FT-IR and Raman Spectroscopy Data Fusion with Chemometrics for Simultaneous Determination of Chemical Quality Indices of Edible Oils During Thermal Oxidation

Huan Liu,Yi Chen,Ce Shi,Xinting Yang,Donghai Han

DOI: https://doi.org/10.1016/j.lwt.2019.108906

2020-01-01

Abstract:A rapid, non-destructive and robust method for measuring peroxide values (PVs) and acid values (AVs) of common edible oils (soybean, rapeseed, sunflower and peanut) simultaneously under various thermal oxidation was explored by FT-IR and Raman spectroscopy data fusion strategy. Uninformative variable elimination (UVE) and successive projections algorithm (SPA) methods were used for feature variables extraction, quantitative models for prediction of chemical quality indices were established using partial least squares regression (PLSR) algorithm. The bands associated with vibration of C=O and C=C stretching were highly correlated with PVs and AVs, data fusion of the two spectra showed the best modeling results when variables identified by SPA were used. For modeling of PVs, the resulting Rc2 and Rp2 were 0.964 and 0.939, RMSEC and RMSEP were 0.060 and 0.080. For modeling of AVs, the resulting Rc2 and Rp2 were 0.955 and 0.919, RMSEC and RMSEP were 0.025 and 0.027.