Identification of olive oil in vegetable blend oil by one-dimensional convolutional neural network combined with Raman spectroscopy

Abstract

Vegetable blend oil is popular because it has more balanced nutrients, which also provides a target for fraud. In this paper, One-dimensional convolutional neural network (1D CNN), support vector machine (SVM) and improved partial least squares discriminant analysis (PLS-DA) combined with Raman spectroscopy were used to identify corn olive blend oil, peanut olive blend oil and corn peanut olive blend oil. The overall performance of 1D CNN model based on 500 Raman spectral data of three types of vegetable blend oil is significantly higher than that of SVM and PLS-DA. By comparing the changing trend of loss curve before and after data preprocessing, it is proved that the data preprocessing process can accelerate the convergence speed of 1D CNN model. Finally, partial least squares regression (PLSR) model was established to identify the content of olive oil in vegetable blend oil. The results show that 1D CNN combined with Raman spectroscopy has great application potential in the field of vegetable oil identification.