Ant colony algorithm with magnetic attractor and its application

Abstract

An modified version of ant colony optimization (ACO), global-route-update ant colony algorithm (GRUACA), is presented in this paper by introducing a new impact factor to the ACO algorithm inspired by Kohonen Self- Organizing Maps (KSOM), and is applied to the tour routing problem (TRP) in this paper. In traditional ACO, the ants cooperate and find the best solution to their tasks guided by pheromone trails of scent. But the convergence speed is slow, and the system is easy to get stuck in local minimum with the increase in the number of iterations. A new impact factor, the so-called magnetic attractor is introduced and proved to be effective for speeding up the convergence. And we divide GRUACA into two similar but different algorithms, GRUACA-I and GRUACA-II, depending on the factor's mechanism. We still propose a corresponding mixed strategy of ant colony to enhance the robustness in GRUACA performance. In order to access the GRUACA performance, one of its applications, which was taken as numerical example in previous research using traditional ant colony algorithm (ACA), genetic algorithm (GA) and simulated annealing (SA) is chosen as a benchmark case study. The experiment shows both of the proposed GRUACA-I and GRUACA-II can escape from the trap of local minimum to a extent and achieve better solutions. And GRUACA-II shows better performance than GRUACA-I in terms of convergence speed and mean value of solutions, while GRUACA-I is better in terms of best solutions and randomness. At last, the results are both found to be satisfactory and demonstrate the effectiveness of the proposed algorithms. © 2012 American Scientific Publishers All rights reserved.