燃料油中有机硫化物在不同色谱柱上的气相色谱保留行为——定量结构色谱保留关系(QSRR)的研究
详细信息 查看官网全文
- 英文论文题名：Investigation of GC Retention Mechanism of Organic Sulfur Compounds in Fuel Oil Based on Quantitative Structure-Retention Relationship
- 论文作者：张晓彤 ; 石丽华 ; 宋丽娟 ; 孙兆林 ; 孙铤
- 英文论文作者：Xiaotong Zhang ; Lihua Shi ; Lijuan Song ; Zhaolin Sun ; Ting Sun ; Liaoning Key Laboratory of Petrochemical catalytic Science and Technology ; Liaoning Shihua University ; College of Sciences ; Northeastern University
- 年：2016
- 作者机构：辽宁石油化工大学辽宁省石油化工催化科学与技术重点实验室;东北大学理学院;
- 论文关键词：遗传算法-多元线性回归法(GA-MLR) ; 列文伯格-马夸尔特-人工神经网络算法(L-M ANN) ; 燃料油 ; 气相色谱-硫化学发光检测法(GC-SCD)
- 会议召开时间：2016-07-01
- 会议录名称：中国化学会第30届学术年会摘要集-第二十五分会：化学信息学与化学计量学
- 语种：中文
- 分类号：TE626.2;O657.71
- 学会代码：ZGHY
- 会议名称：中国化学会第30届学术年会-第二十五分会 ; 化学信息学与化学计量学
- 会议地点：中国辽宁大连
- 主办单位：中国化学会
- 学会名称：中国化学会
- 页数：1
- 文件大小：166k
- 原文格式：D
- 会议级别：全国
摘要
硫组分的含量是表征燃料油品质的重要指标。采用遗传算法-多元线性回归法(GA-MLR)、遗传算法-偏最小二乘法(GA-PLS)、列文伯格-马夸尔特-人工神经网络算法(L-M ANN)对56种有机硫化物在四种不同极性固定相上的气相色谱保留指数分别进行了定量结构-气相色谱保留关系研究。采用遗传算法-多元线性回归法(GA-MLR)选取模型的输入参数。结果表明:上述三种方法建立的定量模型均具有较强的稳定性和良好的预测能力,但L-M ANN模型的预测结果稍好于其它两种方法。本课题首次将L-M ANN算法用于燃料油中有机硫化物定量结构-气相色谱保留关系的研究,效果十分理想。这表明L-M ANN算法可以作为一种替代性的建模方法用于物质的定量结构性质相关/保留关系(QSAR/QSPR)的研究中。
Sulfur component is an important index for characterization of fuel oil.Genetic algorithm and multiple linear regression(GA-MLR),partial least square(GA-PLS) and Levenberg-Marquardt artifical neural network(L-M ANN)techniques were used to investigate the correlation betweent retention indices(RI) and descriptors for 56 organic sulfur compounds in fuel oil.The GA-MLR method was used to select the input variables in QSRR modeling.The results indicate that three QSRR models by above methods all have strong stability and good predictive performance,but the L-M ANN models is better.This is the first research on the QSRR of organic sulfur compounds in fuel oil against the RI using the L-M ANN models.The models perform well indicates that L-M ANN method can be used as an alternative modeling tool for quantitative structure-property/retention relationship(QSPR/QSRR).
Sulfur component is an important index for characterization of fuel oil.Genetic algorithm and multiple linear regression(GA-MLR),partial least square(GA-PLS) and Levenberg-Marquardt artifical neural network(L-M ANN)techniques were used to investigate the correlation betweent retention indices(RI) and descriptors for 56 organic sulfur compounds in fuel oil.The GA-MLR method was used to select the input variables in QSRR modeling.The results indicate that three QSRR models by above methods all have strong stability and good predictive performance,but the L-M ANN models is better.This is the first research on the QSRR of organic sulfur compounds in fuel oil against the RI using the L-M ANN models.The models perform well indicates that L-M ANN method can be used as an alternative modeling tool for quantitative structure-property/retention relationship(QSPR/QSRR).
引文