h1

TY  - THES
AU  - Wiesch, Marian
TI  - Lernende Prozesskraftmodellierung zur Planung und Überwachung in der Einzelteilfertigung für Fräsanwendungen; 1. Auflage
VL  - 2025,27
PB  - Rheinisch-Westfälische Technische Hochschule Aachen
VL  - Dissertation
CY  - Aachen
M1  - RWTH-2026-00064
SN  - 978-3-98555-324-2
T2  - Ergebnisse aus der Produktionstechnik
SP  - 1 Online-Ressource : Illustrationen
PY  - 2025
N1  - Druckausgabe: 2025. - Auch veröffentlicht auf dem Publikationsserver der RWTH Aachen University 2026. - Weitere Reihe: Werkzeugmaschinen. - Weitere Reihe: Edition Wissenschaft Apprimus
N1  - Dissertation, Rheinisch-Westfälische Technische Hochschule Aachen, 2025
AB  - The dissertation develops a novel, learning approach to predicting process forces in NC milling, thereby addressing key challenges in modern single-part production. In view of high component variance, changing boundary conditions, and increasing quality requirements, classic process force models are increasingly reaching their limits: They are often too complex in their parameterization, only transferable to a limited extent, or do not sufficiently reflect essential relationships. The aim of this work is therefore to develop a data-based process force model that can be used across processes and components and combines the advantages of artificial neural networks with the established knowledge of classic modeling approaches. To this end, the constantly growing database of the CAx process chain is being systematically developed. Information from CAD, CAM, NC, and CAQ is contextualized, structured, and refined from a tool perspective to create a consistent, comprehensive picture of the manufacturing process. Based on this, a learning, hybrid overall model is created that precisely captures complex relationships between process variables and resulting process forces. This model is integrated into a milling simulation demonstrator developed at WZL to validate its practical usefulness in an industrial environment. The focus is on two use cases: virtual, a priori verification of component quality in CAM planning and location-based live process monitoring by comparing simulated and actual measured process forces. The validation results clearly show that the learning approach outperforms conventional models in terms of accuracy, robustness, and transferability. The dissertation thus contributes to the further development of digital manufacturing systems and opens up new potential for more efficient, flexible, and reliable machining of complex components.
LB  - PUB:(DE-HGF)11 ; PUB:(DE-HGF)3
DO  - DOI:10.18154/RWTH-2026-00064
UR  - https://publications.rwth-aachen.de/record/1024444
ER  -