硝基芳香族含能材料静电感度的理论预测研究

Theoretical Prediction of Electrostatic Sensitivity of Nitroaromatic Energetic Materials

本文针对39个硝基芳香族含能材料的静电感度进行了理论预测研究。在构型优化的基础上计算了1461种分子结构描述符,采用改进的遗传算法GA-PLS筛选出8种与静电感度相关的描述符,建立了基于支持向量机算法的理论预测模型。模型的拟合优度R2、内部检验Q2LOO,外部检验Q2ext均达到0.92以上,表现出很高的稳定性与泛化能力。

In this paper,the electrostatic sensitivity of 39 nitroaromatic energetic materials was predicted.On the basis of configuration optimization,1461 kinds of molecular structure descriptors were calculated.Eight descriptors related to electrostatic sensitivity were selected by improved genetic algorithm GA-PLS,and a theoretical prediction model based on support vector machine algorithm was established.The goodness of fitting R2,internal test Q2LOO and external test Q2ext of the model are all above 0.92,which show high stability and generalization ability.