A method for quantifying and discriminating the colour index of refined sugar using a portable NIR spectrometer

Colour is one of the main quality parameters of sugar, playing a crucial role in determining its commercial value. This study employed near-infrared (NIR) spectroscopy combined with chemometric tools to assess sugar colour and classify it as refined or unrefined. Calibration models based on Partial Least Squares Regression (PLS) and Multiple Linear Regression (MLR) were developed using variable selection algorithms, such as the Successive Projections Algorithm (SPA), Genetic Algorithm (GA), and Interval SPA (iSPA). The spectra of 89 samples were obtained using two different devices, one bench-top and the other portable. The best predictive performances were achieved with the portable equipment. Classification models based on PLS-DA and LDA coupled with SPA and iSPA achieved up to 98.9 % accuracy and an RMSEP of 5.087. This approach stands out for its speed, low cost, and sample preservation, particularly when using portable devices.