人工神经网络用于连钱草中总黄酮和三萜酸类成分的含量预测

以连钱草的毛细管电泳指纹图谱为输入数据,以总黄酮和三萜酸类成分含量为输出数据,构建了反向传播网络、径向基函数网络和广义回归网络三种人工神经网络模型.采用三种网络模型和两种预测方法对未知样本的总黄酮和三萜酸类成分含量进行了预测,并分别比较了三种网络和两种预测方法的预测结果.另外,结合聚类分析结果和输入数据的相似度,分析了预测误差的来源.结果表明:三种网络对大部分样本的预测值与实际值都比较接近,而广义回归网络的预测效果最优;扣除奇异值后,广义回归网络的两种预测方法对未知样本的总黄酮和三萜酸类成分含量的平均预测误差分别为10.9%和0.00073%.

Prediction for the Contents of Total Flavonoids and Triterpene Acids in Herba Glechomaes Based on Artificial Neural Network

Back-propagation network,radial basis function network and generalized regression network(GRNN) were developed by using capillary electrophoresis fingerprint data of Herba Glechomaes as input values of networks and the contents of total flavonoids and triterpene acids in Herba Glechomaes as output values of networks.Based on the three networks above and two predicting methods,the contents of total flavonoids and triterpene acids of new specimens were predicted.The prediction results of three networks and two predicting methods were compared with each other,respectively.In addition,the origin of prediction error was analyzed according to the results of cluster-analysis and similarity.It was indicated that the values predicted by three networks were all close to the real values for most specimens,and GRNN can give the most effective prediction.After removing singular value,the average prediction errors for the contents of total flavonoids and triterpene acids of new specimens are 10.9% and 0.00073%,respectively,produced by two predicting methods of GRNN.