摘要
目的建立毛细管电泳迁移时间的人工神经网络的预测方法。方法运用人工神经网络 ,通过毛细管区带电泳 (CZE)的实验电压 (V)和缓冲溶液的离子强度 (I)对维生素B1等 7种药物的迁移时间 (tmig)进行预测 ,网络采用三层结构即输入层 隐含层 输出层 ( 2 4 1型 ) ,权值修正采用误差反向传播算法 ,每种药物的样本数为 5 0 ,以“留一法”预测其迁移时间 ,网络学习次数为 1 0 0 0 0。结果当电压在 6~ 2 6kV以及硼砂溶液的离子强度为 1 0~ 1 0 0mmol/L时 ,网络预测的相对误差绝对值小于 1 2 %的概率占 82 3 %。
Objective To develop a predicted method of the migration time by using the artificial neural network.Methods The artificial neural network was used to predict the solutes migration time of vitamin B 1 and other six drugs according to both the varied experimental voltage and ionic strength of background electrolyte by the capillary zone electrophoresis(CZE).The back-propagation (BP) ANN employed was a three-layer net, namely a net with the input layer, hidden layer and output layer (2-4-1).The 'Leave-One-Out' method was applied and the net was trained by the 10 000 epochs for each drug whose sample number was 50.Results The relative errors of the 82.3% predicted values are less than 12% as the voltage is changed from 6 kV to 26 kV at the same time,the ionic strength of the borate solution is increased from 10 to 100 mmol/L.Conclusion The method can accurately predict the migration time of the drugs in the CZE.
出处
《沈阳药科大学学报》
CAS
CSCD
2003年第1期23-26,55,共5页
Journal of Shenyang Pharmaceutical University
关键词
人工神经网络
预测
药物
毛细管电泳迁移
capillary zone electrophoresis
migration time
artificial neural network
drugs