Simple ant colony algorithm for combinatorial optimization problems

Abstract

Ant colony optimization (ACO) as a kind of distributed intelligent bionic optimization algorithm has been widely used to solve a variety of optimization problems, especially combinatorial optimization problems. The model and management mechanism of pheromone are very important to the performance of ACO. The Max-Min ant system (MMAS) is a classical ACO algorithm and has unique characteristics in terms of pheromones management. In this paper, we analyze the advantages and disadvantages of MMAS. Then we propose a novel ACO algorithm called simple ant colony optimization (SACO). In SACO, constant pheromone bounds are used and the update amount and initialization of pheromone are also set a constant. One of benefit is can reduce the coupling of parameters. Then we study the parameters setting about the initial value of pheromone and evaporation rate. The effect of parameters on the algorithm performance is also studied by experimental method based on traveling salesman problems. Finally, the performance of SACO is compared with other novel algorithms based on traveling salesman problems to show the feasibility and effectiveness of improvements.