Authors:
Alexander Gerling
1
;
Ulf Schreier
2
;
Andreas Hess
2
;
Alaa Saleh
1
;
Holger Ziekow
2
and
Djaffar Ould Abdeslam
3
Affiliations:
1
Business Information Systems, Furtwangen University of Applied Science, 78120 Furtwangen, Germany, IRIMAS Laboratory, Université de Haute-Alsace, 68100 Mulhouse, France, Université de Straßbourg, France
;
2
Business Information Systems, Furtwangen University of Applied Science, 78120 Furtwangen, Germany
;
3
IRIMAS Laboratory, Université de Haute-Alsace, 68100 Mulhouse, France, Université de Straßbourg, France
Keyword(s):
Reference Model, Machine Learning, Assembly Line, Manufacturing, Requirements.
Abstract:
The importance of machine learning (ML) methods has been increasing in recent years. This is also the reason why ML processes in production are becoming more and more widespread. Our objective is to develop a ML aided approach supporting production quality. To get an overview, we describe the manufacturing domain and use a visualization to explain the typical structure of a production line. Within this section we illustrate and explain the as-is process to eliminate an error in the production line. Afterwards, we describe a careful analysis of requirements and challenges for a ML system in this context. A basic idea of the system is the definition of product testing meta data and the exploitation of this knowledge inside the ML system. Also, we define a to-be process with ML system assistance for checking production errors. For this purpose, we describe the associated actors and tasks as well.