Hybrid Computational Framework for Fault Detection in Coil Winding Manufacturing Process Using Knowledge Distillation

Izhar Oswaldo Escudero-Ornelas*, Divya Tiwari, Michael Farnsworth, Ze Zhang, Ashutosh Tiwari

*Corresponding author for this work

Research output: Chapter in Book or Report/Conference proceedingConference Proceedingpeer-review

Abstract

This paper proposes a hybrid computational framework for fault detection during the coil winding manufacturing process by using a combination of Discrete Event Simulation (DES) model with a Supervised Machine Learning (SML) algorithm. The conventional End of the Line (EoL) tests are insufficient in detecting faults during process resulting in increased manufacturing costs and lead times. The proposed methodology utilises a Knowledge Distillation (KD) approach to address the challenges associated with the technique and optimise the student model's performance by employing architecture search and data augmentation. Multiple SML algorithms were evaluated to determine their effectiveness in predicting faults during manufacturing. The random forest algorithm demonstrated superior performance due to its ability to handle complex data and identify the impact of interdependencies of process parameters on the final product quality. The method was validated by conducting physical experiments on a linear coil-winding machine, and the results indicated that the random forest algorithm has the potential to decrease simulation time from 2 minutes to less than a second. The proposed methodology has the potential to reduce manufacturing time, enhance stator quality, and ultimately improve their reliability and safety.

Original languageEnglish
Title of host publication2023 IEEE 21st International Conference on Industrial Informatics, INDIN 2023
EditorsHelene Dorksen, Stefano Scanzio, Jurgen Jasperneite, Lukasz Wisniewski, Kim Fung Man, Thilo Sauter, Lucia Seno, Henning Trsek, Valeriy Vyatkin
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781665493130
DOIs
Publication statusPublished - 2023
Externally publishedYes
Event21st IEEE International Conference on Industrial Informatics, INDIN 2023 - Lemgo, Germany
Duration: 17 Jul 202320 Jul 2023

Publication series

NameIEEE International Conference on Industrial Informatics (INDIN)
Volume2023-July
ISSN (Print)1935-4576

Conference

Conference21st IEEE International Conference on Industrial Informatics, INDIN 2023
Country/TerritoryGermany
CityLemgo
Period17/07/2320/07/23

Keywords

  • Discrete Event Simulation
  • Knowledge Distillation
  • modelling
  • Supervised Machine Learning
  • winding faults

Fingerprint

Dive into the research topics of 'Hybrid Computational Framework for Fault Detection in Coil Winding Manufacturing Process Using Knowledge Distillation'. Together they form a unique fingerprint.

Cite this