多氯联苯色谱保留指数QSPR建模的研究

QSPR study of the chromatographic retention time(RRT) of polychlorinated biphenyls(PCBs)

以E-Dragon软件计算的拓扑指数和连接性指数作为变量,随机将209种多氯联苯化合物(PCBs)样本数据划分为训练集、验证集和预测集,采用微粒群-v-支持向量机(PSO-v-SVM)对其色谱保留指数建立QSPR模型,选定的最佳模型入选变量仅5个,对训练集、验证集和预测集计算结果的R^2分别为0.999、0.998和0.999,预测的准确性很高。本文选定的模型较文献[16-19]的计算结果好,预测结果更可靠。

Chose the topology index and connectivity index calculated by E-Dragon software as variables, and divided 209 PCBs compounds （PCBs） into training set, validation set and prediction set randomly. The PSO-v-SVM method was adopted to establish a QSPR model of the retention of PCBs. There were only five variables in the best model, and the （ R2 ） of the computation results for training, validation and prediction set is 0. 999, 0. 998 and 0. 999 respectively, which shows very high accuracy for the forecast set. In this paper, compared with the results in References [ 16- 19], the results show that model of the authors not only has better statistical resuits, but also gets more reliable forecast results.