香叶天竺葵精油香气成分保留指数的神经网络研究  

Research on Predicting Retention Index of Aroma Components of Essential Oil from Pelargonium graveolens by Neural Network

作　　者：堵锡华[1] 王超[1] 李靖[1] 宋明[1] 田林[1] DU Xi-hua;WANG Chao;LI Jing;SONG Ming;TIAN Lin(School of Chemistry and Chemical Engineering,Xuzhou Institute of Technology,Jiangsu Xuzhou 221018,China)

机构地区：[1]徐州工程学院化学化工学院,江苏徐州221018

出　　处：《西南农业学报》2018年第7期1379-1385,共7页Southwest China Journal of Agricultural Sciences

基　　金：国家自然科学基金(21472071);江苏省自然科学基金(BK20171168);徐州市科技创新项目(KC16SG246)

摘　　要：【目的】为研究香叶天竺葵精油香气成分气相色谱保留指数与其结构之间的定量结构-保留相关关系。【方法】在分子拓扑理论基础上,计算了61个香叶天竺葵精油香气成分的分子连接性指数和电性拓扑状态指数,将其中5种结构参数(~0X、~4X、~5Xc、E_3和E_8)作为理论描述符,引入到与香气成分的气相色谱保留指数相关的回归分析中,构建了拟合度高、预测能力强的QSRR模型,再将这5种分子结构参数作为神经网络法的输入层变量,采用5∶6∶1的神经网络结构,构建了良好的神经网络预测模型。【结果】模型的总相关系数rt达到0.9987,预测的色谱保留指数值与相关文献值的相对平均误差仅为0.67%,吻合度令人满意。【结论】香叶天竺葵精油香气成分的色谱保留指数与5种分子结构参数具有良好的非线性关系,模型能较好地解释香气成分色谱保留指数的递变规律。【Objective】The present paper was conducted to study the quantitative structure-retention relationship（ QSRR） between the gas chromatography retention index and the molecular structure of aroma components of the essential oil of Pelargonium graveolens.【Method】The molecular connectivity index and electrotopological state index of 61 aroma components of essential oil of Pelargonium graveolens were calculated based on the molecular topological theory. Thus,5 structural parameters（-0 X,-4 X,-5 Xc,E_3 and E_8） of the aroma components were consequently taken as theoretical descriptors and correlated with their gas chromatographic retention indices,so as to develop the relevant quantitative structure-retention relationship（ QSRR） model. Using the 5 structural parameters as input layer variables of the neural network,a satisfying neural network model,whose network structure was 5∶ 6∶ 1,was constructed by neural network method. 【Result】The total correlation coefficient rtwas 0. 9987. The average relative error between the data in literature and the predicted values（ RI） was 0. 67 %. 【Conclusion】The results showed that there was good nonlinear relationship between the gas chromatography retention index and the 5 structural parameters. The models can better elucidate the changing rule of gas chromatography retention indices of the aroma components.

关 键 词：香叶天竺葵 精油 香气成分 气相色谱保留指数 神经网络 分子结构参数 

分 类 号：TQ91[化学工程]