Qualitative identification of edible oil storage period using a homemade portable electronic nose combined with multivariate analysis.

BACKGROUND Edible oil storage period is one of the important indicators to evaluate the intrinsic quality of edible oil. This study attempted to develop a portable electronic nose device for qualitative identification of edible oil storage period. First, four metal oxide semiconductor (MOS) gas sensors, which were TGS2600, TGS2611, TGS2620, and MQ138, were selected to prepare a sensor array to assemble a portable electronic nose device. Second, the homemade portable electronic nose device was used to obtain the odor change information of edible oil samples in different storage periods, and the sensor features were extracted. Finally, three pattern recognition methods, which were linear discriminant analysis (LDA), K-nearest neighbors (KNN) and support vector machines (SVM), were compared to establish a qualitative identification model of edible oil storage period. The input features and related parameters of the model were optimized by a five-fold cross-validation during the process of model establishment. RESULTS The research results showed that the recognition performance of the nonlinear SVM model was significantly better than that of the linear LDA and KNN models, especially in terms of generalization performance, which had a correct recognition rate of 100% when predicting independent samples in the prediction set. CONCLUSION The overall results demonstrate that it is feasible to apply the homemade portable electronic nose device with the help of the appropriate pattern recognition methods to achieve fast and efficient identification of edible oil storage period, which provides an effective analysis tool for the quality detection of edible oil storage. This article is protected by copyright. All rights reserved.