偏最小二乘法-BP神经网络光度法同时测定复方氨基酸中色氨酸、酪氨酸和苯丙氨酸  

Simultaneous Spectrophotometric Determination of Tryptophan,Tyrosine and Phenylalanine in Compound Aminoacids by Applying New Algorithm of PLS-ANN-BP

作　　者：于洪梅[1] 王志有[1] 李井会[1] 王海洋[1] 臧娜[1] 

机构地区：[1]鞍山科技大学化工学院,鞍山114051

出　　处：《理化检验（化学分册）》2008年第5期411-413,共3页Physical Testing and Chemical Analysis(Part B:Chemical Analysis)

基　　金：辽宁省教委基金资助项目(2004003)

摘　　要：用偏最小二乘法与BP神经网络联用辅助吸光光度法,不经分离同时测定复方氨基酸注射液中色氨酸、酪氨酸和苯丙氨酸。使用改进的BP算法,避免了神经网络学习过程中可能产生的麻痹现象;提出了目标向量的简单交换方法;采用均匀试验设计法,对网络参数进行了优化,避免了过拟合现象。A new algorithm of partial least square （PLS） artificial neural network （ANN） and back propagation （BP） in combination was applied to the simultaneous spectrophotometric determination of tryptophan, tyrosine and phenylalanine in amino-acid compound injections without separation. The paralysis in the ANN training was avoided by the improved back propagation algorithm. The simple transformation for target vactors was proposed. The network parameters were optimized by uniform design and over-fitting was avoided.

关 键 词：神经网络 偏最小二乘法 苯丙氨酸 酪氨酸 色氨酸 

分 类 号：O656.3[理学—分析化学]