基于主成分分析和人工神经网络的五味子质量鉴定方法研究 |
投稿时间:2009-07-23 修订日期:2009-08-05 点此下载全文 |
引用本文: |
摘要点击次数: 4620 |
全文下载次数: 363 |
|
基金项目:国家自然科学基金项目(面上项目,重点项目,重大项目) |
|
中文摘要:提出了一种采用近红外光谱技术结合人工神经网络对中药五味子质量进行鉴别的新方法。利用近红外光谱仪获得了3种不同来源地五味子合计90个样本的光谱曲线,光谱数据采用主成分分析法进行了聚类分析,结合人工神经网络技术建立了五味子甲素、五味子乙素和五味子醇甲三种木脂素类化合物的分析模型。主成分分析表明:前5个主成分的累积贡献率为98.75%,具有很好的聚类作用。在主成分分析的基础上,取前5个主成分的18个吸收峰作为网络的输入节点,3项指标作为输出节点,建立一个18(输入节点)—10(隐含层节点)—3(输出节点)的三层人工神经网络模型。五味子甲素、五味子乙素和五味子醇甲三项指标人工神经网络模型预测值的平均相对误差分别为4.07%、2.65%和6.15%,与高效液相色谱法测定值的符合程度很高,该模型具有很好的预测能力。新模型可用于大批量五味子的质量检测和五味子生产加工过程中的质量控制。 |
中文关键词:近红外光谱,五味子,主成分分析,人工神经网络,木脂素类化合物 |
|
Studies on the Discrimination of Schisandra Chinensis Quality using Near Infrared Reflectance Spectra based on Principal Component Analysis and Artificial Neural Network |
|
|
Abstract:A new non-destructive and rapid method was developed to discriminate quality of schisandra chinensis using near infrared reflectance spectra(NIRS)based on principal component analysis(PCA)and artificial neural network model.The NIRS curve of 90 schisandra chinensis samples from 3 different sources was obtained. The principal component analysis(PCA)was used to compress the NIR spectra. The artificial neural network(ANN) model of Schizandrin A, Schizandrin B and Schisantherin A was established. The PCA results showed the cumulative reliabilities of the first 5 principal components were 98.75%. The PCA had good clustering role. Based on PCA results, the 18 absorption peaks of first 5 principal components were input nods and the 3 predictive indexes were output nods, the ANN model with 18(input nods)—10(hidden layer nods)—3(output nods) was established. The average relative error of ANN models of Schizandrin A, Schizandrin B and Schisantherin A were 4.07% and 2.65% and 6.15%. The predicted value nearly was equal to HPLC value. The ANN model had good predictability to analyze schisandra chinensis quality. This NIRS ANN model can be used on detecting a large scale schisandra chinensis quality and quality controlling of schisandra chinensis production processing. |
keywords:NIRS,schisandra chinensis,PCA,ANN,lignans compounds |
HTML> 查看/发表评论 下载PDF阅读器 |