Process data-driven modelling for the robustification of shear-cutting-collar-drawing processes by means of effective tool surface design under consideration of edge crack sensitivity

Subproject SPP 2422

true" ? copyright : '' }

Project Overview

Principle Investigators:

Prof. Dr.-Ing. habil. Verena Kräusel

Prof. Dr. Agnes Koschmider

Project Team:

M.Sc. Jakub Korenek

M.Sc. Dominic Langhammer

Research Institutions:

Fraunhofer Institute for Machine Tools and Forming Technology IWU Chemnitz

Institute of Information Systems and Process Analytics, University of Bayreuth

Semi-finished Material(s):

S500MC

Manufacturing Processes:

Shear cutting (punching) and collar drawing

Motivation

  • Improvement of the process quality and robustness in the shearing-cutting-collar-drawing process by optimising the die surfaces to increase process reliability with regard to edge cracks that occur
  • Integration of forming domain knowledge into a hybrid process model and the application of advanced analysis methods to create a well-founded decision-making basis for process optimisation
  • Interdisciplinary collaboration between partners with different research competences for a holistic view of the process
  • Creation of comprehensible correlations through Explainable Artificial Intelligence (XAI)

Aims

As part of the research project Data4Collar, which is part of the SPP 2422, the interdisciplinary interaction of forming technology and data science is being used to investigate the formation of edge cracks and the effective design of the die surface by means of a digital representation of the process chain of shear cutting and collar drawing.

Initially, a data-driven modelling of the shearing-cutting-collar drawing process is pursued. The integration of in-line sensor technology into a test tool capable of continuous operation enables comprehensive recording and analysis of process data. This forms the basis for the development of a data-based model, which is extended to a hybrid model through the integration of formalised domain knowledge in order to promote explainability and transparency in the sense of XAI. By synthesising the data-driven model with domain knowledge through the integration of white/grey box models and the development of a hybrid model, the aim is to enable the explanation of correlations between process data and product quality in addition to transparency and interpretability.

Description of the process design and the steps of data-driven modeling

Working Program

Work Package

Description

WP1

Experimental and data analytical basics

WP2

Design of the continuous shear-cutting-collar-drawing process for in-process data acquisition

WP3

Analytical and numerical process analysis (domain knowledge)

WP4

Development of a data generator for the systematic evaluation of developed processes

WP5

Process data acquisition in continuous operation

WP6

Process data-driven modelling

WP7

Hybrid process model

Expected Results

The aim of this project is to create transferable system knowledge which, taking into account the sensitivity to edge cracking, enables an effective design of the mould working surfaces in the process design and thus contributes to a robustification of shearing-collar drawing processes.

Contact

Prof. Dr. Agnes Koschmider

Fraunhofer Institute for Applied Information Technology FIT

Wittelsbacherring 10

95444 Bayreuth

E-mail: agnes.koschmider@uni-bayreuth.de

Website: https://www.wi.fit.fraunhofer.de

Prof. Dr.-Ing. habil. Verena Kräusel

Fraunhofer Institute for Machine Tools and Forming Technology IWU

Reichenhainer Street 88

09126 Chemnitz

E-mail: info@iwu.fraunhofer.de

Website: https://www.iwu.fraunhofer.de/de/forschung/leistungsangebot/kompetenzen-von-a-bis-z/umformen.html

M.Sc. Jakub Korenek

Sheet Metal Working Department

Fraunhofer Institute for Machine Tools and Forming Technology

Reichenhainer Straße 88

09126 Chemnitz

E-Mail: jakub.korenek@iwu.fraunhofer.de

Website: www.iwu.fraunhofer.de

M.Sc. Dominic Langhammer

To the top of the page