Optimization of Polyphenolic Compounds from Bene Hull (Pistacia atlantica) Using Subcritical Water by Artificial Neural Network

Bene hull antioxidant activity is mainly related to polyphenolic compounds. One of the new methods to extract polyphenolic compounds is subcritical water extraction. The present study focused on modelling the amount of bene hull (Pistacia atlantica) polyphenolic compounds extracted with subcritical water method, using back propagation neural network. Process modelling is done with 27 sets of experimental data and temperature (110-200°C), process time (30-60 min) and solvent mixing ratio (1:10 - 1:50) were selected as neural network input variables. Also, the amount of polyphenolic contents (mg gallic acid/kg) was considered as the output of the process. The results showed that Back propagation Neural Network (BPNN) with 12 neurons in the hidden layer by sigmoid activation function, as the best network, could predict the amount of polyphenolic compounds with a correlation coefficient of 95%. In the next stage developed model was optimized using Enumeration method. Optimization results indicated that phenolic compounds extraction process can be achieved at 166 °C for 60 min and the ratio of 36:1 for water to Bene hull (w/v). Experimental results obtained from experiments with optimal conditions indicates proper function of the proposed method.