摘要
提出一种自适应的多种群蚁群算法用于求解频率分配问题.算法将蚂蚁群体划分为若干个子群体,每个子群体的蚂蚁并行地进行优化.在寻优过程中,算法为每个蚂蚁子群体定义一个收敛系数,根据收敛系数来决定子群体内部的路径的选择和信息量的更新.算法同时根据各个子群体的解的质量和分布情况来自适应地决定信息交流策略,包括选择信息交流的对象和调节信息交流的周期以及信息更新策略,以取得各蚂蚁子群体中解的多样性和收敛性之间的动态平衡.对固定频率分配和最小跨度频率分配问题在并行计算机上的实验结果表明,本文算法不仅具有较快的全局收敛速度,而且有高质量的解和高的效率.
An adaptive multi colony ant algorithm for solving frequency assignment problem was presented. The algorithm divided the ant population into several subgroups, each of which made the optimization independently. In the procedure of optimization, the algorithm computes the convergence coefficient of each subgroup which was used for choosing the path for the ants and updating the pheromone in the paths. We also presented an adaptive strategy to adjust the time interval and the partner of information exchange adaptively according to the quality and the diversity of the solutions so as to keep the balance between the diversity and convergence in each subgroup. We tested our algorithm on fixed and minimal span frequency assignment problems using parallel computer, the experimental results showed that it had high global convergence speed, high quality of solution and efficiency.
出处
《小型微型计算机系统》
CSCD
北大核心
2006年第5期837-841,共5页
Journal of Chinese Computer Systems
基金
国家自然科学基金(60074013)资助
国家科技攻关项目(2003BA614A-14)资助
江苏省自然科学基金(BK2005047)资助
南京大学软件新技术国家重点实验室开放基金资助
关键词
多种群
收敛系数
自适应
multi colony
coefficient of convergence
adaptive
