Parameter Analysis and Simulation Experiment of Ant Colony Optimization on Small-scale TSP Problem

Abstract

Ant colony optimization (ACO) is a simple and efficient bionic intelligent algorithm, which can be applied to various optimization problems. There are many controllable parameters in the ant colony optimization, so the parameter setting has a great impact on the efficiency of the algorithm. In order to further improve the efficiency of the algorithm, many researchers have proposed various improvements. However, the current improved method of ACO ignored the consideration of its original parameter settings. In order to analyze the effect of parameter setting on the efficiency of the ant colony optimization, this article uses the ant colony optimization to solve small-scale TSP problems as an example for simulation experiments. Explore the impact of three parameter pairs on the efficiency of algorithm optimization by setting variable pairs. The experimental results show that when the ant colony size is set to approximately 0.6 times the TSP problem size, a better algorithm efficiency can be obtained. Within the value range of [1, 10], setting the information heuristic factor to 1, and the expected heuristic factor to a range of [5, 10] can get a better efficiency. In the last experiments, the pheromone volatility coefficient is generally better to be set greater than 0.2, and in a small-scale TSP problem, totally randomness has little effect on the optimization accuracy of the algorithm, and the value of pheromone enhancement coefficient has little effect on the efficiency of the algorithm.