论文部分内容阅读
基于分子距边矢量(MDE),借助多元线性回归技术(MLR)建立起描述卤代烃沸点变化规律的定量结构-性质相关关系(QSPR)模型,其复相关系数R=0.9935,均方根误差RMS=10.07K。最后5次随机选取25个化合物作预测集,以余下的62个化合物作校正集建QSPR模型(R=0.9928,RMS=10.88K),并有效地预测了沸点(R=0.9925,RMS=11.78K)。结果表明,模型的预测能力良好。另外,还运用所建立的模型对30个未知沸点的卤代甲、乙烷的沸点值进行了预测,根据预测出的沸点数据指出了CFC11,CFC12和CFC113的一些可能的代用品。
Based on the molecular distance vector (MDE), a quantitative structure-property relationship (QSPR) model was established by multivariate linear regression (MLR) to describe the boiling point variation of halohydrocarbons. The complex correlation coefficient R was 0.9935, Root-error RMS = 10.07K. In the last 5 times, 25 compounds were randomly selected as the predictive set, and the remaining 62 compounds were set as the calibration set QSPR model (R = 0.9928, RMS = 10.88K), and the boiling point (R = 0.9925 , RMS = 11.78K). The results show that the model has good predictive ability. In addition, the boiling point values of 30 haloforms of halogenated methane and ethane with unknown boiling points were predicted by using the established model. Some possible substitutes of CFC11, CFC12 and CFC113 were pointed out based on the predicted boiling point data.