Predicting solubility index of roller dried goat whole milk powder using Bayesian regularization ANN models

A predictive model for predicting solubility index of roller dried goat whole milk powder using artificial neural network is proposed. The model takes into account solubility index of the product as a function of roller dried goat milk. Feedforward networks with one hidden layer were used with Bayesian regularization algorithm. The best fitting with the training data set was obtained with 451 topology, which made possible to predict solubility index of roller dried goat whole milk powder with accuracy, at least as good as the experimental error, over the whole experimental range. On the validation data set, simulations and experimental kinetics test were in good agreement. The developed model can be used for predicting solubility index of roller dried goat whole milk powder.

Keywords:

Solubility index , Artificial neural network , Bayesian regularization algorithm , Goat , Milk powder