基于违约解转化法的遗传算法及其应用

Genetic Algorithm Based on Converting Infeasible Individuals into Feasible Ones and its Application

遗传算法求解约束优化问题时,面临的关键问题之一就是如何处理约束条件。目前最常用的方法是惩罚函数法,但也有一些其它方法。该文提出了一种较为通用的处理约束条件的方法—违约解转化法,也就是把所有违反约束的解逐渐拉向可行域之内,使之转化成为一个可行解。把此方法应用到遗传算法中,就形成了基于违约解转化法的遗传算法。测试实例计算表明,基于违约解转化法的遗传算法在求解约束优化问题时是可行的。

As a Genetic Algorithm handles the constraint optimization problem,the dif ficulty is how to solve constraints.Although there are many other methodologies to handle constraints,the most common one is the penalty function.A new metho d is proposed in this paper which transforms an infeasible individual into feas ible one,and can be used in Genetic Algorithms.So the Genetic Algorithms Base d on Converting Infeasible individuals into Feasible ones(CIFGA)come into bei ng which can handle constraints in general purpose.The experimental results sho w that CIFGA is valid for constraint optimization.