Feasibility of Visible/Near Infrared Spectroscopy in order to detect pomace olive oil fraud with LDA and SVM detection methods

Articles in Press

Document Type : Research Article

Authors

1 Biosystem engineering. تربیت modares university, tehran , iran

2 Tarbiat Modares University

3 Associate Professor, Agricultural Engineering Research Institute, Agricultural Research, Education and Extension Organization (AREEO)

Abstract

Olive is a Mediterranean fruit that is cultivated mainly for its oil. Extra virgin olive oil has always been the attention and demand of users. Therefore, fraud in virgin and extra virgin olive oils is observed by adding valuable food oils and lower prices such as canola, sunflower, olive pomace, etc. Fraud detection with conventional methods such as gas chromatography is difficult, time-consuming, and requires sample and operator preparation. For this reason, the use of non-destructive technologies for fraud detection is important. In this research, detection of adulteration of pomace oil (olive pomace oil) was investigated using visible/near-infrared spectroscopy (Vis/NIR) technology. Three types of extra virgin olive oil, virgin and refined olive oil were used for sampling. Moreover, the samples were made in six categories of pure, 11, 20, 33, 50 and 100% fraud. Each treatment was prepared and tested in ten samples. Next, in order to analyze the qualitative features and classify the data extracted from the spectrometer, pattern recognition methods including linear discriminant analysis (LDA) and support vector machine (SVM) were used. The obtained results showed that visible/near infrared (Vis/NIR) spectroscopy is able to distinguish olive oil samples based on different percentages of pomace adulteration. Although the LDA method was able to classify olive oil samples with acceptable accuracy according to the adulteration rate, the SVM method had a better accuracy and fit with a training accuracy of 96.69% and a validation of 94.21%. According to the results, the linear function was suggested as the best function for building the classification models using the SVM method.

Keywords

Main Subjects

Innovative Food Technologies

Articles in Press, Accepted Manuscript
Available Online from 27 May 2024

Files

History

  • Receive Date: 19 October 2023
  • Revise Date: 14 May 2024
  • Accept Date: 27 May 2024

Share

How to cite

Statistics