摘要
应用人工BP神经网络紫外分光光度法,不经分离,同时测定了色氨酸、酪氨酸和苯丙氮酸.采用均匀实验设计法,确定了最佳网络运行参数;比较了三层和四层BP神经网络的预测能力;用预测能力较好的三层BP神经网络同时测定了复方氨基酸注射液中色氨酸、酪氨酸和苯丙氨酸,相对平均误差分别为1.09%、3.71%和2.40%.
ANN-BP algorithm is applied to simultaneous determination of the tryptophan、tyrosine and phenylalanine by spectrophotomeytry without separation. The best network parameters are determined by uniform design; the ability of forecasting between three-layer network and four-layer network is compared; three-layer network that forecasted better is applied to simultaneous determination of the tryptophan 、tyrosine and phenylalanine in the compound amino acid injection without separation with satisfactory results, and relative mean errors of tryptophan、tyrosine and phenylalanine are 1.09%、3.71% and 2.40% respectively.
出处
《生物数学学报》
CSCD
北大核心
2005年第2期240-244,共5页
Journal of Biomathematics
基金
辽宁省教育厅资助项目(2004003)
