|本期目录/Table of Contents|

[1]陈锡进,潘勇.应用电性拓扑状态指数预测烷烃自燃点*[J].中国安全生产科学技术,2009,5(6):16-20.
 CHEN Xi jin,PAN Yong.Prediction of autoignition temperatures of alkanes by electrotopological state indices[J].JOURNAL OF SAFETY SCIENCE AND TECHNOLOGY,2009,5(6):16-20.
点击复制

应用电性拓扑状态指数预测烷烃自燃点*()
分享到:

《中国安全生产科学技术》[ISSN:1673-193X/CN:11-5335/TB]

卷:
5
期数:
2009年6
页码:
16-20
栏目:
出版日期:
2009-12-31

文章信息/Info

Title:
Prediction of autoignition temperatures of alkanes by electrotopological state indices
文章编号:
1673-193X(2009)-06-0016-05
作者:
陈锡进潘勇
南京工业大学城市建设与安全工程学院
Author(s):
CHEN Xijin PAN Yong
College of Urban Construction & Safety Engineering, Nanjing University of Technology
关键词:
定量结构-性质相关(QSPR)电性拓扑状态指数人工神经网络自燃点烷烃
Keywords:
quantitative structureproperty relationship (QSPR) electrotopological state indices artificial neural network autoignition temperature alkane
分类号:
O621.21;X937
DOI:
-
文献标志码:
A
摘要:
建立了一个基于人工神经网络的定量结构-性质相关性模型,用于52种烷烃化合物自燃点的预测研究。应用原子类型电性拓扑状态指数作为表征分子结构特征的描述符。该指数既能表征分子的电子特性,又反映其拓扑特征,同时易于计算,并有较强的同分异构体区分能力。采用误差反向传播(BP)神经网络方法对烷烃自燃点与电性拓扑状态指数间可能存在的非线性关系进行拟合。将52种烷烃样本随机划分为训练集(30种)、验证集(8种)和测试集(14种),并通过“试差法”确定网络的最优参数。运用最佳网络结构[641]对实验样本进行模拟,结果表明,多数样本的自燃点预测值与实验值符合良好,对于测试集,平均预测绝对误差为84℃,均方根误差为118,优于多元线性回归方法和传统基团贡献法所得结果。该方法的提出为工程上提供了一种根据分子结构预测有机物自燃点的有效方法。
Abstract:
A quantitative structureproperty relationship (QSPR) model was constructed to predict the autoignition temperature (AIT) of 52 alkanes by means of artificial neural network (ANN). Atomtype electrotopological state indices were used as molecular structure descriptors which combined together both electronic and topological characteristics of the analyzed molecules. Moreover, these indices were easily to calculate and also had good discrimination ability for isomers. The backpropagation (BP) neural network was employed for fitting the possible nonlinear relationship existed between the structure and property. The dataset of 52 alkanes was randomly divided into a training set (30), a validation set (8) and a testing set (14). The optimal condition of the neural network was obtained by adjusting various parameters by trialanderror. Simulated with the final optimum BP neural network[641], the results showed that most of the predicted AIT values are in good agreement with the experimental data, with the average absolute error being 84℃, and the root mean square error (RMS) being 118 for the testing set, which were superior to those obtained by multiple linear regression analysis and traditional group contribution method. The method proposed can be used to predict the AIT values of organic compounds based on molecular structures for engineering.

参考文献/References:

[1]Suzuki, T. Quantitative structureproperty relationships for autoignition temperatures of organic compounds. Fire and Materials.1994,18(2):81~88.[2]Tetteh, J.; Metcalfe, E., Howells, S. Optimization of radial basis and backpropagation neural networks for modeling autoignition temperature by quantitative structureproperty relationships. Chemometrics and Intelligent Laboratory Systems.1996,32:177~191.[3]Mitchell, B. E.; Jurs, P. C. Prediction of autoignition temperatures of organic compounds from molecular structure. Journal of Chemical Information and Computer Sciences.1997,37:538~547.[4]Albahri, T. A. Flammability characteristics of pure hydrocarbons. Chemical Engineering Science.2003,58:3629~3641.[5]Albahri, T. A.; George, R. S. Artificial neural network investigation of the structural group contribution method for predicting pure components auto ignition temperature. Industrial & Engineering Chemistry Research.2003,42(22):5708~5714.[6]Hall, L. H.; Kier, L. B. Electrotopological state indices for atom types:A novel combination of electronic, topological, and valence state information. Journal of Chemical Information and Computer Sciences.1995,35(6):1039~1045.[7]Hall, L. H.; Story, C. T. Boiling point and critical temperature of a heterogeneous data set:QSAR with atom type electrotopological state indices using artificial neural networks. Journal of Chemical Information and Computer Sciences.1996,36(5):1004~1014.[8]Huuskonen, J.; Livingstone, D. J.; Tetko, I. V. Neural network modeling for estimation of partition coefficient based on atomtype electrotopological state indices. Journal of Chemical Information and Computer Sciences.2000,40(4):947~955.[9]Huuskonen, J. QSAR modeling with the electrotopological state indices:predicting the toxicity of organic chemicals. Chemosphere.2003,50(7):949~953.[10]http://www.inchem.org/pages/icsc.html.[11]http://ptcl.chem.ox.ac.uk/MSDS/.[12]http://ull.chemistry.uakron.edu/erd/index.html.[13]http://www.msdsxchange.com/english/index.cfm.[14]Dean, J. A. Langes Handbook of Chemistry, 15th ed. New York:McGrawHill,1999.[15]Pan, Y.; Jiang, J. C.; Wang, Z. R. Quantitative structureproperty relationship studies for predicting flash points of alkanes using group bond contribution method with backpropagation neural network. Journal of Hazardous Materials.2007,147:424~430.

相似文献/References:

备注/Memo

备注/Memo:
收稿日期:2009-09-24作者简介:陈锡进,男,博士研究生。*基金项目:国家自然科学基金资助项目(编号:20976081);
更新日期/Last Update: