Innovation Projects
Machine Learning for Production and its Products (ML4Pro²)
Making machine learning sustainably available
Thanks to machine learning (ML), knowledge can be generated from data to create added value at all levels of business processes. Products such as mobile platforms, robots or vehicles are utilising data to optimise their behaviour.
However, production systems are also drawing on it more and more often in order to enable them to react to new market developments and client needs in a more agile manner, and to produce the best products using available resources. In the process, using ML methods close to the source of data is especially promising. The project objective is to make ML available for intelligent products and production methods.
To that end, state-of-the-art ML methods are expected to be integrated into products and production chains. The project is also about increasing business awareness of how to utilise ML for agile business models. The main areas of focus are hybrid learning methods, integration of expert knowledge, data interpretability, learning from data streams, as well as Cognitive Edge Computing. ML methods are considered across applications, using three industrial use cases: State monitoring, process optimisation and product quality improvement. Results and methods are made available to companies on a ML platform. For instance, this platform consists of reference implementations, data preparation and data visualisation methods as well as application knowledge on typical processes when using ML methods.
Facts
Project name: Machine Learning for Production and its Products (Ml4Pro²)
Project duration: 01/12/2018 to 31/03/2022
Project budget: EUR 5.38 million
Publications
- Pelkmann, David, Alaa Tharwat and Wolfram Schenck: How to Label? Combining Experts’ Knowledge for German Text Classification. In: 2020 7th Swiss Conference on Data Science (SDS), IEEE, 2020, pp. 61-62.
- Holst, Christoph-Alexander and Volker Lohweg: A Redundancy Metric based on the Framework of Possibility Theory for Technical Systems. In: 2020 25th IEEE International Conference on Emerging Technologies and Factory Automation (ETFA) (Vol. 1). IEEE, 2020, pp. 1571-1578.
- Wissbrock, P., David Pelkmann and B. Tölle: Automate Quality Prediction in an End-of-Line Test of a Highly Variant Production of Geared Motors – Discussion of a Full Concept. 6th European Conference of the Prognostics and Health Management Society (PHM Society), vrsl.2021.
- Redeker, Magnus, Klarhorst, Christian, Göllner, Denis, Quirin, Dennis, Wißbrock, Peter, Althoff, Simon und Hesse, Marc: Towards an Autonomous Application of Smart Services in Industry 4.0. In: 2021 26th IEEE International Conference on Emerging Technologies and Factory Automation (ETFA ), Schweden 2021. https://doi.org/10.1109/ETFA45728.2021.9613369
- Lenz, Cederic, Henke, Christian, and Trächtler, Ansgar: Anomaly detection in hot forming processes using hybrid modeling. In: 2021 26th IEEE International Conference on Emerging Technologies and Factory Automation (ETFA ), Schweden 2021. https://doi.org/10.1109/ETFA45728.2021.9613629
- Voigt, T., Migenda, N., Schöne, M., Pelkmann, D., Fricke, M, Schenck, W. und Kohlhase, M.: Advanced Data Analytics Platform for Manufacturing Companies. In: 2021 26th IEEE International Conference on Emerging Technologies and Factory Automation (ETFA ), Schweden 2021. https://doi.org/10.1109/ETFA45728.2021.9613499
- Schmidt, M, Lohweg, V.: Interval-based Interpretable Decision Tree for Time Series Classification. In: Schulte, H.: Proceedings - 31. Workshop Computational Intelligence: Berlin, 25. - 26. November 2021. KIT Scientific Publishing, 2021. https://doi.org/10.5445/KSP/1000138532
- Hesse, M.; Hunstig, M.; Timmermann, J. and Trächtler, A.: Batch Constrained Bayesian Optimization for Ultrasonic Wire Bonding Feed-forward Control Design. In Proceedings of the 11th International Conference on Pattern Recognition Applications and Methods, 2022, pages 31-42.
