【摘要】 建立了基于免疫遗传算法的人工神经网络结构,用于预测α-和β-环糊精与单取代或1,4-二取代苯衍生物包结物的稳定常数,其中神经网络用于建立取代苯中2个取代基的摩尔折射率R_m,疏水性参数π和Hammett常数σ与包结物稳定常数之间的QSAR模型,而免疫遗传算法则用于优化网络的权重系数。结果表明,由于免疫机制的引入使遗传算法的优化效率提高,神经网络的学习功能得到明显改善。更多还原

【Abstract】 An artificial neural network based on an immune genetic algorithm was established to predict the association constants for the inclusion complexation of α- and β-cyclodextrin with mono- or 1, 4-disubstituted benzenes. The neural network was used to construct the QSAR model of the association constants with the substituent molar refraction Rm, hydrophobic constant ir, and Hammett constant σ of the substituents in benzene derivatives, and the immune genetic algorithm was employed to optimize the weights in the network. Results showed that the efficiency of the genetic algorithm has been enhanced due to introduction of immune mechanism, and the leaming ability of the neural network has been significantly improved.更多还原