引入随机子群蚁群算法的JSP仿真与研究  

Simulation and study on JSP by using the ant colony optimization algorithm with a random subgroup

作　　者：胡中华[1] 赵敏[1] 姚敏[1] 撒鹏飞[1] 

机构地区：[1]南京航空航天大学自动化学院,南京210016

出　　处：《兰州大学学报（自然科学版）》2009年第1期120-125,共6页Journal of Lanzhou University(Natural Sciences)

基　　金：国家自然科学基金项目(60477031)

摘　　要：针对标准蚁群算法易于出现早熟停滞等缺陷,在原有算法基础上引入一定比例的随机蚁群.在优化过程中随机蚁群以一定概率做随机搜索,以扩大了解的搜索空间;在信息素更新策略上,为了兼顾当代和历代的搜索成果,采取了信息素混合更新策略,同时增强随机子群的最佳路径信息及剩余全部蚁群路径信息,有效抑制了收敛过程中的早熟停滞现象,提高了算法收敛速度.通过典型作业车间调度问题(JSP)实例进行了仿真实验,表明该算法不仅能够克服早熟现象,而且能够加快收敛速度.To solve the problem that any standard colony algorithm often gets stuck into premature stagnation after the iteration process, an improved algorithm was presented in this paper, which can enlarge the searching space of results by adopting a scouting subgroup, search the route at a certain probability, restrain premature stagnation and speed up constringency by the omnibus pheromone＇s updating strategy. The strategy used the iteration-best-ant of scouting subgroup and remaining subgroup and global-ant at the same time to make use of both the current and past results. Three typical job-shop schedule problem （JSP） samples were practiced and the results showed that this algorithm can avoid pre-maturity and advance constringency.

关 键 词：蚁群优化算法 随机子群 作业车间调度问题 组合优化 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP301[自动化与计算机技术—控制科学与工程]