摘要：This work aims to propose an alternative methodology to measure acidity index and chlorophyll content in ground soybean grains using an ultra-compact near-infrared (NIR) spectrometer. The experiment was conducted using 592 soybean samples for acidity index and 348 for chlorophyll content, harvested in southern Brazil. A partial least squares (PLS) model was developed for both parameters, being optimized by outlier detection and validated through its merit figures. For the acid index model the root mean squared error for calibration (RMSEC) was 0.1974 and validation (RMSEP) 0.1694, correlation coefficient (Rcal) 0.8589. Moreover, residual prediction deviation for calibration (RPDcal; 1.7661) and validation (RPDval; 1.8525) were also determined. The results for chlorophyll content were RMSEC (623.2972 mg kg-1 ) and RMSEP (621.9778 mg kg-1 ), Rcal (0.8709), RPDcal (1.5956) and RPDval (1.6295). Results suggest that the NIR spectroscopy can be used to provide a suitable tool for the measurement of the acidity index and the chlorophyll content of the soybean, bringing a significant improvement mainly on the speed of analysis, what is an important goal to industrial routine purposes, besides advantages as non-invasively and non-destructive characteristics, exempt from waste generation and chemical reagents, being a potential alternative methodology for online monitoring in the soybean processing industries.
