Indonesia Conference Directory

The spectral analysis method of Fourier transform near Infrared (FT-NIR) to determine protein, total carbohydrate, and fat in soybean was investigated, and the performance was compared with those derived from high-performance liquid chromatography (HPLC). The samples used in this study were 40 varieties of soybean obtained from Rural Development Administration (RDA), Korea. The spectral data were acquired using Thermo Scientific Antaris II FT-NIR Analyzer at wavelength of 1000-2500 nm. A Partial Least Square (PLS) algorithm with several pre-processing methods was applied to develop model for calibration and validation. The correlation coefficient (R2) and standard error prediction (SEP) in the data set were used as evaluation parameters for optimal models. The result shows that satisfactory accuracy of prediction was achieved. The FT-NIR method was suggested to be applicable for quantification and prediction of nutritional component of soybean.