个体速度差异的蚁群算法设计及仿真  

Design and simulation of an ant colony algorithm based on individual velocity differences

作　　者：印峰[1] 王耀南[1] 刘炜[2] 周良[1] 

机构地区：[1]湖南大学电气与信息工程学院,湖南长沙410082 [2]湖南科技职业学院软件学院,湖南长沙410118

出　　处：《智能系统学报》2009年第6期528-533,共6页CAAI Transactions on Intelligent Systems

基　　金：国家科技支撑计划资助项目(2008BAF36B01);国家"863"计划资助项目(2008AA04Z214)

摘　　要：针对如何提高蚁群算法搜索速度及防止算法停滞问题,提出一种改进的蚁群优化算法VACO(ACO algorithm based on ant velocity),通过构造与局部路径和蚂蚁个体速度相关的时间函数,并建立与时间函数相关的动态信息素释放机制,加快信息素在较优路径上正反馈过程,从而提高了算法的收敛速度;采取一种连续小区间变异策略,在加快局部搜索过程的同时可有效防止算法陷入局部最优.对典型TSP问题的仿真研究结果表明,改进后的算法在收敛性和对较好解的探索性能得到一定程度的提高.A new implementation of the ant colony optimization（ACO） algorithm was primarily focused on improving search speed and preventing stagnation.To resolve these two issues,improvements based on velocity were proposed,producing a VACO algorithm.By constructing a time-function for local paths and ant velocity,and building a dynamic release mechanism for pheromones in the time-function,it accelerated positive feedback from the accumulation of pheromones,leading to better paths and improved convergence speed.A strategy of continuous inter-cell mutation sped up local searches and at the same time effectively prevented the algorithm being trapped in local optimums.The results showed that the proposed algorithm improves convergence and increases the possibility of finding optimal solutions.

关 键 词：蚁群算法 旅行商问题 信息素 NP-难解 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]