RSM and ANN approach for optimization of ultrasonic assisted extraction of pumpkin seed oil and their quality assessment

Abstract

The current study focuses on optimization of process parameters for ultrasonic assisted extraction (UAE) of pumpkin seed oil (PSO) using response surface methodology (RSM) and artificial neural network (ANN). Three parameters of UAE process, amplitude, time and solvent to seed (S:S) ratio were selected in the range of 20–40 %, 15–45 min and 2–6 mL/g respectively and optimized on the basis of responses viz. yield, total phenolic compounds (TPC), squalene content and induction time of oil. The optimum process conditions of UAE were obtained as amplitude- 34.76 %, time- 34.37 min and the S:S ratio 6 mL/g. Under these optimum conditions, the oil showed good yield 39.05 %, TPC 45.02 mg GAE/g, squalene content 447.4 mg/100 g and induction time 5.27 h. The ANN model was developed using multilayer perceptron (MLP) and trained with back propagation algorithm utilizing training data set. The results indicate that the predicted values of responses by ANN model were in good agreement with the experimental values than the RSM model. The UAE-PSO showed better retention of bioactive compounds along with far better oil stability as compared to SE-PSO. The Scanning electron microscopy (SEM) analysis showed the random porous pores and rupture cell wall matrix in UAE treated seed powder.