Predictive model of moisture content in dry parchment coffee beans using near-infrared spectroscopy (FT-NIR)

Abstract

Moisture content is a key quality parameter in grain storage. Near infrared spectroscopy (NIRS) is a non-destructive technique, with fast and simple measurements, that can be a method to implement for monitoring this parameter. The objective of this research was to evaluate the NIRS technique for the prediction of the moisture content of dry samples of dry parchment coffee (DPC), for this it was necessary to carry out the study in different drying times in order to obtain several points of moisture of the grain that could exist, once the spectrum is taken in the FT-NIR, it is taken to the oven so that through the gravimetric method the real moisture of the grain can be known. The Spectrum Quant software was used to develop the predictive mathematical model by means of principal component regression (PCR) and partial least squares (PLS), using a set of randomly divided data for calibration and validation. The results showed that a better goodness of fit was found with a PLS model and the application of the baseline and second derivative correction, obtaining a coefficient of determination (R2) of 0.99 and a predictive standard error (SEP) of 0.34. Finding a good correlation between the real data with those estimated by the NIR equipment, emerging a fast and practical way in full-scale monitoring in DPC grain moisture control.