A novel strategy of initializing the population size for ant colony optimization algorithms in TSP

The ant colony optimization (ACO) algorithm belonging to swarm intelligence methods has been used to solve quantities of optimization problems. Among those problem, the travelling salesman problem (TSP) is a very essential application of ACO algorithm, which displays the great ability of ACO algorithm to find short paths through graphs. However, the existing ant colony optimization algorithms still perform a low efficiency in solving TSP within a limited time. In order to overcome these shortcomings, a hypothesis about initializing the population size for ACO algorithms is put forward, based on the analysis of the relationship among the initial number of ant, the average optimal solution and the computational cost. Furthermore, some experiments are implemented in six datasets, and the results prove that the hypothesis is reasonable and reveal that the initial population size is relevant to the number of cities in a dataset. Based on the hypothesis, this paper proposes a novel strategy of initializing the number of ants for ACO algorithms in TSP, so that the relative high-quality optimal solutions can be obtained within a short time.