摘要
针对基本蚁群算法收敛速度慢、容易出现停滞等缺陷,提出一种新的蚁群优化算法——带有侦察子群的蚁群系统.该算法从整个蚁群中分离出一部分蚂蚁组成侦察子群,在优化过程中侦察子群以一定概率做随机搜索,提高了解的多样性;在信息素更新策略上同时使用本代和全局最优蚂蚁,兼顾了本代和历史的搜索成果;同时还采用LK变异算子,对每次搜索的解进行局部优化.最后对三个典型TSP实例进行了仿真实验,结果表明新的算法不仅能够克服早熟现象,而且能够大大加快收敛速度.
To solve the disadvantages of the basic ant colony algorithm including slow convergent speed and incidental stagnation behavior, a new ant colony optimization algorithm, named the ant system with scouting subgroup （ASSS）, was proposed. In the algorithm a small part of ants were separated and formed a scouting subgroup that random moved at a certain probability to increase results diversity. The pheromone update strategy used the iteration-best-ant and global-best-ant at the same time to make use of both iteration-fruit and history-fruit. LK mutation factor was employed to locally optimize the search results of each step. Three typical traveling salesman problems （TSP） were tested, and the results show that this proposed algorithm can avoid prematurity and speed up convergence.
出处
《北京科技大学学报》
EI
CAS
CSCD
北大核心
2006年第8期794-798,共5页
Journal of University of Science and Technology Beijing
基金
国家自然科学基金资助项目(No.70371057)
关键词
蚁群系统
蚁群算法
蚁群优化
随机搜索
变异算子
ant system
ant colony algorithm
ant colony optimization
random search
mutation factor
