A Machine Learning Based Method to Efficiently Analyze the Cogging Torque under Manufacturing Tolerances

Andrea Reales, Werner Jara, Gabriel Hermosilla, Carlos Madariaga, Juan Tapia, Gerd Bramerdorfer

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

Abstract

This paper addresses a new technique based on machine learning which reduces the number of evaluations required to perform robustness analysis of permanent magnet synchronous machines. This methodology is based on the logical behavior of possible faulty magnet combinations produced by manufacturing tolerances. Groups of faulty combinations with a similar structure and cogging output are identified by means of a fuzzy-logic algorithm. Subsequently, only a single faulty combination of each group needs to be evaluated through the finite element method, which severely decreases the computational burden of the tolerance analysis. A 6-slot 4-pole and a 9-slot 6-pole machine were subject to tolerance analysis considering the displacement of the magnets. Both machines were evaluated through the proposed method and the results were validated by means of the finite element method (FEM).

Original languageEnglish
Title of host publication2021 IEEE Energy Conversion Congress and Exposition, ECCE 2021 - Proceedings
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1353-1357
Number of pages5
ISBN (Electronic)9781728151359
DOIs
StatePublished - 2021
Externally publishedYes
Event13th IEEE Energy Conversion Congress and Exposition, ECCE 2021 - Virtual, Online, Canada
Duration: 10 Oct 202114 Oct 2021

Publication series

Name2021 IEEE Energy Conversion Congress and Exposition, ECCE 2021 - Proceedings

Conference

Conference13th IEEE Energy Conversion Congress and Exposition, ECCE 2021
Country/TerritoryCanada
CityVirtual, Online
Period10/10/2114/10/21

Keywords

  • Fuzzy logic
  • machine learning
  • permanent magnet
  • robustness
  • tolerance analysis

Fingerprint

Dive into the research topics of 'A Machine Learning Based Method to Efficiently Analyze the Cogging Torque under Manufacturing Tolerances'. Together they form a unique fingerprint.

Cite this