饱和链烷烃结构与沸点、热力学性质的QSPR研究

QSPR study between the space structure and boiling point, thermodynamic properties of saturated alkanes

  • 摘要: 利用密度泛函理论(DFT)在B3LYP/6-31g(d, p)水平下计算一系列饱和链烷烃的最优几何构型,获取空间坐标计算自定义的空间距离指数Yw'Yh、HF能量,利用多元线性回归(MLR)和BP神经网络算法构建饱和链烷烃的空间距离指数Yw'Yh、HF能量分别与沸点(Tb)、热力学性质-ΔfH298θg)的QSPR模型,BP算法的预测值与文献值相关性系数大于0.99,MLR方法的预测值与文献值相关性系数大于0.9,预测结果准确性高,模型可靠,对饱和链烷烃的物化性质研究具有理论指导意义。

     

    Abstract: Density functional theory(DFT) calculations employing the Becke-Lee-Yang-Parr(B3LYP) functional combined with a popletype 6-31 g(d, p) basis set were performed on the saturated chain alkanes. Accessed to coordinate data to calculate the new defined spatial distance index Yw',Yhand Hartree-Fock(HF). The multiple linear regression(MLR) methods and back propagation neural network(BP) are used to establish the quantitative structure property correlation(QSPR) model between the structure characteristic index Yw',Yh, HF and boiling point(Tb) or thermodynamic properties -ΔfH298θg)of saturated alkanes. The correlation coefficient between predicted value and literature value of BP method is greater than 0.99, the correlation coefficient between predicted value and literature value of MLR method is greater than 0.9, the accuracy of prediction result is high, the model is reliable, and the physicochemical property of saturated alkanes is good qualitative research has theoretical significance.

     

/

返回文章
返回