Multi-ant colony algorithm based on cooperative game and dynamic path tracking

Abstract

For the problems of slow convergence and low stability in traditional ant colony algorithm when solving large-scale Traveling Salesman Problem (TSP), a multi-ant colony algorithm based on cooperative game and dynamic path tracking (CDMACA) is proposed. Firstly, a novel heterogeneous multi-population is formed by Ant Colony System (ACS) and Max–Min Ant System (MMAS). Secondly, a cooperative game is introduced. In the isomorphic populations, all the populations cooperate and then the pheromone income will be distributed to the participants based on the contribution. Populations with higher contribution will obtain more income, which can improve the accuracy of the solutions; On the other hand, in the heterogeneous populations, a learning mechanism is introduced to enable heterogeneous populations to learn the optimal solution from each other to further improve the accuracy. Thirdly, a dynamic path tracking mechanism is proposed to reward or punish the public paths with pheromone based on the similarity of the optimal solutions, which can enhance the concentration of pheromone on the optimal path, thus improving the convergence. Finally, when the algorithm stalls, a pheromone balance mechanism is introduced. By dynamically cutting the pheromone on the current optimal path, the probability of the optimal path being chosen will decrease, which can help the algorithm jump out of the local optimum. Through a large number of TSP instances, the proposed algorithm in this paper can further improve the accuracy when solving large-scale TSP compared with other improved algorithms.