论文部分内容阅读
基于遗传算法(genetic algorithm,GA)的变量筛选和支持向量机(support vector machine,SVM),提出了一种改进的定量结构-性质相关(quantitative structure detonation relationship,QSPR)建模方法——遗传-支持向量机(GA-SVM),并用其建立含能材料的定量结构-爆轰性能关系(QSDR)模型,此外还应用标准SVM方法建立了QSDR模型,并用这2种模型进行呋咱系含能化合物密度的预测,随机选取85%化合物作为训练集,用来建立模型,其余化合物作为测试集来测试模型的预测能力。预测结果的交互检验的相关系数平方分别为0.9887和0.9885,平均相对误差分别为1.16%和2.12%,表明了2种建模方法的有效性。通过对2种模型的预测能力进行比较,GA-SVM方法建立的QSDR模型能更好地预测呋咱系含能化合物的密度,更利于实际应用。
Based on the genetic algorithm (GA) -based variable screening and support vector machine (SVM), an improved quantitative structure-detonation relationship (QSPR) modeling method is proposed, Support vector machine (GA-SVM) is established. The quantitative structure-detonation performance relationship (QSDR) model of energetic materials is established. In addition, the QSDR model is established by using the standard SVM method. The compound density was estimated by randomly selecting 85% of the compounds as the training set to build the model and the remaining compounds as the test set to test the predictive power of the model. The correlation coefficients of the test results were 0.9887 and 0.9885 respectively, and the average relative errors were 1.16% and 2.12% respectively, indicating the validity of the two modeling methods. By comparing the predictive power of the two models, the QSDR model established by the GA-SVM method can better predict the density of energetic compounds of furazan system, which is more conducive to practical application.