Solving TSP problems with hybrid estimation of distribution algorithms

Abstract

In this paper, a hybrid Estimation of Distribution Algorithms is proposed to solve traveling salesman problem, and a greedy algorithm is used to improve the quality of the initial population. It sets up aBayes probabilistic model of the TSP. The roulette method is adopted to generate the new population. In order to prevent falling into local optimum, the mutation and limit were proposed to enhance the exploitation ability. At the same time, three new neighborhood search strategies and the second element optimization method are presented to enhance the ability of the local search. The simulation results and comparisons based on benchmarks validate the efficiency of the proposed algorithm. © 2014 Springer International Publishing Switzerland.