Abstract: Research on swarm intelligence provides a better way for distributed control and optimization. Much study has been done on swarm intelligence such as ACO(ant colony optimization), and many applications also have been made in the field of combinatorial optimization. However, when solving combinatorial optimization problems, especially those problems with large scale, slow convergence and easy stagnation still restrain the algorithms being much more widely used. The DEAHACO algorithm is presented, in which a mechanism of dynamic evaporation rate is used to achieve better balance between solution efficiency and solution quality, avoiding algorithm falling into local optimum. To speed up the convergence of the algorithm, we redefine the heuristic information to guide the algorithm converge fast. A boundary symmetric mutation strategy is introduced to get variation of iteration results symmetrically, which not only enhances the mutation efficiency, but also improves the mutation quality. Experimental results show that the DEAHACO algorithm has better performance than other algorithm, and its convergence rate increases by 20% or more. Furthermore, the DEAHACO algorithm in other classic TSP instances also showes good performance.