Abstract: The effects of four independent variables, namely microwave intensity, extraction time, ethanol concentration and solid-to-solvent ratio, on the extraction efficiency of anthocyanins from blueberry were investigated by one-factor-at-a-time method. Subsequently, these variables were optimized using response surface methodology (RSM) and genetic algorithm-artificial neural network (GA-ANN) based on Box-Behnken design. The results showed that the extraction efficiency increased to a maximum followed a decrease with increasing level of each variable in the experimental range. GA-ANN showed better prediction and optimization abilities than RSM with lower relative error value (1.43% versus 2.71%) and higher determination coefficient R2 (0.904 4 versus 0.877 3). The optimal process parameters were obtained by using GA-ANN as follows: microwave intensity 155 W/g, extraction time of 53 s, ethanol concentration 56% and solid-to-solvent ratio 1:30 (g/mL). The yield of blueberry anthocyanins extracted was 85.12%, under the optimized conditions, which was higher than that (83.32%) calculated by RSM. The results from this research can provide an effective method for the optimization of process parameters in food processing.

Key words: microwave extraction, blueberry, anthocyanins, response surface methodology, neural network, optimization