umu.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Modeling of Surface Quality Defects Related to Production Process Data: A Data Analysis Approach in the Paint Shop at Volvo GTO Umeå
Umeå University, Faculty of Science and Technology, Department of Mathematics and Mathematical Statistics.
Umeå University, Faculty of Science and Technology, Department of Mathematics and Mathematical Statistics.
2016 (English)Independent thesis Advanced level (professional degree), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

The painting of cabs consists of a lot of processes that ultimately affect the quality of the coating. Even if the quality is monitored closely, knowledge of its relation to certain subprocesses is inadequate. Hence, this work aims to determine the relation between process parameters in the paint shop and the quality outcome of the primer coating at Volvo GTO Umeå, by applying statistical classification models and regression for count data. The implemented models were logistic regression, linear discriminant analysis, random forest, Poisson regression and negative binomial regression. The models were all compared and evaluated using cross validation with different error measures. The classification models some what struggled in separating cabs with zero defects from the ones with few defects, but did show tendencies of more accurate predictions as the number of actual defects on the cab increased. Concerning the

regression for count data, the negative binomial model estimated the number of cabs with different number of defects well. The quality of the available data was however not high enough to safely draw conclusions regarding the relation between the process parameters and the quality outcome. Hence, it was recommended that further research reproduces this work using data of higher quality.

Abstract [sv]

Målningen av en lastbilshytt innefattar många olika delprocesser som i slutändan påverkar lackeringens kvalitet. Även om kvaliteten övervakas noga så finns det idag ingen klar bild om hur dessa processer påverkar utfallet av målningen. Med detta i åtanke syftar detta arbete till att undersöka orsakssambandet mellan olika processparametrar och kvalitetsutfallet i måleriet på Volvo GTO Umeå, genom användning av klassificering och regression med räknedata. De modeller som implementerades var logistisk regression, linjär  diskriminantanalys, random forest, Poisson regression och negativ binomial regression. Prestandan av modellerna jämfördes och utvärderades med hjälp av korsvalidering. Resultatet visade att klassificeringsmodellerna  hade problem att separera hytter med noll fel från de med få fel, men att det fanns tendenser på en bättre prediktiv förmåga när antalet fel på hytten ökade. Resultatet av heltalsregressionen visade att negativ binomial regressionsmodellen uppskattade antalet hytter med olika antal defekter väl. Då kvaliteten på tillgängligt data inte var tillräckligt hög för att på ett säkert sätt dra slutsatser om sambandet mellan processparametrar och kvalitetsutfallet, rekommenderades det att i framtida studier återskapa detta arbete med data av högre kvalitet.

Place, publisher, year, edition, pages
2016. , 73 p.
National Category
Mathematics
Identifiers
URN: urn:nbn:se:umu:diva-125857OAI: oai:DiVA.org:umu-125857DiVA: diva2:972424
External cooperation
Volvo GTO, Umeå
Educational program
Master of Science in Engineering and Management
Supervisors
Examiners
Available from: 2016-10-06 Created: 2016-09-20 Last updated: 2016-10-06Bibliographically approved

Open Access in DiVA

No full text

By organisation
Department of Mathematics and Mathematical Statistics
Mathematics

Search outside of DiVA

GoogleGoogle Scholar

urn-nbn

Altmetric score

urn-nbn
Total: 126 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf