Adulteration detection model of Tea Oil research based on FTIR back-propagation neural network

Aiming at the research progress of Fourier transform infrared spectroscopy (FTIR) in edible oil adulteration detection in infrared spectroscopy, this paper uses FTIR combined with Backpropagation Neural Network (BPNN) to establish a tea oil blending Classification model for false detection. After performing convolution smoothing (Savitzky-Golay, SG) pretreatment on tea oil and adulterated oil mixed with soybean oil and corn oil in the spectral range of 4000-400cm-l, that is, the infrared spectra of 105 samples Principal Component Analysis (PCA) and random forest (RF) feature extraction were performed on the spectrum, and two camellia oil adulteration detection models, SG-PCA-BPNN and SG-RF-BPNN, were established. Comparing the classification accuracy of the two models in the test set samples, the analysis shows that the SG-RF-BPNN tea oil adulteration detection model achieves the best accuracy of 0.93, which meets the detection classification requirements and is a fast and easy detection of adulterated tea for the market oil provides a technical reference.