基于BP神经网络算法的古代名方透脓散提取工艺优化

Back-propagation Neural Network for Optimization of Extraction Technology of Ancient Famous Prescriptions Tou’nong Powder

目的:构建BP神经网络系统优化透脓散提取工艺。方法:以加水倍量,煎煮时间,提取次数为主要影响因素,以黄芪甲苷总量为评价指标,正交实验数据为样本,构建BP神经网络系统,对透脓散提取工艺主要影响因素进行仿真优化。结果:最佳提取工艺条件为:14倍量水,煎煮1.0h,煎煮3次。结论:采用BP神经网络可以不增加实验次数进行实验结果预测,误差<1%,不仅得到最佳条件,且能预测条件变化对结果的影响,较好地优化透脓散的提取工艺,亦为其他古代名方的提取工艺优化提供新的参考依据。

Objective:Building the Back-propagation(BP)neural network system optimization Tou’nong Powder extraction process.Methods:To add water,decoction time,extraction times as the main influence factors,using Astragalus armour glycosides as the evaluation index,orthogonal test data as sample,building a BP neural network system optimization extraction process.Results:The optimal extraction conditions were verified to be 14 volumes of water,1.0 h decoction,decoction for 3 times.Conclusion:BP neural network can carry out the experimental results of forecasting without any increase in the experiments,experiments and error<1%,not only can get the best conditions,also can directly observe conditions change in its influence on the result.