Application of response surface methodology coupled with artificial neural network to predict kinetic of food product under different drying conditions

Drying has been used to extend the shelf life of foods. For monitoring the drying process of zucchini, different models of neural networks such as percentron, radial basis function and hybrid model of neural network and response surface methodology were utilized. Drying time, air drying temperature and thickness were considered as input parameters. On the other hand, furrier number, activation energy, effective moisture diffusivity and shrinkage were as network output. The results showed that perceptron neural network with logsig-logsig activation function as a goodness activation function can be estimated activation energy, furrier number, effective moisture diffusivity and shrinkage with R2 value 0.999, 0.992, 0.999 and 0.991, respectively.