橄榄油挥发性香气成分的神经网络定量结构-保留相关性(QSRR)研究  

Quantitative Structure-Retention Relationship(QSRR)Study of Volatile Aroma Components in Olive Oil by Neural Network Method

作　　者：堵锡华[1] 宋明[1] 田林[1] 徐艳[1] 陈艳[1] 李靖[1] Du Xihua;Song Ming;Tian Lin;Xu Yan;Chen Yan;Li Jing(School of materials and chemical engineering,Xuzhou University of Technology,Xuzhou 221018)

机构地区：[1]徐州工程学院材料与化工学院,徐州221018

出　　处：《中国粮油学报》2020年第12期148-155,共8页Journal of the Chinese Cereals and Oils Association

基　　金：国家自然科学基金(21472071);江苏省自然科学基金(BK20171168),江苏省高校自然科学基金重大项目(18KJA430015)。

摘　　要：橄榄油富含不饱和脂肪酸及多种抗氧化营养物质,有良好的预防疾病和保健作用,故具有极高的保健和经济价值。为建构橄榄油挥发性香气成分保留指数神经网络的结构-保留关系模型,计算了橄榄油挥发性成分的连接性指数(m X)、分子形状指数(K m)、电性拓扑状态指数(E m)及电性距离矢量(M i),选取连接性指数的0 X、2 X、5 X,分子形状指数的K 2、K 3,电性拓扑状态指数的E 1、E 2和电性距离矢量的M 2共8种分子结构参数,作为BP神经网络方法的输入层变量,保留指数作为输出层变量,采用8-6-1的网络结构,构建模型总的相关系数R为0.9967,根据该模型计算得到的橄榄油挥发性成分保留指数预测值与实验值的平均相对误差仅为1.56%;结果表明,橄榄油挥发性成分保留指数与8种分子结构参数之间具有良好的非线性关系,—C、—C—、—C等基团片段是影响橄榄油挥发性香气成分保留指数大小的主要因素。Olive oil contains a lots of unsaturated fatty acids and a variety of antioxidant nutrients,so that it is commendable for health,which prevents people from a lot of diseases.Therefore,olive oil has very high healthcare value and economic value.In order to establish neural network model of structure-retention relationship for the chromatographic retention indices of volatile aroma components of olive oil,the molecular connectivity index(m X),the molecular shape index(K m),the electrical topological state index(E m)and the electronegativity distance vectors(M m)of aroma components of olive oil were calculated.Eight configuration parameters,including the connectivity indices 0X,2X,5X,molecular shape indices K 2,K 3,electrical topological state indices E 1,E 2 and the electronegativity-distance vectors M 2,were selected.Then,these eight parameters were used as input layer variables,while the retention index was used as the output layer variable.Besides,the neural network model was established by means of 8-6-1 network structure and BP neural network method.The total correlation coefficient R was 0.9967.The mean relative error between the predicted values and the experimental values of retention index was only 1.56%.The results showed that there was commendable non-linear relationship between the retention index and the eight molecular structure parameters.In addition,—C,—C—,—C iso-group fragments were the main factors affecting the retention index of volatile aroma components in olive oil.

关 键 词：橄榄油 挥发性香气成分 色谱保留指数 分子结构参数 神经网络 定量结构-保留相关 

分 类 号：TS221[轻工技术与工程—粮食、油脂及植物蛋白工程]