Online detection of distilled spirit quality based on laser Raman spectroscopy
In order to detect the online alcohol degree and methanol content in a distilled spirit accurately, a new system is proposed based on the model of a quantitative relationship between the alcohol concentration and its characteristic peak area to predict the alcohol degree utilizing the Gaussian fitting of Raman spectroscopy and the least square method. Moreover, the difference spectrum method was employed to determine the peak value of methanol located at 1074.6 cm−1 (caused by bending vibration of C–O–H) for building the predictive model of methanol content in distilled spirits. The experimental results showed that the RMSEP (root-mean-square error of prediction) was 0.131 for the alcohol degree, the response time for switching to different samples was almost 1.8 s and the most relative deviation was <1.1% for continuous detection of same distilled spirit sample; the RMSEP for methanol concentration in distilled spirit was 0.122. The online detection method proposed based on Gaussian fitting is of great significance to the online measurement of the alcohol degree and trace methanol in distilled spirits.