The quantitative relationships existed between flash points, boiling points and molecular structures of hydrocarbon and their derivatives were investigated based on the quantitative structure-property relationship ( QSPR) study. 384 molecular descriptors of hydrocarbon and their derivatives were calculated by CODESSA, and these descriptors were pre-selected by best multilinear regression method. Then QSPR models about flash points and boiling points were built. As a result, the five-descriptor linear models were developed to describe the relationship between the molecular structures and the flash points or the boiling points. The non-linear regression models were built based on support vector machine using the five descriptors selected by best multilinear regression method. The compounds were divided into a training set and a test set. The squared correlation coefficient, cross-validation coefficient and mean squared error of each model were calculated. The test set was used to validate the prediction performance of the resulting models. The predicted results indicated that, the prediction results were in good agreement with the experimental values. The models of flash points had robustness, strong generative ability and small prediction error. The predicted results were satisfactory. But the predicted results of boiling points remained to be improved. Compared to the models of hydrocarbons, the performance of the models which added derivatives was decreased. It can be very helpful to expand the applied scope of QSPR study.%基于定量结构-性质相关性( QSPR)原理,研究了烃类及其衍生物闪点、沸点与其分子结构间的内在定量关系.应用CODESSA软件计算384种烃类及其衍生物的分子结构描述符,建立了闪点和沸点的QSPR模型.用最佳多元线性回归(B-MLR)方法筛选得到的分子描述符建立了线性回归模型.用B-MLR方法所选择的5个描述符作为支持向量机(SVM)的输入建立了非线性模型.所有的化合物被分为训练集和测试集,对每个模型的训练集和测试集的复相关系数、交互验证系数、均方根误差等进行了计算,并用测试集对模型的预测能力进行检验,预测结果表明:预测值与实验值均符合良好,所建立的闪点模型稳健,泛化能力强,预测误差小,预测的效果令人满意,但沸点的模型预测效果有待加强.相比烃类物质的模型,加了衍生物的模型性能均有所下降.
展开▼
