Multiple Ant Colony Optimization Based on Pearson Correlation Coefficient

Abstract

Ant Colony Optimization algorithms have been successfully applied to solve the Traveling Salesman Problem (TSP). However, they still have a tendency to fall into local optima, mainly resulting from poor diversity, especially in those TSPs with a lot of cities. To address this problem, and further obtain a better result in big-scale TSPs, we propose an algorithm called Multiple Colonies Ant Colony Optimization Based on Pearson Correlation Coefficient (PCCACO). To improve the diversity, first, we introduce a novel single colony termed Unit Distance-Pheromone Operator, which along with two other classic ant populations: Ant Colony System and Max-Min Ant System, make the final whole algorithm. A Pearson correlation coefficient is further employed to erect multi-colony communication with an adaptive frequency. Besides that, an initialization is applied when the algorithm is stagnant, which helps it to jump out of the local optima. Finally, we render a dropout approach to reduce the running time. The extensive simulations in TSP demonstrate that our algorithm can get a better solution with a reasonable variation.

Keywords

• Multiple colonies
• TSP
• Pearson correlation coefficient
• cities-dropout
• Unit Distance-Pheromone Operator