The capability of different spectroscopic techniques in estimation of adulteration of extra virgin olive oil with mild deodorized and refined olive oils was investigated. The visible absorption and ultraviolet-visible fluorescence spectra were distinctly different for oils of different quality grades. Differences in the near- and mid-IR absorption spectra were less pronounced and could only be revealed by principal component analysis. The calibration models for estimation of adulterant oil in the blends obtained using partial least squares (PLS) and principal component regression (PCR) showed similar performance using all of the examined spectroscopic techniques. The simultaneous analysis of combined spectra using concatenated matrix did not improve the calibration results. The prediction parameters for the set of 20 samples were: R2>0.96, the relative error of prediction (REP) was in the range 5.6-12.0%. The RPD (ratio of the SD of reference values to RMSEP) were in the range of 5.1-10.9, indicating that all models were at least suitable for quality control.
Spectroscopic techniques and chemometrics in analysis of blends of extra virgin with refined and mild deodorized olive oils
Paradiso V. M.;
2015-01-01
Abstract
The capability of different spectroscopic techniques in estimation of adulteration of extra virgin olive oil with mild deodorized and refined olive oils was investigated. The visible absorption and ultraviolet-visible fluorescence spectra were distinctly different for oils of different quality grades. Differences in the near- and mid-IR absorption spectra were less pronounced and could only be revealed by principal component analysis. The calibration models for estimation of adulterant oil in the blends obtained using partial least squares (PLS) and principal component regression (PCR) showed similar performance using all of the examined spectroscopic techniques. The simultaneous analysis of combined spectra using concatenated matrix did not improve the calibration results. The prediction parameters for the set of 20 samples were: R2>0.96, the relative error of prediction (REP) was in the range 5.6-12.0%. The RPD (ratio of the SD of reference values to RMSEP) were in the range of 5.1-10.9, indicating that all models were at least suitable for quality control.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
