Abstract: To enhance the yield of pumpkin polysaccharides, this study employed an ultrasound-assisted extraction method and conducted single-factor experiments focusing on three variables: liquid-to-material ratio, ultrasonic power, and ultrasonic time. Based on the Box-Behnken design, both response surface methodology (RSM) and a genetic algorithm–backpropagation neural network (GA–BP neural network) model were applied to optimize the extraction process. The results showed that under the optimal conditions predicted by the GA–BP neural network and RSM, the relative errors between the actual and predicted yields were 0.75% and 0.96%, respectively, indicating that the GA–BP neural network exhibited superior predictive performance. The optimized extraction parameters were a liquid-to-material ratio of 31:1, ultrasonic power of 252 W, and extraction time of 10 minutes, under which the polysaccharide yield reached 25.17%. This study provides a valuable reference for the development and utilization of pumpkin resources.
李新胜, 杨璐, 荣爽, 王慧竹. 遗传算法—BP神经网络优化南瓜多糖的提取工艺[J]. 吉林化工学院学报, 2025, 42(7): 4-10.
LI Xinsheng, YANG Lu, RONG Shuang, WANG Huizhu. Optimization of Pumpkin Polysaccharide Extraction Process Using Genetic Algorithm-Backpropagation Neural Network. Journal of Jilin Institute of Chemical Technology, 2025, 42(7): 4-10.
