Detect Level of Methanol in Alcohol using Near-Infrared (NIR) Spectrometer Imaging

• Published in: 2023 International Conference on Advanced Technologies for Communications (ATC) pp. 488 - 492
• Main Authors: Huynh, Quoc T., Nguyen, Uyen D., Nguyen, Huy N.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 19-10-2023
• Subjects: Alcoholic beverages; Approximation algorithms; Classification algorithms; NIR spectrometry; Root mean square; Root Mean Square Error (RMSE); SCIO near-infrared spectrometer; Standard Normal Validate (SNV); Training; Training data; Vietnamese alcoholic beverages
• ISSN: 2162-1039
• DOI: 10.1109/ATC58710.2023.10318925

Near-infrared (NIR) spectrometry method had been utilized for the detection and verification of methanol content in 300 adulterated samples of methanol with ethanol, and Vietnamese's vodka. The characterization of beverages was analyzed by hyperspectral imaging. For this study, ethanol and alcoholic beverages samples adulterated with 5% changed steps in methanol content were used to validate the predictive algorithm. The spectral data of each alcoholic adulterated sample was collected and recorded by commercial-off-the-shelf(COTS) SCIO near-infrared spectrometer in the wavelength range of700 nm and 1070 nm. In addition, the differentiation of the adulterated samples was concentrated in the spectral shift from 900 nm to 920 nm. The Standard Normal Variate (SNV) algorithm was applied to scale the NIR spectra. Based on the different shapes of the NIR spectrum between 900 nm and 920 nm, the Root Mean Square Error (RMSE) was implemented to detect the unknown samples. The proposed algorithms successfully worked in the classification of alcoholic adulteration with approximately 98% accuracy.