Abstract
Oil oxidation is an undesirable series of chemical reactions involving oxygen that degrades the quality of an oil. Oxidation eventually produces rancidity in oil, with accompanying off-flavors and smells. An electronic nose was used in this study to detect the adulterations in edible oils. The acidity, peroxide, anisidine and Totox values of the edible oil samples were measured according to the official American Oil Chemist Society (AOCS) standard. The results were analyzed using cluster analysis, principal component analysis, support vector machine, quadratic discriminant analysis, and Partial least squares regression technique. In the sensor array, the TGS2602, and MQ136 sensors had the highest values of the Loudness coefficient and the MQ9, TGS822, TGS813, and TGS2620 had the lowest values. Based on the results obtained, the accuracy of the three methods; Support vector machine (SVM), Quadratic discriminant analysis and Partial least squares were 97%, 98.33%, and 100%, respectively. The results for the linear vector kernel support machine, training accuracy and validation for C-SVM and Nu-SVM were 98, 97, 97 and 95%, respectively. The results also indicated that the proposed method can be used as an alternative to the official AOCS methods to innovatively detect the edible oil oxidation with high accuracy