神经网络法计算镉(Ⅱ)、羟基及碳酸根三元体系的形态分布

Calculation of Species Distribution in Cadmium (Ⅱ), Hydroxide and Carbonate System by Artificial Neural Network Methods

采用前馈线性网络ＢＰ算法，计算了Ｃｄ２＋－ＯＨ－ＣＯ２－３三元体系的累积稳定常数。用Ｈｏｐｆｉｅｌｄ反馈网络研究了体系中络合物的形态分布。溶液中溶解的ＣＯ２对Ｉｇβ１的计算结果有重要影响，对Ｉｇβ２，Ｉｇβ３，Ｉｇβ４的结果影响不大。通过对ＣＯ２－３ 影响的校正，较好地改善了神经网络法计算Ｉｇβ１的结果。

The accumulative stabilization constants and the species distribution in Cd2- OH--CO2-3 system were calculated respectively by the feed-forward linear neural network with back-propagation algorithm and by Hopfield feedback network. The presence of CO2 in the solution has serious effect on the calculation results of Igβ1, and of species distribution, while having no obvious effect on that of Igβ2, Igβ3, Igβ4. The calculation result of Igβ1 can be well improved by calibrating the effect of carbonate in aquatic system.