Study on Bitterness Analysis and Evaluation of the Isolated Soybean Protein Using Electronic Tongue

The bitterness of the isolated soybean protein was analyzed using Astree electronic tongue. Principal component analysis (PCA) and discriminant factor analysis (DFA) were used for qualitative analysis of the acquired taste information. Based on partial least squares and RBF neural network, the quantitative prediction model of bitterness was established. The results showed that both principal component analysis and discriminant factor analysis could be used to determine the bitterness degree of the formula. RMSE of RBF neural network prediction model were 0.010 and 0.007, respectively; and RMSE of partial least square prediction model were 0.035 and 0.093, respectively. The results showed that the prediction model established by RBF neural network was very effective. The results also provided a new method for the bitterness evaluation system of the isolated soybean protein.