摘要
将 Levenberg-Marquardt BP 算法用于五味子甲素、五味子醇甲、五味子乙素含量的同时测定。通过对网络结构和参数的优化,获得了较满意的结果,网络训练仅34即可达到要求。人工神经网络分光光度法对试样各组分的平均回收率在98.11%～105.8%,试样不经分离即可同时测定。
Artificial Neural Networks spectrophotometry is used to analyze three components. Deoxyschizandrin, Schisandrin, and γ-Schisandrin. The computer program of ANN is based on Matlab. The difficulty of the overlapping of the absorption spectra of the three components is overcome by this procedure. The experimental results showed that the average recovery of each component was in the range from 98. 11% to 105.8% and each component could be determined satisfactorily without any pre-separation. Compared with conventional methods, this method is simple, rapid and suitable for the computer-aid analysis.
出处
《精细石油化工》
CAS
CSCD
北大核心
2008年第5期68-71,共4页
Speciality Petrochemicals
基金
浙江省自然科学基金项目(编号Y404082)
浙江省高校中青年学科带头人资助项目
关键词
人工神经网络分光光度法测定
五味子甲素
五味子醇甲
五味子乙素
五味子
artificial neural networks spectrophotometry
simultaneous determination
deoxyschizandrin
schizandrin
γ-schizandrin
Schisandra chinensis and Schisandra sphenanthera
