论文部分内容阅读
采用自适应模糊神经网络的方法,以金属离子的价电子结构、电负性、电荷半径比及失屏参数为参变量,关联金属- HEDTA配合物稳定常数。利用减法聚类算法以确定模糊神经网络的结构,并结合模糊推理系统调整其参数。30种已知的金属-HEDTA配合物稳定常数logK值预测结果令人满意,比函数连接网络要好些。在此基础上,预测了迄今尚缺的22种金属- HEDTA配合物的稳定常数值。
Using the adaptive fuzzy neural network method, the stability constants of the metal-HEDTA complexes are correlated with the valence electron structure, the electronegativity, the charge-radius ratio and the loss-of-screen parameters of the metal ions. Subtractive clustering algorithm is used to determine the structure of fuzzy neural network, and its parameters are adjusted with fuzzy inference system. The estimated log K values of the 30 known metal-HEDTA complexes are satisfactory, which is better than the function of the network. Based on this, the stability constants of twenty-two metal-HEDTA complexes, which are still missing, were predicted.