Prediction of Auto-ignition Temperatures of Organic Compounds by SVM Based on the Genetic Algorithm

SHI Jing-jie,CHEN Li-ping,SHI Ning,XU Wei,YANG Hui,CHEN Wang-hua

DOI: https://doi.org/10.3969/j.issn.1003-3033.2011.07.021

2011-01-01

Abstract:The quantitative relationship between AIT and molecular structures of organic compounds was investigated based on QSPR principle.The dataset of 265 kinds of organic compounds was randomly divided into a training set(238),and a testing set(27).GA was adopted in variable selection,and MLR model and SVM model were employed in research for figuring out the relationship between AIT and molecular structure of organic compounds.Through in-depth analysis,authors find out that it is the problem of experimental data that resulted in unreasonable prediction of AIT values.By comparing the two models,GA-SVM model is obviously superior to GA-MLR model,which indicates that there is a non-linear relationship between AIT and molecular structure of organic compounds.