改进人工蜂群算法求解多目标连续优化问题  

Improved Artificial Bee Colony Algorithms for Multi-objective Continuous Optimization Problem

作　　者：葛宇[1] 梁静[2] 王学平[3] 谢小川[1] 

机构地区：[1]四川师范大学基础教学学院,成都610068 [2]成都工业学院网络中心,成都610031 [3]四川师范大学数学与软件科学学院,成都610068

出　　处：《计算机科学》2014年第6期254-259,286,共7页Computer Science

基　　金：四川省教育厅项目:人工蜂群算法及其在多目标优化问题中的应用研究(12ZB112)资助

摘　　要：针对多目标连续优化问题,依据人工蜂群算法原理给出其求解流程,并指出算法中更新策略存在盲目搜索和丢失优秀个体的不足,随后提出改进方案。改进方案包含两部分:首先,设计一种自适应搜索算子,使算法在运行过程中能根据个体质量自动调节搜索范围,让算法搜索行为准确高效;其次,利用外部集合记录下新产生的个体,一次迭代完成后结合外部集合重新构造种群,让算法能有效地保存进化过程中产生的优秀个体。实验中将改进人工蜂群算法与NSGA2算法、改进前算法以及文献报道的同类优秀算法进行了比较,结果说明:改进人工蜂群算法在求解多目标连续优化问题中具有良好的收敛性和均匀性。In order to solve multi-objective continuous optimization problem,this paper gave the solving process according to artificial bee colony algorithm theory,and pointed out that the updating strategy in the algorithm has defect of blind searching and missing good individuals,thus proposed an improved strategy.The improved strategy has two parts.First,a self-adapting searching operator is designed to enable the algorithm to adjust the searching range automatically according to individual quality during the iterative process,leading to a more accurate and efficient algorithm searching process.Second,the newly produced individuals are recorded by external archive,and external archive is combined to reconstruct the colony after a iteration,which can save good individuals in the iterative process.The experiment compares improved artificial bee colony algorithm with NSGA2 algorithm,artificial bee colony algorithm and superior algorithm alike in papers.The comparison result indicates the improved artificial bee colony algorithm has good convergence and uniformity in solving multi-objective continuous optimization problem.

关 键 词：人工蜂群算法 多目标连续优化 更新策略 自适应搜索算子 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]