求解约束优化问题的一种复合形遗传算法  

A Complex-genetic Algorithm for Solving Constrained Optimization Problems

作　　者：李旻松[1] 王惠平[1] 章芬芬[1] 钟若武[1] 

机构地区：[1]韶关学院计算机科学学院,广东韶关512005

出　　处：《计算机仿真》2011年第6期259-261,370,共4页Computer Simulation

基　　金：广东省自然科学基金博士科研启动项目(9451200501002983);韶关市技术创新项目(韶科(成)2008-03)

摘　　要：研究约束优化问题是科学和工程应用领域经常会遇到的一类数学规划问题。现有的约束优化进化算法,通常的解决办法是将等式约束条件转化为成对的不等式约束条件来处理,转换会使得可行域的拓扑结构变化显著,直接影响了算法性能和解的精度。为解决上述问题,提出了一种改进的处理约束优化问题的新算法。新算法将约束优化问题转化为多目标优化问题,把复合形法嵌入到遗传算法中,通过将全局搜索和局部搜索机制有机地结合,利用遗传算法全局性好和复合形法快速高效的特点,以加快最优解的搜索进程。仿真结果表明,方法既有复合形法快速高效的特点,又有遗传算法全局性好的特点。与标准遗传算法相比,方法具有良好的求解约束优化性能和精度效果。Constrained optimization problems（COPs） are a kind of mathematic programming problems frequently seen in science and engineering application realm.In dealing with all kinds of COPs,the existing constrained optimization evolutionary algorithms（COEAs） usually convert equality constraints into pairs of inequality constraints,and the conversion causes a noticeable change in the topological structure of feasible region,which has a direct influence on the performance of algorithm and solution precision.Considering the deficiency of existing COEAs,a new improved algorithm,which converts COPs into multi-objective optimization problems and effectively combines multi-objective optimization with global and local search,was proposed to handle COPs.Complex-GA combines the advantages of the two methods and overcomes the disadvantages of both and increases the speed of optima search remarkably.Compared with standard GA,the results from the experimentation show that this method has a good ability of problem solving.

关 键 词：约束优化 多目标优化 复合形 遗传算法 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]