An improved genetic algorithm for the multiple traveling salesman problem

Abstract

In this paper, an improved genetic algorithm for the multiple traveling salesman problem was proposed. In the algorithm, a pheromone-based crossover operator is designed, and a local search procedure is used to act as the mutation operator. The pheromone-based crossover can utilize both the heuristic information, including edge lengths and adjacency relations, and pheromone to construct offspring. Experimental results for benchmark instances clearly show the superiority of our genetic algorithm.