Self-evolving Ant Colony Optimization and Its Application to Traveling Salesman Problem

Ant colony optimization (ACO) algorithms are a recently developed, population-based approach which has been successfully applied to combinatorial optimization problems. However, in the ACO algorithms, it is difficult to adjust the balance between intensification and diversification and thus the performance is not always very well. In this paper we proposed an improved ACO algorithm in which a self-evolving strategy is adopted so as to acquire a trade-off between intensification and diversification. With this strategy we make each ant evolve by means of multiple pheromone deposition, the total concentration of which includes two parts, that is, one part is deposited by several best ants of the last iteration and the ant that has searched the global-best solution by the current iteration, and the other part is deposited temporarily by each ant itself of the last iteration as well. The proposed algorithm is tested by simulating Traveling Salesman Problem (TSP) and experimental results show that Self-evolving Ant Colony Optimization has superior performance when compared with other existing ACO algorithms.