Determination of alcohol degree in Chinese spirit with terahertz time-domain transmission spectroscopy

In this paper, we have determined water-ethanol mixtures and Chinese spirit with terahertz time-domain transmission spectroscopy (THz-TDS). Eleven samples of water-ethanol mixtures with varying alcohol degree from 0% to 100% (v/v), increased in 10%-steps, were used as the calibration set. Several Chinese spirit samples were used as the prediction set. We measured alcohol degree by alcohol meter, with measurement accuracy of 0.2% (v/v). Partial least square (PLS) model was employed for regression. The refractive index, absorption coefficient and mixed models by PLS regression together with alcohol degree were established, respectively. The predictive values of alcohol degree were very close to the reference values determined by alcohol meter. To assess the quality of the model, we calculated the correlation coefficient R-square, the root mean square errors of the calibration set (RMSEC) and the prediction set (RMSEP) of the three models. The results show that the absorption coefficient PLS regression model is the best method for determining alcohol degree of Chinese spirit with THz-TDS transmission measurement.