To overcome the stagnation of the search in classical ant colony algorithms, the paper defines a new directed
pheromone to represent the global information of searching. Accordingly, the global searching ability and the convergence
speed of the proposed algorithm are enhanced. Furthermore, the probability of premature convergence is low due to the
introduction of the new explore-rate parameter and the modification of global chosen rule, which can increase the probability
of selecting the arcs with low pheromone trail. Finally, the improved algorithm and the ant colony system(ACS) algorithm are
used in several different travelling salesman problems(TSP) for comparing experiments. The results show that the proposed
algorithm has more accurate searching results and faster convergence speed.