常州大学学报（自然科学版）2017，Vol.29Issue(1)：86-92,7.DOI:10.3969/j.issn.2095-0411.2017.01.015
基于遗传算法对有机物热导率的预测研究
Prediction of the Thermal Conductivity of Organic Compounds Based on the Genetic Algorithm
摘要
Abstract
The quantitative relationship between the thermal conductivity and the molecular structure of the organic compound was investigated based on the QSPR principle.The datasets of 178 kinds of organic com-pounds were randomly divided into the train set (142)and the test one (36).GeneticAlgorithm (GA)was well adapted to the variable selection.As a result,five descriptors were screened out from a large pool of calculated descriptors as input parameters for the model.Coupled with these descriptors,Multiple Linear Regression (MLR)method was employed to build the GA-MLR model.In the model,the square correla-tion coefficient (R2 )of the train set and the test one were 0.808 0 and 0.742 2,and the Root Mean Square Error(RMSE)were 0.109 8 and 0.129 3,respectively.The above results showed that the built modelis robust are very satisfying.Then,the sample was analyzed with the residual analysis as to further validate no systematic error in the process of the model.In addition,Y-randomization was applied to determine that there is no chance correlation in the model.It was seen that the established model had strong stability.This research provides a new and effective method for predicting the thermal conductivity of organic com-pounds.关键词
热导率/遗传算法/多元线性回归/预测/定量构效关系Key words
thermal conductivity/genetic algorithm/multiplelinearregression/prediction/quantitativestruc-ture-property relationship分类
资源环境引用本文复制引用
时静洁,袁雄军,邵辉,王凯全,陈海群..基于遗传算法对有机物热导率的预测研究[J].常州大学学报（自然科学版）,2017,29(1):86-92,7.基金项目
常州大学科研启动基金(ZMF15020112) （ZMF15020112）
常州市科技支撑计划项目(社会发展)(CE20155025) （社会发展）
建筑消防工程技术公安部重点实验室开放课题(KFKT2014MS02)。 （KFKT2014MS02）
