Machine learning based defect detection in a low automated assembly environment

Procedia CIRP, vol. 104, pp. 265–270

Abstract

Machine learning algorithms offer great potential for quality control in manufacturing. A great share of manufacturing operations are assembly tasks, which are often performed manually. Unlike automatic processes, manual assembly offers a limited amount of recorded data, that can be used as an input for a machine learning model. This study investigates, if the assembly times that are recorded by the conveyor technology, are enough to train machine learning models and how different models perform on this limited data input. To conduct this investigation, the study utilizes a real data set from the automotive industry.

Authors 4

1. G. Schuh Aachen Laboratory for Machine Tools and Production Engineering (WZL)

   RWTH Aachen University

   Affiliation as printed

   Laboratory for Machine Tools and Production Engineering (WZL) of RWTH Aachen University, Aachen, NRW 52074 Germany

2. Andreas Gützlaff Aachen Laboratory for Machine Tools and Production Engineering (WZL)

   RWTH Aachen University

   Affiliation as printed

   Laboratory for Machine Tools and Production Engineering (WZL) of RWTH Aachen University, Aachen, NRW 52074 Germany

3. Katharina Thomas Aachen Laboratory for Machine Tools and Production Engineering (WZL)

   RWTH Aachen University

   Affiliation as printed

   Laboratory for Machine Tools and Production Engineering (WZL) of RWTH Aachen University, Aachen, NRW 52074 Germany

4. Martin Welsing corresponding Aachen Laboratory for Machine Tools and Production Engineering (WZL)

   RWTH Aachen University

   Affiliation as printed

   Laboratory for Machine Tools and Production Engineering (WZL) of RWTH Aachen University, Aachen, NRW 52074 Germany