提出了一种求解子集问题的基于图的蚂蚁系统。针对子集问题,定义了构造图和等效路径,提出了基于等效路径增强的信息素更新策略,将问题的无序信息转化为对蚂蚁的有序影响,增加蚂蚁搜索路径的信息量。引入路径变异机制,通过路径的改良调节信息素分布,防止算法陷入停滞状态。将信息素更新分为三种情况:本次迭代最优更新、变异更新和本次迭代不更新,兼顾算法的收敛速度和搜索能力。对算法进行了描述并分析了算法复杂度。以多维背包问题为例,对该蚂蚁系统的性能进行了测试,验证了系统的有效性和优越性。
A graph-based ant system was proposed for solving subset problems. The structure graph and equivalent routes were defined. A new updating pheromone policy based on strengthening the pheromone on equivalent routes was proposed. Ants were influenced orderly by the out-of-order information of the problem, and the information in process of ants’ searching routes was increased. A route mutation mechanism was introduced, so the pheromone distribution was adjusted by amending the routes and the algorithm could be kept from running into the stagnation behavior. After one cycle, the best route of this cycle updating, the best mutated route updating and no updating are three instances of updating pheromone. As a result the convergence speed and searching capability of the algorithm were improved. The algorithm was described, and its complexity was analyzed. Finally, the effectiveness and superiority of the algorithm were illustrated with multidimensional knapsack problem.