基于遗传神经网络的双目标甘草提取工艺优化

Optimization of extraction process of bi-objective glycyrrhiza uralensis based on genetic neural networ

目的:确定甘草的工艺条件,使三萜皂苷含量和总黄酮含量综合最优。方法:根据熵权法和理想算法分别得到甘草提取含量评分,并利用D-S证据理论糅合两评分得到综合评价值。运用遗传神经网络,对提取工艺和综合评价值进行回归,得到它们的关系网络,最后利用遗传算法寻求关系网络中的最佳提取工艺。结果:甘草的综合最佳提取条件为:氨浓度0.64%,乙醇浓度62%,回流时间2.06h,液固比11.2∶1。在此条件下的平均综合评价值为0.044,模型预测值为0.042,两者相对误差为4.55%,且所得综合评价值远高于实验组,故优化具有实际应用价值。结论:将遗传神经网络应用于优化甘草提取的最佳工艺条件具有可行性,可为中药提取工艺的研究提供新思路。

Objective:To determine the technological conditions of Glycyrrhiza uralensis so as to optimize the contents of triterpenoid saponins and total flavonoids.Methods:According to the entropy weight method and the ideal algorithm,the content score of liquorice extract was obtained,and the comprehensive evaluation value was obtained by combining the two scores of D-S evidence theory.Using genetic neural network,the extraction process and comprehensive evaluation value are regressed,and their relationship network is obtained.Finally,genetic algorithm is used to find the best extraction process in the relationship network.Results:The optimum extraction conditions were as follows:ammonia concentration 0.64%,ethanol concentration 62%,reflux time 2.06 h,liquid-solid ratio 11.2:1.Under these conditions,the average comprehensive evaluation value is 0.044,and the model prediction value is 0.042.The relative error between them is 4.55%,and the comprehensive evaluation value obtained is much higher than that of the experimental group.Therefore,the optimization has practical application value.Conclusion:It is feasible to apply genetic neural network to optimize the optimum extraction conditions of licorice,which provides a new idea for the study of extraction technology of traditional Chinese medicine.