摘要
在测定、收集和计算出一组氨基酸的拓扑指数和各种理化参数之后，再通过相关分析选择其中最有代表性的几个参数作为反向传播人工神经网络的输入参数，用于正相薄层色谱中氨基酸保留规律的研究。结果表明，氨基酸的色谱保留值与其结构之间呈现较强的非线性关系，采用人工神经网络方法比用多元线性回归方法能够更精确地描述这种关系。
The relationship between the thin layer chromatographic retention values and the molecular structures of fifteen amino acids was studied by using back propagation artificial neural networks (ANNs). In this paper, firstly, lots of parameters of amino acids have been determined, accumulated and computed. Then, correlation coefficients of all parameters were computed by taking advantage of correlation analysis. Taking the correlation coefficient approaching one as the criteria in the correlation analysis, all parameters were classified into three kinds. Three parameters were selected from each kind, respectively, to consist of one group. Then, the optimized groups of parametes, which have clearer physicochemical meanings, were used as the inputting parameters of artificial neural networks. Correlation coefficients of experimental retardation values and those calculated by using ANNs were computed and showed good agreement. The present work shows that the ANNs method may take an important role in the study of the relationship between TLC behavior and compound structure.
出处
《色谱》
CAS
CSCD
北大核心
1999年第1期14-17,共4页
Chinese Journal of Chromatography
关键词
薄层色谱法
人工神经网络
氨基酸
保留值
thin layer chromatography, artificial neural network, amino acid, retention value
