Publikation
Concept Drift in Industrial Material Processing
Pascal Marijan; Sebastian Igel; Tatjana Legler; Achim Wagner; Martin Ruskowski
In: 2025 11th International Conference on Control, Decision and Information Technologies (CoDIT). International Conference on Control, Decision and Information Technologies (CoDIT-2025), Pages 647-652, Vol. 1, IEEE, 2025.
Zusammenfassung
Process optimization is particularly essential in areas with high energy demand, such as industrial material processing. One approach to this problem is the application of neural networks. However, in plants of large dimensions used in industrial material processing, material wear is very high which again results in a change of distribution of the data. Therefore, the performance of the neural networks and, thus, the optimization of the specific energy decreases over time. Against the context of a change in the distribution of data, the aim of the present study is to improve the modeling, prediction and optimization of industrial material processing technologies. Using approaches from continual learning in the field of concept drift, changing conditions are captured and implemented to improve the performance. In the field of industrial material processing, this paper represents the first contribution to adaptation to changing conditions. The proposed approach is validated on real data.