Improved Strategies of Ant Colony Optimization Algorithms

Ant Colony Optimization (ACO) algorithms, inspired by the foraging behavior of real ants, have achieved great success in tackling discrete combinational optimization problems. Since the first ant algorithm—Ant System was introduced in early 1990s, various improved versions of ant algorithms have been proposed and most of them share similar improving ideas. In this paper, we analyze and compare several typical ACO algorithms that employ different improving methods, and then conclude two sorts of strategies (improvement on the construction of solutions and improvement on the update of pheromone trails) from them. Based on plentiful experiments, we analyze the performance and usage of the two strategies, and prove the effectiveness of them. Largely, the two strategies can guide the design of new ant algorithms or can adopt directly in the new application of ACO algorithms.