Fast Discrimination of Irradiation Doses of Rice Flour Based on Vis/NIR Spectroscopy

Abstract

Irradiation has been used for food preservation in many areas, however the high dose of irradiation is able to influence the safety of food. Rice flour is a staple food hackneyed in our lives and a part of them are processed by irradiation. Here we offer a new method for fast discriminating the rice flour with different doses of irradiation based on visible-near infrared spectroscopy. We dealt with the rice flour on four dose level before the experiment at first, then used the handheld Vis/NIR spectrometer to obtain the reflection spectrum curve and raw spectral data of the rice flour with different doses of irradiation. Next principal component analysis (PCA) was used for analyzing the clustering. Among the rest, six principal components (PCs) were selected based on accumulative reliabilities (AR), and these six PCs would be taken as the inputs of the back-propagation artificial neural network (BP-ANN), to establish the identification model of different rice samples. The results showed that the identification rate achieved 100% under the standard deviation of ± 0.1. It could be concluded that PCA combined with BP-ANN was an available method for recognition and discrimination of rice flour with different doses of irradiation.