利用遗传神经网络对正规溶液模型组元交互作用能进行了改进, 使之更接近实际溶液活度模型, 并将其应用于 MnO--SiO2及CaO--Al2O3二元渣系的组元活度计算. 由于实际溶液的性质不同于正规溶液, 在模型对交互作用能 Ω Mn-Si和ΩSi-Mn的计算时发现, Ωij不仅是温度和组成的函数, 而且在相同的温度和组 成时, Ωij≠Ωji. 通过将该模型的计算结果与大量文献数据进行对比研究, 发现该模型具有很强的非线性拟合能力, 能准确地预报实际溶液组元活度值.
A method for improving the regular solution model by GA-NN is introduced in this article and the activity of component in MnO-SiO2 and CaO-Al2O3 binary system is estimated by this model. Due to the different properties between real solution and regular solution, it is proved that the interaction energy Ωij is the function of temperature and composition, and Ωij ≠Ωji at the same temperature and composition. With the comparison of results received by calculation and previous studies, a high nonlinear capability is found of the model, so it can be used to accurately predict the activity of solution component.
参考文献
| [1] | |
| [2] | |
| [3] | |
| [4] | |
| [5] | |
| [6] | |
| [7] | |
| [8] |
- 下载量()
- 访问量()
- 您的评分：
-
10%
-
20%
-
30%
-
40%
-
50%