Open AccessDetermination of the Real Contents of Olive Oil in Blend Oils by near Infrared Spectroscopy
Author(s) -
Zihao Cai,
Qiang Xu,
Xingxing Yang,
Zhui Hu,
Xing Zheng,
Jiafan Zhang
Publication year - 2020
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/1673/1/012009
Subject(s) - olive oil , correlation coefficient , support vector machine , mean squared error , regression analysis , canonical analysis , coefficient of determination , mathematics , linear regression , canonical correlation , predictive modelling , statistics , biological system , artificial intelligence , chemistry , computer science , food science , biology
Aiming at the determination of real contents of olive oil in blend oils, near infrared spectroscopy and support vector machine were combined to establish regression models for the determination of olive oil contents. The data were fused on the feature level with canonical correlation analysis to improve the prediction performance of the models. The results showed that the SVM regression models could effectively predict the real contents of olive oil in 135 samples containing 2%-20% olive oil. Among these models, the highest correlation coefficient R2 was 99.74%, and the root mean square error of the prediction set was 0.08. In addition, the prediction performances of some regression models were deteriorated after the curves were smoothed with the Savitzky-Golay method. But, the prediction performances were improved in the models established with data fused with the canonical correlation analysis approach.