基于违约解转化法的遗传算法及其性能分析

遗传算法在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.本文提出了一种基于违约解转化法的遗传算法(CIFGA),也就是遗传算法在处理约束条件时,在每一进化代遗传操作后,把所有违反约束条件的个体逐个转化成满足约束条件的个体,整个遗传群体保持不变,经过一代代的进化,最终求出约束问题的最优解.对于采用二进制编码和实数编码的CIFGA,理论证明了其收敛性.测试试验结果表明:CIFGA有较好的算法性能和解决约束优化问题的能力.

A New Improved Genetic Algorithms Based on Converting Infeasible Individuals into Feasible Ones and Its Property Analysis

As Genetic Algorithms handles the constraint optimization problem,the difficulty is how to solve the constraints.According to this problem,a new improved Genetic Algorithms(CIFGA) is proposed.The key strategy in CIFGA is that the whole infeasible individuals,appeared after each generation,will be transformed into feasible ones.Go through generation after generation,the optimum solution of optimization problem can be founded.The CIFGA,with either binary coding or real coding,is proved to converge to global optimum solution.The experimental results show that CIFGA has great advantage of convergence property over the GAs based on Penalty Function(PFGA),and has good ability of solving constrained optimization in general purpose.