Artificial Intelligence in Automation

SyDaPro: Synthetic data in production

01.10.2021 bis 30.09.2023

Artificial intelligence allows production facilities to improve their maintenance through prediction and optimize entire production processes. One challenge of artificial intelligence is the necessity of providing enough data, which does not exist in the necessary quality and quantity.
The project SyDaPro addresses the generation of synthetic data in industrial production. Based on real data and expert knowledge, we generate probabilistic models. These models admit the generation of realistic synthetic data and also of purposefully crafted anomalies.
In the project SyDaPro we gather and preprocess data, and enhance it with expert knowledge. Based on this, we conduct the modelling and data generation. We deduce best practices and publish those together with the project results. The application of synthetic data is highlighted on demonstrator systems, where artificially generated data is used to optimize production.
Using adaptable algorithms, this project allows data synthesis using artificial intelligence in various application domains, which a-priori suffer from lack of data preventing robust usage of artificial intelligence. This results in manifold opportunities to save time and money by preventing system failures and to realize predictive maintenance, reduction of risks, and optimal processes. In particular, this allows medium sized companies to use artificial intelligence effectively and efficiently in industrial production. This high potential of transferring the project results to additional industries yields a high economic potential: the profiteers of the project are all producing companies.

This project is promoted by:
Bundesministerium für Bildung und Forschung (BMBF)
Sponsors: Das Deutsche Zentrum für Luft- und Raumfahrt e.V. (DLR), Datenwissenschaften/ Software-intensive Systeme
Funding Code: 01IS21066A
Funding Lines: Erzeugung von synthetischen Daten für Künstliche Intelligenz
Promoted by
Datenwissenschaften/ Software-intensive Systeme