A modified shuffled frog leaping algorithm for the traveling salesman problem

Abstract

This paper regards a feasible travel route as a frog and presents a modified shuffled frog leaping algorithm (SFLA) for the traveling salesman problem (TSP). To improve the quality of initial population, some frogs are generated according to the simple near neighborhood rule. To adjust the position of the worst frog, the ROXIM operator, which synthetically takes advantage of the virtues of the revised order crossover (ROX) and the inversion mutation (IM), is developed and employed during the local search procedure. To evaluate the performance of the modified SFLA and the ROXIM operator, the experiments on Br17, P43 and Berlin52 are respectively designed and conducted. Experimental results show that the modified SFLA suits for the given TSPs and the ROXIM operator is beneficial to the algorithm's robustness and the solution's precision both for symmetric and asymmetric TSP.