关键词: 蚁群系统, 参数自适应, 变异算法, 旅行商问题

Abstract: In ant colony algorithm, how to effectively deal with the contradiction between the convergence speed and the precocity and stagnation has been a difficult problem. A parameter adaptive ant colony system is proposed by introducing the crowding degree to strengthen the coordination and cooperation between ants in the search process. In the presented ant colony algorithm, it adopts the proactive strategies to avert the precocity and stagnation in advance, and embeds the crowding degree into the state transition and the pheromone update. The parameter in the local pheromone update adaptively changes with the local search state, and the parameter in the global pheromone update adaptively changes with the global search state. These make its global searching ability enhance remarkably. At the same time a simple and efficient mutation algorithm is adopted to accelerate convergence. Experimental results show that the presented algorithm has much higher quality and stability and convergence speed than that of classical ant colony algorithm

Key words: ant colony system, parameter adaptive, mutation algorithm, traveling salesman problem