文献类型：期刊文献

英文题名：Computational and experimental approaches for investigating membranes diffusion behavior in model diesel fuel

作者：Yang, Zhen[1,2];Gu, Xingsheng[3];Ling, Changjian[1,2];Liang, Xiaoyi[1,2]

机构：[1]East China Univ Sci & Technol, State Key Lab Chem Engn, Shanghai 200237, Peoples R China;[2]Minist Educ, Key Lab Special Funct Polymer Mat & Their Related, Shanghai 200237, Peoples R China;[3]East China Univ Sci & Technol, Informat Sci Inst, Shanghai 200237, Peoples R China

年份：2018

卷号：56

期号：9

起止页码：2724

外文期刊名：JOURNAL OF MATHEMATICAL CHEMISTRY

收录：;WOS:【SCI-EXPANDED(收录号:WOS:000443702300009)】；

基金：This work is partly supported by National Science Foundation of China (21177038), China Scholarship Council Fund (201406745031) and Material Informatics for Engineering Design Research Group of Woodruff School of Mechanical Engineering, Georgia Institute of Technology, Atlanta, GA, USA. The authors would like to thank the anonymous reviewers for their constructive suggestions that have improved the quality of this work.

语种：英文

外文关键词：Diesel components; Desulfurization selectivity; Prediction; Support vector regression

摘要：Genetic algorithms trained support vector regression predicting model is conducted to research diffusion behavior of methylnaphthalene and dibenzothiophene in four different membranes of polymethyl methacrylate, polymethyl acrylate, polyvinyl chloride and polyvinyl alcohol in model diesel fuel. It is found that the polyvinyl chloride is optimal membrane material for improving the diffusion selectivity of methylnaphthalene and dibenzothiophene, which demonstrates that the polyvinyl chloride membrane is favorable to the diesel fuel desulfurization. Also, molecular dynamic simulation is applied to validating the performance of genetic algorithm trained support vector regression model. The results of genetic algorithm trained support vector regression model reveal that the simulation values are well agreed with the experimental data and molecular dynamic simulation results. Meanwhile, the performance of the genetic algorithms trained support vector regression predict model is better than that of the genetic algorithms trained neural network model, which indicates that genetic algorithms trained support vector regression method offers a new prospected decision-theoretic approach to the diesel desulfurization.