救灾物资发放问题的动态遗传算法求解

Dynamic genetic algorithm for problems of distributing goods to disaster areas

与传统启发式优化搜索算法相比,遗传算法的主要本质特征在于利用了群体搜索策略和简单的遗传算子.群体搜索使遗传算法得以突破邻域搜索的限制,可以实现整个解空间上的分布式信息探索、采集和继承.这篇文章针对救灾物资发放问题进行了研究,建立了此类问题的数学模型,在分析标准遗传算法的基础上,采用设置摆动适应度函数与条件交叉、变异概率的方式设计了动态遗传算法,并通过求解实际问题对标准遗传算法与设计的动态遗传算法计算结果进行了对比.结果表明该算法在一定程度上动态解决了群体由于缺乏多样性而陷入局部解的问题,能够更大概率地得到最优解,可以说是对遗传算法改进方面的一个尝试,结论对于解决类似问题具有较大的参考价值.

Compared with conventional searching arithmetic, the primary characterof GA is the colonial searching resource and the simple GA operator. Colonial searching breaches the restriction of neighborhood searching. And it can realize exploring, collection, and inheritng. In this paper, we discuss the problem of distributing goods to disaster areas, and establish the optimizing model. On the basis of analyses of standard GA, we design an advanced GA by using pendular fitness function and conditional dominate parameters. We compare dynamic GA with standard GA by making some experiments by Matlab. The result demonstrates that our method can solve the problem of partial convergence, and can give the best result by bigger probability.