Bitte verwenden Sie diesen Link, um diese Publikation zu zitieren, oder auf sie als Internetquelle zu verweisen: https://hdl.handle.net/10419/323704 
Erscheinungsjahr: 
2024
Quellenangabe: 
[Journal:] Journal of Intelligent Manufacturing [ISSN:] 1572-8145 [Volume:] 36 [Issue:] 6 [Publisher:] Springer US [Place:] New York, NY [Year:] 2024 [Pages:] 3869-3885
Verlag: 
Springer US, New York, NY
Zusammenfassung: 
In the automotive industry, the process of deep drawing is used for producing most of the outer surface panels. There, surface defects can occur while stamping the part. This paper proposes an area of interest (AOI) algorithm to filter possible surface deflection areas of finite element method (FEM) simulation results. The FEM is well established in the area of sheet metal forming and has shown accurate results in showing surface defects like waviness and sink marks. These two defect types are also the targeted systematic defects. In these deep drawing simulations, every manufacturing step of the sheet metal is calculated and the resulting stresses and strains are analyzed. The paper presents a newly developed post processing method for detecting surface in-corrections on basis of FEM simulation results. The focus of the method is to be independent of an experts knowledge. It should be able to be used by a wide range of non-expert applicants, unlike other post-processing methods know in today’s literature. A comparison between several machine learning (ML) approaches is made. It is shown, that the developed method outperforms current state of the art approaches in terms of the recall rate. In addition, a contour tree dataset of a FEM simulation in combination with an ML approach can be successfully used to learn a multidimensional relationship between the nodes.
Schlagwörter: 
Machine learning
Graph neural network
Finite element method
Deep drawing
Contour tree
Persistent Identifier der Erstveröffentlichung: 
Creative-Commons-Lizenz: 
cc-by Logo
Dokumentart: 
Article
Dokumentversion: 
Published Version

Datei(en):
Datei
Größe





Publikationen in EconStor sind urheberrechtlich geschützt.