摘要
为了能够快速、准确地获取含低碳酯二元共沸物的共沸温度,以102种含低碳酯二元共沸物为研究样本,从分子角度出发,基于定量-构效关系原理对共沸温度与其分子结构信息之间的内在定量关系开展了理论研究,建立了多个共沸温度预测模型。通过对模型的拟合能力、显著性及标准误差进行比较和分析,得到最佳的共沸温度预测模型是由6个分子描述符所构建的模型;再利用留一交叉验证法、测试集及Williams图对该模型进行内部验证、外部验证及应用域分析,并将本文所建的模型与文献报道的同类模型及UNIFAC模型进行比较。研究结果显示:共沸温度预测模型具有稳定性好、预测精度高及泛化推广能力强等优点。
In order to quickly and accurately obtain the azeotropic temperature of binary azeotropes containing low carbon esters,the inherent quantitative relationship between the azeotropic temperature and the structures of binary azeotropes was studied based on the quantitative structure-property relationship(QSPR)principles.Several azeotropic temperatures predicted models were established to estimate the azeotropic temperature when 102 binary azeotropes containing low carbon esters were taken as samples.The best azeotropic temperature prediction model,which was established by six molecular descriptors,was selected by comparison of fitting ability,significance and standard deviation.Moreover,the internal validation was executed by leave-one-out cross validation to check the stability and predictability of the model,the external validation was implemented by predicting the azeotropic temperature of binary azeotropes in test set,the applicability domain was also discussed by Williams plot,and the statistical parameters of this model was compared with the similar models reported in literature and UNIFAC model.The results show that the azeotropic temperature prediction model has the advantages of good stability,high accuracy and strong generalization ability.
作者
吕利平
李兵
何树华
李航
徐建华
LYU Li-ping;LI Bing;HE Shu-hua;LI Hang;XU Jian-hua(School of Chemistry and Chemical Engineering,Yangtze Normal University,Chongqing 408100,China;Sinochem Fuling Chongqing Chemical Industry Co.,Ltd.,Chongqing 408100,China;School of Chemistry and Chemical Engineering,Southwest Petroleum University,Chengdu 610500,Sichuan Province,China)
出处
《化学工程》
CAS
CSCD
北大核心
2019年第11期44-49,共6页
Chemical Engineering(China)
基金
重庆市“科技创新领军人才支持计划”资助项目(CSTCCXLJRC201703)
重庆市社会事业与民生保障科技创新专项项目(cstc2017shmsA90016)