Abstract: Ant Colony Algorithm （ACA） is meta-heuristic which has some typical shortcomings, such as long-computation-time, stagnation behavior. For avoiding these, the references such as [α,β,ρ,Q] and [m] etc. are studied for discovering their inertial relation, and their dynamic models are constructed with interest level, acceleration and uniformity. Based on these models, an Acceleration Ant Colony Algorithm （AACA） is proposed, which is improved from ACA by modifying the pheromone updating rule and the transition rule. Simulation shows that the AACA can solve the contradictory between convergence speed and stagnation behavior efficiently and has a better solution.

Key words: ant colony algorithm, convergence, acceleration, interest level, uniformity

摘要： 为解决蚁群算法的收敛速度和全局最优性的矛盾，通过引入均匀度、兴趣度以及加速度等概念，对算法中[α、][β、][ρ、][Q、][m]等参数进行分析，研究了参数的内在联系，建立了参数的动态模型，对算法的转移策略和更新策略进行改进，构造了具有自适应功能的蚁群算法。实验结果表明，该算法在性能上优于基本蚂蚁系统。

关键词: 蚁群算法, 收敛性, 加速度, 兴趣度, 均匀度