Near Infrared Spectroscopy Sugar Quantification in Intact Orange

Abstract

One of the critical points in the orange processing industry is the fruit quality evaluations by destructive methods. Therefore, non-destructive analysis can represent an advance in the system. The reflectance spectroscopy technique of near infrared (NIR) was employed with spectrophotometer model Spectrum 100N associated with chemometric method regression of Partial Least Squares (PLS) to quantify the total sugars concentration in intact oranges (cv. Valencia). Reference values for total concentration sugars were obtained by high-performance liquid chromatography with chromatograph equipped Refractive Index detector (RI). Spectral data were pre-processed by Normal Standard Variation (SNV), centering on the middle and 1st derivative with 21 points. Chemometric data analysis was done using the software Pirouette for the calibration models construction with regression method of Partial Least Squares (PLS). After this step, the prediction was performed. The model presented Pearson correlation coefficients (r) prediction of 0.88 and Root Mean Square Error of Prediction (RMSEP) of 14.12 g L-1. Preliminary results show spectroscopy technique as an alternative non-destructive analysis for oranges quality determination, showing a strong correlation and low prediction error value.