Microwave vacuum drying of dragon fruit slice: Artificial neural network modelling, genetic algorithm optimization, and kinetics study

Abstract Microwave vacuum drying of dragon fruit (Hylocereus undatus) slice was investigated in this study and the drying process was modelled with the help of feed-forward back-propagation artificial neural network (ANN). ANN model was architecturally composed of an input layer with three nodes, an output layer with four nodes, and a hidden layer with eleven nodes. The slices were pre-treated with citric acid (0.5–1.5%) and afterward dried in the microwave vacuum dryer at different combinations of microwave power (200–600 W), vacuum level (3–9 kPa). Based on the factorial design, sixty experiments were conducted at different combinations of independent variables. The effect of independent variables on four different responses such as total phenolic content, drying efficiency, color change, and rehydration ratio, was analyzed for optimization by application of genetic algorithm (GA). The optimized conditions of microwave power, vacuum, and citric acid concentration obtained by integrated ANN and GA were found to be 450 W, 9 kPa, and 1.35 % , respectively. The predicted response at optimum condition for total phenolic content, drying efficiency, color change, and rehydration ratio of dried dragon fruit slice was found to be 8.324 mg GAE/g, 59.099%, 14.577, and 3.463 respectively. The predicted values of the ANN-GA model were in strong agreement with the experimental data with a low relative deviation value of 1.557–2.936%. The effective moisture diffusivity values for microwave vacuum drying of dragon fruit were found to be 3.418 × 10 - 9 - 10.771 × 10 - 9 m2/s and activation energy was estimated to be 6.766 W/g for dragon fruit slice.