New transition probablity for Ant Colony Optimization: global random-proportional rule

Aiming at the disadvantage (premature convergence) of the Ant Colony Optimization (ACO), a mechanism called global random-proportional rule (GRP) is presented. The mechanism adjusts the transition probability proportionally and facilitates the exploration by increasing the probability of selecting solution components with low pheromone trail. It can avoid premature convergence of ACO and exploit more strongly solutions. The results show that ACO with GRP are superior to the existing ACO and the mechanism is useful to improve the performance of any versions of ACO by investigating the functioning of GRP in the Traveling Salesman Problem (TSP).