The Identification of Acetic Acid-Ethanol Mixture Using Gas Sensor Array and Ensemble Regression

Abstract

Identification of acetic acid-ethanol mixtures using a commercial gas sensor array equipped with ensemble regression has been carried out. The gas sensor analysis was simple, rapid, and fast since it did not require any sample preparation. A quantitative analysis of the acetic acid-ethanol mixture was carried out to determine the sensitivity and selectivity of the sensor in distinguishing the concentration of the acetic acid and ethanol mixture. This study focuses on the coefficient of determination of 80% of the calibration data set and recovery of 20% of the testing data set. The models showed excellent performance,specifically, the Bagging and Random Forest r2 for the ethanol calibration data reached 0.91 and 0.94, respectively. The corresponding ethanol test recoveries were 99.95% and 97.84%, indicating the robustness of the model in accurately predicting ethanol concentration. Acetic acid test recoveries were 100.56% and 101.38% with r2 of 0.89 and 0.93 for Bagging and Random Forest regression, respectively. Hence, the commercial gas sensor array equipped with ensemble regression can be applied to the quantification of the acetic acid – ethanol mixture and demonstrate opportunities for the practical use of this gas sensor array in analyzing real samples, i.e. human breath or environmental monitoring samples.