A simple method using Fourier Transform Near-Infrared (FT-NIR) was investigated to determine quality of sugarcane/sugarcane juice in terms of â—¦brix value (total soluble solids), sucrose content (polarization %) and purity. Correlation equations for the chemometric models to predict the quality parameters of sugarcane juice were developed using partial least square (PLS) regression and were evaluated by cross validation and external test set validation to test their reliability for the calculation of â—¦brix (TSS), sucrose content and purity. The PLS regressions were performed on NIR spectra in the range of 9997.7 to 5774.1 cm−1. This spectral region was further divided into two interactive spectral regions from wavenumber 9997.7 to 7498.3 cm−1 and 6102 to 5774.1 cm−1. The prediction models developed were checked by cross validation to test its reliability. The root means square errors for cross validation (RMSECV) and correlation coefficients (R2cv) were 0.0783 and 99.78 for â—¦brix, 0.258 and 98.78 for sucrose content 1.05 and 95.3 for purity. The root means square errors of prediction (RMSEP) of test samples were 0.0851, 0.319 and 0.962, whereas the determination coefficients (R2val) for prediction were 99.7, 97.82 and 96.05 for â—¦brix, sucrose content and purity respectively. High values of determination coefficients (R2) and low values of root mean square errors indicate high predictive accuracy of optimized models. Thus the FT-NIR offers a viable alternative for the quality evaluation of sugarcane/sugarcane juice.

Download this article as