基于神经网络和响应面法对比优化富硒绿豆芽蛋白提取工艺研究

Comparison and optimization of protein extraction from selenium-enriched mung bean sprouts based on an artificial neural network and a response surface method

该研究从多角度评估了响应面(response surface methodology,RSM)和神经网络(artificial neural network,ANN)模型性能以优化富硒绿豆芽蛋白的提取工艺。在此基础上,检测了富硒绿豆芽蛋白的氨基酸含量和有机硒形态。试验结果表明,与RSM相比,ANN模型具有更高的拟合相关系数、更小的误差值和更精确的预测性能。应用该模型得到的最优提取工艺为:碱液浓度0.15 mol/L,提取温度43℃,提取时间133 min,液料比50∶1(mL∶g)。以此条件提取原料3次后,其蛋白质提取率为90.32%,得率为62.23%。富硒绿豆芽蛋白的总氨基酸含量为49.89 g/100 g,必需氨基酸含量为19.92 g/100 g,其中谷氨酸的含量最高(7.85 g/100 g),符合联合国粮农组织/世界卫生组织提出的理想蛋白质条件。此外,硒形态分析结果显示,硒元素主要以硒代甲硫氨酸(SeMet)的形式存在于该蛋白质中。该研究可为人工智能技术用于功能性富硒食品的开发提供一定的理论参考。

To explore the extraction process of selenium-rich mung bean sprout protein,this study evaluated the performance of the constructed response surface methodology(RSM)and artificial neural network(ANN)models from multiple perspectives.The amino acid content and organic selenium form of selenium-rich mung bean sprout protein were determined.The experimental results showed that compared with RSM,the ANN model had a higher fitting correlation coefficient,smaller error value,and more accurate prediction performance.The optimal extraction process was as follows:lye concentration of 0.15 mol/L,extraction temperature of 43℃,extraction time of 133 min,and liquid-solid ratio of 50∶1(mL∶g).Under this condition,the protein extraction rate was 90.32%,and the yield was 62.23%after three times of extraction.The total amino acid content and essential amino acid content of Se-rich mung bean sprouts protein were 49.89 g/100 g and 19.92 g/100 g,among which glutamic acid content was the highest(7.85 g/100 g),which met the ideal protein condition proposed by FAO/WHO.Besides,it was found that selenomethionine(SeMet)was the main binding mode between amino acids and selenium in selenium-rich mung bean sprouts.This study hoped to provide a theoretical reference for developing and applying functional Se-enriched foods.