遗传算法结合反向传播算法神经网络优化党参多糖的提取工艺

Genetic algorithm combined with back propagation neural network to optimize the extraction process of Codonopsis pilosula polysaccharide

中文关键词:  党参  遗传算法  反向传播算法神经网络  正交实验  多糖

英文关键词:Codonopsis pilosula  genetic algorithm  back propagation neural network  orthogonal test  polysaccharide

中文摘要:

      目的 通过遗传算法结合反向传播算法(back propagation, BP)神经网络, 与正交实验结果对比, 优化党参多糖的提取工艺。方法 以党参多糖的提取率为指标, 采用三因素(提取次数、提取时间、料液比)水平对BP神经网络模型参数进行优化, 建立网络模型, 并采用遗传算法对BP神经网络进行目标寻优, 得到党参多糖的最佳提取工艺。结果 得到的最优提取工艺为提取次数3次, 提取时间2 h, 料液比为1:10(m:m), 在此条件下党参多糖得率预测值为55.29 mg/g, 与实际测量值的相对误差仅为1.10%, 具有较好的网络预测性。结论 利用遗传算法结合BP神经网络对党参多糖的提取工艺进行优化具有较好的预测性, 可以为党参的进一步研究开发提供新的思路。

英文摘要:

      Objective To optimize the extraction process of polysaccharides from Codonopsis pilosula by genetic algorithm combined with back propagation(BP)-based neural network and orthogonal experiments. Methods Taking the extraction amount of polysaccharide as the index, the parameters of BP neural network model were optimized by 3 factors (extraction times, extraction time, ratio of material to liquid), and the network model was established. Then the genetic algorithm was used to optimize the BP neural network and obtain the optimal extraction process of the Codonopsis pilosula polysaccharide. Results The optimal process conditions were as follows: extraction times three times, extraction time 2 h, the ratio of material to liquid is 1:10(m:m), the predicted value of this condition was 55.29 mg/g, and the relative error of the actual measured value was only 1.10%, which had good network prediction. Conclusion Using genetic algorithm combined with BP neural network to optimize the extraction process of Codonopsis pilosula polysaccharide has good predictability, which can provide new ideas for further research and development of Codonopsis pilosula.