论文部分内容阅读
在求解分子相互作用体积模型(MIVM)中的分子对位能参数Bij,Bji时,为避免出现计算超量问题,基于遗传算法,提出了一种新算法—多个体参与交叉遗传算法,并给出了新算法的实现方案。在新算法中,采用了轮盘选种法与优秀个体保存、多个体参与交叉和多点变异等策略;选择Bij,Bji个体时,按照适应度函数值愈大,个体被选中的概率愈大的原则,选择二元合金溶液中一组元活度的计算值与实验测定值之间偏差(σ)建立的适应度函数fx最大为优化目标。采用Matlab软件编制新算法下的Bij,Bji计算程序,并分别求算了Cd-Bi,Cd-Pb,Cd-Sn,Bi-Pb,Bi-Sn,Pb-Sn等6组二元合金溶液在773 K下合金溶液的参数Bij,Bji优化值。由Bij,Bji计算优化值和其他一些资料数据,使用MIVM进一步预测773 K下Cd-Bi-Pb,Cd-Bi-Sn,Cd-Pb-Sn,Cd-Bi-Pb-Sn多元合金溶液中组元Cd的活度,模型预测值与实验测定值之间的相对误差低于4.6%,偏差低于0.022,两者吻合很好。结果表明,所提出的多个体参与交叉遗传算法为求解MIVM中的参数Bij,Bji提供了一种有力工具,从而也能提高MIVM对多元合金溶液中组元活度的预测效果。
In order to avoid the problem of overcompensation, a new algorithm based on genetic algorithm was proposed when solving the molecular alignment parameter Bij, Bji in MIVM (Multiple Interactions Model) A new algorithm to achieve the program. In the new algorithm, strategies such as roulette selection and excellent individual preservation, multiple individuals involved in crossover and multi-point mutation are adopted. When individuals of Bij and Bji are selected, according to the fitness function value is larger, the probability of individuals being selected is larger , The maximum fitness function fx established by the deviation (σ) between the calculated value of a group of elemental activity in the binary alloy solution and the experimentally determined value is the optimal target. The Bij and Bji calculation programs under the new algorithm were compiled by using Matlab software and six groups of binary alloy solutions of Cd-Bi, Cd-Pb, Cd-Sn, Bi-Pb, Bi-Sn and Pb- The parameter Bij, Bji of alloy solution under 773 K is optimized. Bij, Bji calculated the optimized value and some other data to further predict the group of Cd-Bi-Pb, Cd-Bi-Sn, Cd-Pb-Sn and Cd-Bi-Pb-Sn multi-alloy solution at 773 K using MIVM The relative error between elemental Cd activity, model predictive value and experimentally measured value was less than 4.6% and the deviation was less than 0.022, which was in good agreement with each other. The results show that the proposed multiple-body crossover genetic algorithm provides a powerful tool for solving the Bij and Bji parameters in MIVM, and also improves MIVM’s predicting the activity of elements in a multi-alloy solution.