Modeling and optimization of pectinase-assisted low-temperature extraction of cashew apple juice using artificial neural network coupled with genetic algorithm

Abstract

In this study, pectinase-assisted extraction of cashew apple juice was modeled and optimized using a multi-layer artificial neural network (ANN) coupled with genetic algorithm (GA). The effect of incubation time, incubation temperature, and enzyme concentration on different responses such as yield, turbidity, ascorbic acid content, polyphenol content, total soluble solids, and pH was also determined. The developed ANN has minimum mean squared error values of 0.83, 40.92, 29.01, and 8.95 and maximum R values of 0.9999, 0.9972, 0.9995, and 0.9996 for training, testing, validation, and all data sets, respectively, which shows good agreement between the actual and predicted values. The optimum extraction parameters obtained using the developed ANN-GA were as follows: an incubation time of 64 min, incubation temperature of 32 °C, and enzyme concentration of 0.078%. The measured value of responses at the optimized process conditions were in accordance with the predicted values obtained using the developed ANN model.

Keywords: Artificial neural network; Enzymatic extraction; Genetic algorithm; Modeling; Optimization.