IMPROVING PERFORMANCE OF ACO ALGORITHMS USING CROSSOVER MECHANISM BASED ON BEST TOURS GRAPH

摘要

Several algorithms have been proposed that are based on the ant colony optimization (ACO) meta-heuristic in literature. This paper proposes an extra data structure that we called best tours graph feeding the pheromone trail information for ACO algorithms. Best tours graph is a table that blends the information on the global best tours encountered statistically during iterations and includes the strengths of edges. The table is crossover of favorable edges, and the technique provides a simple crossover mechanism. A powerful pheromone reinforcement mechanism is also developed based on the best tours table to increase the performance of ACO algorithms in this study. Algorithms are tested on Traveling Salesman Problem using TSPLIB. Our experiments and comparisons show that the method improves the performance of almost all original ACO algorithms.