咪唑类离子液体捕集CO_2性能的定量结构-性质相关性(QSPR)研究  

Predictable study of CO_2 capture performance of the imidazolium ionic liquids based on the quantitative structure-property relationship(QSPR)

作　　者：薛妮[1] 蒋军成[1] 潘勇[1] 

机构地区：[1]南京工业大学安全科学与工程学院江苏省危险化学品本质安全与控制技术重点实验室,南京210009

出　　处：《安全与环境学报》2017年第2期694-700,共7页Journal of Safety and Environment

基　　金：国家自然科学基金项目(21436006)

摘　　要：CO_2是主要的温室气体,大量CO_2的存在严重影响着人类的生存环境和生态平衡,而咪唑型离子液体具有独特的气体溶解性,在CO_2的捕集分离中有很好的应用前景。基于定量结构-性质相关性(QSPR)原理,研究了咪唑类离子液体捕集CO_2的性能与其结构参数之间的内在定量关系。应用遗传算法获得与捕集量最为密切相关的一组描述符作为输入参数,随后,分别采用多元线性回归算法及支持向量机结合粒子群优化算法建立了咪唑类离子液体捕集CO_2的性能与其描述符之间的线性和非线性模型。多元线性回归算法得出训练集和测试集的复相关系数分别为0.765和0.814,支持向量机算法得出训练集和测试集的复相关系数分别为0.987和0.933。对预测模型进行了评价验证以及稳定性分析,结果表明,2种模型具有良好的稳定性能和预测能力。This paper is inclined to devote itself to a predictable study of CO2 capture performance of the imidazolium ionic liquids based on PSO-SVM（particle swarm optimization and support vector machine）.The large number of CO2 have a negative impact on the living environment of human beings and the ecological balance.For the unique gas solubility,the imidazolium ionic liquids have the prospective potential of CO2 capture.In this paper,there exist eighteen kinds of imidazolium ionic liquids with 343 groups of experimental data being used as the modeling samples.The quantitative relationship between CO2 capture performance and the molecular structures of imidazolium ionic liquids have been investigated based on the quantitative structure-property relationship（QSPR） studies.From various calculated structure parameters,a set of parameters which have significant contribution to CO2 capture performance have been chosen as the input variables by employing the genetic algorithm method.Then the multiple linear regression（MLR） method and PSO-SVM method have been employed to model the possible linear and non-linear relationships of the selected descriptors and the CO2 capture performance.The results show that R2 of the training set and the test set are 0.765 and 0.814 by MLR,0.987 and 0.933 by SVM,respectively.Subsequently,the stability and the prediction ability of the models have been verified.Besides,we have also found the residual errors scattering randomly on the both sides of the zero calibration in the plot of residuals.The above results of our investigation and analysis show that the models we have proposed enjoy a very high stability and predictability.The relative importance of each descriptor contributing to CO2 capture performance of the imidazolium ionic liquids in MLR and PSO-SVM models can be determined by the mean effect and MMDI（method for measure of descriptor importance）.The important order of the four calculated descriptors is RDF030u〉 MATS5v〉 Ms 〉Mor29 p.Among the two experimental descrip

关 键 词：环境工程学 离子液体 CO2 预测模型 定量结构-性质相关性(QSPR) 

分 类 号：X511[环境科学与工程—环境工程]