Clustering of Milling Strategies Using Machine Internal Data

IEEE International Conference on Intelligent Engineering Systems (INES), pp. 000077–000082

Abstract

Increasing the efficiency of milling processes demands the automated evaluation of time series data collected from the machine. Solutions based on Artificial Intelligence are becoming popular for this purpose, which require tailored datasets for model training. The programmed strategies for milling particular features divide the process into subsections reflected by repeating patterns in multiple signals. In this paper, a clustering model is developed, which differentiates similar milling strategies based on these patterns. The model is combined with a classification model, so that recurring milling strategies of future processes are continuously assigned with corresponding cluster labels. Storing these labels in combination with the raw data enables finding customized data in growing data lakes.

Authors 4

1. Christian Brecher Aachen Lab. for Machine Tools and Production Engineering (WZL)

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University,Lab. for Machine Tools & Production Engineering (WZL),Aachen,Germany

2. Janis Ochel Aachen Lab. for Machine Tools and Production Engineering (WZL)

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University,Lab. for Machine Tools & Production Engineering (WZL),Aachen,Germany

3. Muhammad Mehmood Ahmed Aachen Lab. for Machine Tools and Production Engineering (WZL)

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University,Lab. for Machine Tools & Production Engineering (WZL),Aachen,Germany

4. Marcel Fey Aachen Lab. for Machine Tools and Production Engineering (WZL)

   RWTH Aachen University

   Affiliation as printed

   RWTH Aachen University,Lab. for Machine Tools & Production Engineering (WZL),Aachen,Germany