Germany. Data such as machine speed, power consumption or which workpieces are currently in use are automatically recorded – as are malfunctions. This smart factory data is applied to optimize processes in real time; for example, by having an Al suggest a suitable time for maintenance – before any damage occurs. The underlying knowledge graph makes it possible to link the various data sources and thus provides the system context for individual data points, which the Al can then take into account.


Welektronik – a wiki for electronics hardware without security gaps

In computing parlance, a backdoor is an unwanted access point built into electronic hardware that hackers use to bypass security systems so they can spy on, say, corporate networks. To ensure that trustworthy suppliers are reliably recognized by the German microelectronics industry, we are developing new software as part of a Fraunhofer-wide consortium project on trustworthy electronics, called Velektronik. With knowledge graphs as a database, we are building Welektronik: an open, collaborative wiki for describing supply networks and analyzing them for trustworthiness. The data for this is obtained from public sources such as Wikidata or manufacturer websites, but can also be entered by supply chain participants themselves.