期刊文献+

基于遗传算法对有机物热导率的预测研究 被引量:3

Prediction of the Thermal Conductivity of Organic Compounds Based on the Genetic Algorithm
下载PDF
导出
摘要 根据定量构效关系(Quantitative Structure-Property Relationship,QSPR)原理,研究热导率与其分子结构间的内在定量关系。以178种有机化合物作为样本集,随机选择142种作为训练集,36种作为测试集,采用遗传算法(Genetic Algorithm,GA)进行变量选择,得到5个特征描述符作为模型的输入变量,结合多元线性回归(Multiple Linear Regression,MLR)方法建立了遗传-多元线性回归(GA-MLR)预测模型。研究结果表明:GA-MLR模型的训练集和测试集的复相关系数分别为0.808 0和0.742 2,其均方根误差分别为0.109 8和0.129 3,预测效果令人满意。随后采用残差分析图对样本集进行了残差分析,进一步验证模型在建立过程中未产生系统误差。采用"Y-随机性检验"方法对模型进行了研究,发现预测模型不存在"偶然相关"现象,具备较强的稳定性。该研究提供了一种有效预测有机化合物热导率的方法。 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 compounds 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 compounds.
出处 《常州大学学报(自然科学版)》 CAS 2017年第1期86-92,共7页 Journal of Changzhou University:Natural Science Edition
基金 常州大学科研启动基金(ZMF15020112) 常州市科技支撑计划项目(社会发展)(CE20155025) 建筑消防工程技术公安部重点实验室开放课题(KFKT2014MS02)
关键词 热导率 遗传算法 多元线性回归 预测 定量构效关系 thermal conductivity genetic algorithm multiplelinearregression prediction quantitativestructure-property relationship
  • 相关文献

参考文献13

二级参考文献171

共引文献82

同被引文献13

引证文献3

二级引证文献5

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部