基于相对距离和关联度的多任务联盟的蚁群算法

Ant Colony Optimization Algorithm Based on Relative Distance and Association Frequency for the Multi-task Coalition Problem

针对蚁群算法在求解多任务联盟问题(multi-task coalition problem,MTCP)时存在的迭代次数多、求解精度不高的问题,提出了一种基于相对距离和关联度的蚁群算法.该算法针对蚁群算法搜索机制和信息素增量模型,提出了2种策略.首先,为提高资源利用效率,减少Agent的能力浪费,引入了相对距离的概念,提出了基于相对距离的搜索机制;其次,为强化蚂蚁间的协作,利用已获得的解信息,给出了一种基于关联度的信息素增量模型.仿真实验结果表明,与已有的一些算法相比,本文算法不仅能获得更好的联盟结构,而且具有较快的收敛速度.

When solving the multi-task coalition problem （MTCP）, the ant colony optimization （ACO） algorithm showed deficiencies such as too many iterations and low solution accuracy. For problems above, the ACO algorithm based on relative distance and association frequency was proposed, which adopted two strategies in view of search mechanism and pheromone increment model. First, in order to improve the utilization of resources, the concept of relative distance was introduced, based on which, a more effective search mechanism was proposed. Then, to strengthen the collaborations among ants and make full use of answer information obtained, a pheromone increment model based on association frequency was established. Experiment shows that the proposed algorithm can not only get much more optimal solutions but also greatly enhance convergence speed compared with related algorithms.