<div class="csl-bib-body">
<div class="csl-entry">Pfeiffer, P., Ronai, B., Vorlaufer, G., Dörr, N., & Filzmoser, P. (2022, September 15). <i>Prediction of engine oil degradation based on FTIR spectroscopic data</i> [Conference Presentation]. Symposium 2022 der Österreichischen Tribologischen Gesellschaft (ÖTG), Wr. Neustadt, Austria. http://hdl.handle.net/20.500.12708/152959</div>
</div>
-
dc.identifier.uri
http://hdl.handle.net/20.500.12708/152959
-
dc.description.abstract
By means of artificial oil alteration, large quantities of degraded oils can be produced under laboratory-controlled conditions. An understanding of the relationship between different degradation mechanisms is essential for the choice of parameters for artificial alteration. As FTIR spectroscopy is widely used for the analysis of fresh and degraded lubricants, we demonstrate an analysis procedure based on 58 FTIR spectra of automotive engine oils in different conditions (fresh, vehicle, artificial alteration) to quantify the connection between degradation mechanisms. First, an automatic preprocessing step for filtering non-informative variables is introduced. Then, experts’ knowledge is integrated with a statistical model using a weighted LASSO regression. From this, a quantitative relation between degradation pathways in engine oils is derived. This leads to sparse models with high predictive power between the duration in artificial large-scale alteration and mileage in a vehicle.
en
dc.description.sponsorship
FFG - Österr. Forschungsförderungs- gesellschaft mbH
-
dc.language.iso
en
-
dc.subject
chemometrics
-
dc.subject
LASSO regression
-
dc.subject
variable selection
-
dc.subject
FTIR spectroscopy
-
dc.title
Prediction of engine oil degradation based on FTIR spectroscopic data
-
dc.type
Presentation
en
dc.type
Vortrag
de
dc.contributor.affiliation
AC2T Research (Austria), Austria
-
dc.relation.grantno
RV-TUW-01
-
dc.type.category
Conference Presentation
-
tuw.project.title
Merkmalerkennung in mehrdimensionalen Datensätzen von geschmierten Kontakten
