A new kind of data capture process was on display at this year’s Hannover Messe, one that deploys a novel programming language and temporal graph solutions to rapidly collect and analyze sensor data. 

Developed by the University of Luxembourg spinoff DataThings, the system – dubbed Greycat – uses machine learning algorithms to develop digital twins of industrial assets. While the concept itself is not new, the manner through which it is achieved is novel in that it works far more rapidly than traditional methods of data analysis and works on any server or computer. 

The group currently provides its services to the CREOS; the national grid manager for Luxembourg – which has ramped up its domestic smart meter rollout to almost 99% deployment. While a small country, the uptake remains significant and foreshadows some of the problems larger grids will run into as we see increased instances of renewable technologies and domestic energy production adding to its load. While the move to cleaner energies is a growing ambition for nations, much grid infrastructure remains incapable of withstanding the anticipated increase in capacity. Predictive supply management and maintenance will become a crucial tool to manage this; as will rapid data analytics tools.

DataThings believes Greycat can be the answer to this growing need. The company says its system can profile every smart meter in a network and assess trends over time – with the whole process taking around 90 seconds for the whole of Luxembourg. Contextual data is also taken into account for these assessments, including weather, time of day and day of the week. Creating digital twins out of this data, the group creates a virtual map of a network to monitor possible issues and, in the case of CREOS, ensure the flow of electricity is moving in the right way at the right times.

As the group’s code communicates directly with a computer’s central processing unit, companies don’t require specialist systems to run the program, with the capacity to create digital twins of their assets available regardless of their background.

“As more and more houses are equipped with smart meters, the electricity grid is going to become harder to manage, with readability and predictability becoming increasingly difficult,” said business development manager Franck-Alexandre Sallebant-Bessone. “To help the grid, you don’t necessarily need specialist systems, you can deploy machine learning and predictive features very efficiently all on a consumer’s machine.”

The group also has applications beyond energy systems, with ambitions to continue adding to its repertoire to prove the versatility and applications of digital twin and data processing solutions.