基于人工鱼群算法的溶液热力学模型及其在真空蒸馏中的应用

Calculation of Key Parameters of Thermodynamic Models of Molten Binary Alloys for Vacuum Distillation in Artificial Fish Swarm Algorithm

鉴于传统的实验数据优化方法求解溶液热力学模型参数较为困难且效果较差,本文应用人工鱼群算法(AFSA)估算溶液热力学模型参数。采用Java软件编制AFSA下的正规溶液模型、分子相互作用体积模型(M]VM)的模型参数Ω_(ij)、B_(ij)和B_(ij)值的计算程序,以平均偏差S最小作为优化目标,求解了Cd-Zn、Ag-Pb、Bi-Tl、Pb-Sb和In-Sn等分属于强正、弱正、强负、弱负和混合偏差五种类型的真空蒸馏常见的二元合金体系的模型参数,并比较算法优化前后的活度系数预测值和文献值之间的差异。结果表明:AFSA优化后模型的拟合效果均有不同程度的提高,即该算法可合理有效地求解正规溶液模型的参数Ω_(ij)和MIVM的参数B_(ij)和B_(ij)值,提高其在二元合金体系中的适用程度。

Thekey parameters of the thermodynamic models of the molten alloyswere calculated in artificial fish swarm algorithm （AFSA） to simplify and improve the data-processing accuracy of the experimental results in conventional methods. The molten alloys commonly involved in vacuum distillation, including the Cd-Zn （strong positive deviation）, Ag-Pb （weak positive deviation）,Bi-TI （strong negative deviation）,Pb-Sb （weak negative deviation） and In-Sn （co-ex- isted deviation）, were selected for the analysis in AFSA and the major parameters, such as Ωij of the molecular interaction volume model （MIVM）, Bij and Bji of the regular solution model, were numerically calculated with self-developed Java software by minimizing the mean deviation S between the analytically predicted and literature reported activity coeffi- cients. The preliminary results show that AFSA considerably improves the prediction precision of Ωij, Bij and Bji, widen- ing their applications in binary alloy systems.