一种基于监工机制的改进蚁群算法

针对基本蚁群算法存在收敛速度慢、易陷入局部最优解等问题,受监工机制的启发,提出了监工蚁群算法,以监工距离作为评价标准,自适应地选择优良的蚂蚁更新信息素,提高了每次迭代中解的质量,指导之后的蚂蚁进行更好的学习。该算法选用优化的全局更新策略,使得信息素在进化前期增加较多,在后期增加较少;同时,自适应地将信息素的值限定在一定范围内,防止某条路径被选择的概率过大或者过小。该算法还添加了发散和收敛机制,当算法陷入局部最优解时,增加探索的概率,有助于跳出局部最优解。仿真结果表明,监工蚁群算法具有较高的全局寻优能力,减少了迭代次数,增强了算法的稳定性。

New ant colony optimization algorithm based on supervisory mechanism

Given such problems as slow convergence speed and premature convergence existing in basic colony optimization algorithm, and enlightened by supervisory mechanism, the supervisor ant colony optimization (SACO) algorithm was introduced. With the supervisory distance as an evaluation criterion, SACO self-adaptively adopted excellent ants to update pheromone trails, thus improving the solution qualities of each iteration, and a better guide was made for the ants later. The optimized global pheromone trail strategy was selected in the prophase of the evolution in SACO, and the pheromone trail was added more, whereas in the anaphase the pheromone trail was added less. Moreover, the pheromone trail was adaptively limited to a certain range, avoiding the selecting probability of a path being too large or too small. When the SACO converges to an optimal solution, the exploring probability is adaptively increased, which helps to jump out of the local optimal solution. The simulation experiments show that SACO not only obtains stable and optimal solutions but also enhances the convergence speed.