混合遗传算法在随机规划问题中的应用  

Applications of Hybrid Genetic Algorithms to Dynamic Programming Problems

作　　者：赵莉萍[1] 王建华[2] 龙刚 

机构地区：[1]华东理工大学计算机科学与工程系,上海200237 [2]华东船舶工业学院电子与信息系,镇江212003

出　　处：《华东理工大学学报（自然科学版）》2000年第5期547-551,共5页Journal of East China University of Science and Technology

摘　　要：以含有机会约束的生产管理动态规划问题为例 ,基于随机模拟技术的混合遗传算法实现最优化决策 ,构造了数学模型。采用 VB编程 ,计算机模拟结果显示该模型能很好解决生产过程中的最优化决策问题 ,它是简单的基于二进制编码的遗传算法所不能解决的。该算法具有很高的鲁棒性 ,避免了在局部最优解附近徘徊 ,且因为随机规划问题要求许多数学知识 ,而该算法本身并不要求对优化问题的性质作一些深入的数学分析 ,从而对那些不太熟悉数学理论和算法的使用者来说 ,无疑是最方便的。Genetic Algorithms (GA) is a method that imitates evolution of life. It is a searching and optimization algorithm. In the past thirty years, genetic algorithms is succeed in solving some complex global optimization problem. In this paper, an improved GA is given. It adopts a new hybrid operator, and a hybrid coding method. At the same time, we discuss a kind of random dynamic programming problems and construct hybrid genetic algorithms for these problems. On the basis of analyzing and simulating examples of manufacturing decision problem, it shows that the algorithm is efficient to solve these dynamic programming problems. The algorithm is more robust and more convenient for user, it not only avoids plunging into local optimum solutions but also solves the dynamic programming problem quickly without any other mathematics knowledges.

关 键 词：遗传算法 随机规划 优化问题 数学模型 

分 类 号：O221.5[理学—运筹学与控制论] O242.23[理学—数学]