h1

%0 Thesis
%A Mann, Samuel Micha
%T Datenbasierte Erfassung und Regelung transienter Qualitätsmerkmale beim Metall-Schutzgasschweißen
%V 2024,2
%I RWTH Aachen University
%V Dissertation
%C Düren
%M RWTH-2024-04642
%@ 978-3-8440-9478-7
%B Aachener Berichte Fügetechnik
%P 1 Online-Ressource : Illustrationen
%D 2024
%Z Druckausgabe: 2024. - Auch veröffentlicht auf dem Publikationsserver der RWTH Aachen University
%Z Dissertation, RWTH Aachen University, 2023
%X Gas metal arc welding technology is confronted with a high demand for quality and competence, one that presently can only be fulfilled by highly trained but poorly available specialists. The overarching goal of this work is therefore to transfer parts of the process competence into the welding system. The focus is further on the compliance with quality features of the non-volatile product quality (weld seam geometry) and the volatile process quality (welding fume emission). With the acquisition of the transient process and product quality followed by the closing of the quality control loop, two research objectives are then specified and investigated using the introduced conceptof data-based quality control. The first part of this work considers suitable sensor technology as well as data pro-cessing to capture meaningful process features that can be used for statistical modeling of quality features. Hybrid process imaging is used to investigate an approach for simultaneously capturing process features from the joint, process zone, and weld seam in one sensor system. Here, the position of the joint can be detected at just 1-2°mm from the weld pool, thus minimizing the lead time error compared to conventional sensor systems. Electrical and optical time series are characterized by high availability yet require a distinct degree of modeling. With an introduced feature extraction methodology, time series are made usable, which is demonstrated using neural networks to identify process deviations. Building on the previously studied sensor technology and feature extraction, the second part demonstrates data-based quality control using two case studies for the control of fillet weld flanks (product quality) and welding fume emission (process quality). In the welding fume emission study, the FER is modeled using current and voltage time series and reduced by 12-40
%F PUB:(DE-HGF)11 ; PUB:(DE-HGF)3
%9 Dissertation / PhD ThesisBook
%R 10.18154/RWTH-2024-04642
%U https://publications.rwth-aachen.de/record/985472