Optimisation of pumpkin mass transfer kinetic during osmotic dehydration using artificial neural network and response surface methodology modelling

In this study, the response surface methodology (RSM) was used to optimise osmo-dehydration of pumpkin cubes. Effect of different parameters including osmotic solution temperature in the range of 5 to 50 °C, the immersion time from 0 to 180 min and the concentration of osmotic solution (from 5% salt + 50% sucrose w/v to 15% salt + 50% sucrose w/v) on water loss (WL), solid gain (SG), weight reduction and final moisture content were investigated by central composite design. The optimum condition for osmotic dehydration was found to be at a temperature of 5 °C, an immersion time of 180 min and an osmotic solution concentration of 15% salt + 50% sucrose w/v. At this optimum condition WL, SG, weight reduction and moisture content were found to be 70.7 g/100 g initial sample, 10.2 g/100 g initial sample, 59.06 g/100 g initial sample and 0.64 g water/g dry matter, respectively. The comparison of the obtained results by artificial neural network and RSM modelling showed that the artificial neural approach has a higher ability in comparison with RSM modelling in predicting final moisture content (R2=0.998 and 0.992, respectively).