Naslov
Prediction of used lubricating oils properties by infrared spectroscopy using multivariate analysis

Autori
Marinović, Slavica ; Jukić, Ante ; Doležal, Dubravka ; Špehar, Branka ; Krištović, Marko

Izvornik
Goriva i maziva (0350-350X) 51 (2012), 3; 205-215

Vrsta, podvrsta i kategorija rada
Radovi u časopisima, članak, znanstveni

Ključne riječi
used lubricating oil; prediction of properties; infrared spectroscopy; multivariate analysis; partial least squares regression; density; viscosity; pour point

Sažetak
Infrared spectroscopy (IR) in combination with partial least squares regression (PLS) algorithm was applied to develop calibration models for rapid and accurate simultaneous determination of the properties of used lubricating oils: density, viscosity, and pour point. Training sets were composed of hundred used lubricating oil samples. PLS calibration models for each property have been developed by using IR transmission spectra without spectral pretreatment. Cross validation technique was applied for validating models. The PLS models showed very good correlation with corresponding reference methods. Results indicated PLS models based on infrared spectra as practical and reliable analytical methods for routine monitoring of used lubricating oil.

Izvorni jezik
Hrvatski, engleski

Znanstvena područja
Kemijsko inženjerstvo

POVEZANOST RADA