Development And Application Of FT-NIR Spectroscopic Technologies In Determination Of Mutritional Contents In Bean Food Products

As a fast high efficiency and non-destructive analytical technique,NIR(Near infrared spectroscopy) has been widely used in rapid determination of foods and agricultural products.The papers introduced systemly technical principle,steps of analysis, characteristics,development course of NIR. It also introduced several NIR spectroscopy pretreatment methods, including derivative algorithm,smoothing algorithm,centeriz ation and standardization,wavelet transform, and usual NIR spectroscopy correction methods including principal component analysis,partial least square,artificial neural networks. The NIR application in beans and bean products at home and abroad was introduced.1. The measurment of bean-curd cakes, total acid, protein and moisture content with fourier transform near-infrared spectroscopy(FT-NIRS) was studied. The original spectra of 63 bean-curd cakes were obtained, partial least squares is used. The regression model of total acid, protein and moisture were constructed. The determination coefficient(R2)of the model were 98.24%,97.85%,99.17%, and its root mean square error(RMSE)were 0.0113,0.122,0.152 in cross-validation.While the R2 and RMSE were93.46%,97.49%,99.39%, and 0.0208,0.121,0.121 in test set validation , respectively. Internal and external cross-certification test have shown that near-infrared quantitative analysis has high accuracy and could meet the accuracy needs of prediction of total acid, protein and moisture content of bean-curd cakes in production.2. The regression model of total acid, protein and moisture were constructed. The determination coefficient(R2)of the model were 99.37%,99.70%,99.73%, and its root mean square error(RMSE)were 0.00871,0.110,0.0714 in cross-validation.While the R2 and RMSE were98.74%,99.38%,99.48%, and 0.00862,0.113,0.0683 in test set validation , respectively. The results have proved that near-infrared quantitative analysis could meet the accuracy needs of prediction of total acid, protein and moisture content of tofus in production.3. The regression model of total acid, protein and moisture were constructed. The determination coefficient(R2)of the model were98.92%,97.18%,99.04%, and its root mean square error(RMSE)were 0.000581,0.0678,0.0848 in cross-validation.While the R2 and RMSE were99.67%,98.84%,99.78%, and 0.00488,0.069,0.0685 in test set validation , respectively. The results have proved that near-infrared quantitative analysis could meet the accuracy needs of prediction of total acid, protein and moisture content of tofus in production.