Abstract: The traditional ant colony algorithm for TSP problem, convergence speed is slower and is easy to fall into local optimum. Aiming at the deficiency of traditional ant colony algorithm, the algorithm is improved, a kind of new self adaptive ant colony algorithm is put forward. Early in the search, pheromone volatilization coefficient is bigger then decreases to a constant value, ensure to search quickly at the early iteration stage and to search accurately at the late stage. The simulation results show that, the improved algorithm has faster convergence speed and higher precision.

Key words: traveling salesman problem, adaptive ant colony algorithm, search, rate of convergence, precision