Anomaly correction driven by extracted expert knowledge and learning systems (ADELeS)

Production systems are increasingly characterized by a higher degree of automation and connectivity. The resulting highly complex production systems contribute to a more information-intensive work environment on the shop floor. The thereby increased knowledge and competence requirements establish the need for shop-floor operator support through assistance systems.

In the ADELeS project, an assistance system is developed for the AI-supported detection and elimination of anomalies in production processes. This intends to improve production quality, reduce the effects of skilled worker shortages, and strengthen the location and future security of Bavarian production facilities. The targeted quality assurance procedure links explainable learning systems with expert knowledge, which is extracted based on experience and data. The assistance system provides both passive assistance, in the form of suggested correction measures, and active assistance – direct intervention in machine control. The methodology is developed and evaluated based on two real application scenarios in an extrusion processes, but is also transferable to other production processes.

The ADELeS Project is a cooperation between Universität Augsburg, REHAU, Xitaso GmbH, and the Friedrich-Alexander-Universität Erlangen-Nürnberg. The project is enabled by the Bavarian State Ministry for Economic Affairs, Energy and Technology and part of the funding line „Digitization“ of the Bavarian Collaborative Research Program (BayVFP).