Abstract:

Fourier transform infrared (FT-IR) spectroscopy coupled with partial least squares (PLS) regression was used to discriminate fermented soy sauce and blended soy sauce with and acid hydrolyzed vegetable protein (acid-HVP). Results indicated that different samples were successfully discriminated from seven classes according to the distribution of geographical areas from principal components of PLS. By comparing the model correlation coefficient (R2) and root mean square error of cross validation (RMSECV), the spectral region of 4000－500 cm-1 and the number of principal components was determined to be 8 and an optimal discrimination model was established. Totally 21 independent samples were successfully discriminated by the established model, which supported its availability about discriminating fermented soy sauce, HVP and blended soy sauce. The maximum deviation is less than 3.5%.

Key words: mid-infrared spectroscopy, fermented soy sauce, blended soy sauce, acid hydrolyzed vegetable protein, principal component analysis, partial least squares