Prediction of Enological Parameters in Chinese Rice Wine Using Chemometrics and FT-NIRS

Abstract

The applicability of Fourier transform near-infrared spectroscopy (FT-NIRS) together with chemometrics methods for determining enological parameters (alcohol content, available acidity (pH), and total acidity (TA)) of Chinese rice wine was investigated. Transmission spectra of 244 Chinese rice wine samples were collected in the spectral range of 800-2500 nm in a 1 mm path-length rectangular quartz cuvette. Calibration models relating near-infrared (NIR) spectra to alcohol content, pH, and TA were developed based on three partial least squares regression (PLSR) methods: linear PLSR (LPLSR), concentration-weighted PLSR (WPLSR), and nonlinear PLSR (NPLSR). The prediction performance of the calibration models in different wavelength regions and different preprocessing methods (original, first, and second derivative, as well as Savitzky-Golay smoothed spectra) was also investigated. The best models gave root mean square errors of prediction (RMSEP) of 0.112% (v v-1), 0.059, and 0.067 g L-1 and correlation coefficients for validation (rval) of 0.966, 0.955, and 0.970 for alcohol content, pH, and TA, respectively. The results demonstrated that FT-NIRS combined with chemometrics is a promising technique for predicting enological parameters of Chinese rice wine.