Emprego da espectroscopia no infravermelho próximo e quimiometria na avaliação de grãos de sorgo para a produção de ração animal

Abstract

This study investigated the use of near-infrared (NIR) spectroscopy combined with multivariate classification models to develop a screening method capable of distinguishing sorghum grains (Sorghum bicolor (L.) Moench) infested by insects from non-infested grains, with a focus on animal feed production. The research was motivated by the importance of sorghum as a cereal of high economic and productive value, used in human food, animal nutrition, and by the need for rapid methods to detect early post-harvest infestations. A total of 101 sorghum grain samples infested with unsexed adult insects of the species Sitophilus zeamais, maintained for periods ranging from 7 to 70 days, were analyzed, along with 95 non-infested samples, all originating from different cities in the states of Minas Gerais, Bahia and Goiás. Spectra were collected using a portable NIR device, and the spectral data were used to build supervised classification models based on PLS-DA (partial least squares discriminant analysis) and SIMCA (soft independent modeling of class analogy). The PLS-DA model showed better performance, with 94.1% sensitivity, 100% specificity, 100% concordance, and 100% accordance for the test set. The results demonstrate that combining spectroscopy with chemometric analysis provides an effective, rapid, economical, and non-destructive alternative for detecting infestations in sorghum grains, contributing to the preservation of the quality of stored grains.