Qualitative and quantitative detection of camellia oil adulteration based on electronic nose

Abstract

Objective: This study aims to realize the adulteration detection of camellia oil mixed with rapeseed oil, soybean oil and corn oil. Methods: The electronic nose detection platform was used for the adulteration detection of camellia oil mixed with different proportion of rapeseed oil, soybean oil and corn oil. Firstly, the linear discriminant analysis （LDA） and support vector machine （SVM） were used for the qualitative identification of camellia oil adulteration. Then multilayer perceptron （MLP） and partial least squares regression （PLSR） were used to establish quantitative prediction models for camellia oil adulteration. Results: The accuracy of SVM for qualitative authentication was higher than that of LDA, and the average precision rate, average recall rate and average F1-score were 94.85%, 96.11% and 95.34%, respectively, which were 5.17%, 4.44% and 5.29% higher than those of LDA. For quantitative prediction, MLP outperforms PLSR. In particular, the determination coefficients of MLP were 0.98, 0.99 and 0.98, and the root mean square errors were 4.02%, 1.45% and 3.74%, respectively, for camellia oil mixed with rapeseed oil, soybean oil and corn oil. Conclusion: The SVM-based identification model and MLP-based prediction model can effectively detect oil adulteration by using electronic nose platform.

Keywords

• electronic nose
• camellia oil
• adulteration
• qualitative identification
• quantitative prediction