论文部分内容阅读
应用人工神经网络构造了2个直链聚合物的结构性质关系模型。一个是直链聚合物的基团均值法描述的结构参数与其12种性质间定量关系的模型(模型1A);一个是直链聚合物的连接性指数描述的结构参数与其12种性质间定量关系的模型(模型2A)。讨论了2个模型的参数设置,而2个模型绘出的聚合物的12种性质的拟合误差(拟合值与实验值间的标准偏差)分别是:V(298K)为18.9(模型1A)/40.5(模型2A)cc/mole,E_(coh)为8.019/11.122 KJ/mole,δ为0.74/2.17(J/cc)~(0.5) ,F_d为228/235 J~(0.5)cm~(1.5)/mole,T_g为27/52 K,P_s为25/37(cc/mole)(dyn/cm)~(1/4),n为0.0140/0.5191,ζ为7.45/5.36 10~(-6)cc/mole,U_R为727/593 cm~(10/3)/(sec~(1/3)mole),U_H为568/674 cm~(10/3)/(sec~(1/3)mole),H_(μsum)为649/719 gJ~(1/3)/mole~(4/3),Y_(d,1/2)为10.6/10.5 K*kg/mole。结果表明,所建立的模型可用于直链聚合物性质的预测,而人工神经网络确实是聚合物结构性质关系研究中的一个有利的数学工具。
The structural properties of two linear polymers were constructed using artificial neural networks. One is the model describing the quantitative relationship between the structural parameters and its 12 properties (model 1A) of the mean value of the linear polymer groups; the one is the quantitative relationship between the structural parameters described by the connectivity index of linear polymers and their 12 properties Model (model 2A). The parameter settings of the two models are discussed, and the fitting errors of the 12 properties of the polymer plotted by the two models (standard deviation between the fitted and experimental values) are: V (298K) 18.9 (model 1A ) / 40.5 (model 2A) cc / mole, E coh is 8.019 / 11.122 KJ / mole and δ is 0.74 / 2.17 (J / cc) (1.5) / mole, T_g is 27/52 K, P_s is 25/37 (dyn / cm) ~ (1/4), n is 0.0140 / 0.5191 and ζ is 7.45 / 5.36 10 ~ (- 6) cc / mole, U_R is 727/593 cm ~ (10/3) / (sec ~ (1/3) mole) and U_H is 568/674 cm ~ (10/3) / ) mole of 649/719 gJ ~ (1/3) / mole ~ (4/3) and Y_ (d, 1/2) of 10.6 / 10.5 K * kg / mole. The results show that the proposed model can be used to predict the properties of linear polymers, while artificial neural network is indeed a favorable mathematical tool in the study of the structural properties of polymers.