混合免疫算法及在相平衡计算中的应用

Hybrid artificial immune system and its application to calculation of chemical reaction and phase equilibrium

带或不带化学反应的相平衡计算为化学、化工领域的重要课题,可将其转换为带有约束的Gibbs自由能最小化问题。常用的序贯二次规划(SQP)收敛速度快,但依赖初始值,易陷入局部极小。人工免疫算法(AIS)具有全局寻优功能,但局部搜优性能差,收敛速度很慢,甚难找到痕量解。为此,在AIS算法中引入SQP操作,汲取两者的优点,构建混合免疫算法(HAIS)。还将相平衡的物质的量变量转换为摩尔分数,并采用适当策略处理约束,以基本可行解为基础,快速生成满足约束的抗体,以提高HAIS的操作速率。多个相平衡实例应用表明HAIS性能良好,优于其他方法(SQP,AIS)。

The calculation of phase equilibrium with or without chemical reaction is an important problem in the field of chemistry and chemical engineering. One formulation of this problem is the minimization of Gibbs free energy with constraints. The often-used method, sequence quadratic programming （SQP） is easily trapped in the local point and depends on the initial point despite its quick convergence. The artificial immune system （AIS） is able to find the global point, but its local searching is poor and finding the trace solution is difficult. In this paper, a hybrid artificial immune system （HAIS） is constructed by making use of the advantages of AIS and SQP and adding the SQP operation to the AIS method. In order to quickly get the feasible solution or antibody and shorten the run time of HAIS, the variables of mole were transformed to the new variables of mole fraction, and an appropriate strategy is employed in dealing with the constraints. The antibodies are based on the basic feasible solution. Its performance is better than other methods, such as SQP and AIS , as shown by analyzing several phase equilibrium cases.