MIVM参数的算法研究

Algorithm Research on the Parameters from the MIVM

采用MIVM模型预测合金溶液中组元的活度时,遇到了瓶颈参数B12、B21求解处理问题,为此,对无限稀偏摩尔混合焓实验数据已知状况不同的两类溶液分别提出了"遍历优算法"和"多目标加权和优算法"两种算法,并给出了算法的实现方案。采用Bi-InI、n-Sn、Fe-Si合金溶液验算"遍历优算法",求算了溶液两参数B12、B21的优化解,计算出的平均相对误差、偏差分别为0.27%、182.4227J.mol-1;采用Au-Zn、Au-Sn、Mn-Ni合金溶液验算"多目标加权和优算法",求算了对应溶液两参数B12、B21的优化解,相应的平均相对误差、偏差分别为10.82%、1063.8667J.mol-1。在进一步的合金溶液组元活度预测中,组元活度的预测值与试验值吻合较好。结果表明,采用"遍历优算法"、"多目标加权和优算法"能有效地解决MIVM模型中参数B12、B21求解问题,两算法体系是有效可行的。

The solutions of the bottleneck parameters of B12 and B21 are presented when the molecular interaction volume model（MIVM） is used to predict the activities of components in liquid alloys.With regarding to this,the traverse excellent algorithm（TEA） and multi-objective weighting and optimization algorithm（MOWOA） were put forward for the liquid alloys with or without the experimental values of infinite dilution partial molar mixing enthalpy.At the same time,the implementation schemes of the algorithm were given.The TEA was checked by using Bi-In,In-Sn and Fe-Si alloys to optimally solve the binary parameters,B12 and B21,where the average relative error is 0.27% and the average standard deviation is 182.4227J·mol-1.And the MOWOA was checked by using the Au-Zn,Au-Sn and Mn-Ni alloys,where the average relative error is 10.82% and average standard deviation is 1063.8667J·mol-1.Further,the prediction values of activity of both components in the binary liquid alloys are in good agreement with the experimental values.The results show that the TEA and MOWOA may solve the solution of MIVM model parameters of B12 and B21 and they are valid and feasible.