A Novel Particle Swarm Optimization With Genetic Operator and Its Application to TSP

Abstract

To solve some problems of particle swarm optimization, such as the premature convergence and falling into a sub-optimal solution easily, the authors introduce the probability initialization strategy and genetic operator into the particle swarm optimization algorithm. Based on the hybrid strategies, they propose an improved hybrid particle swarm optimization, namely IHPSO, for solving the traveling salesman problem. In the IHPSO algorithm, the probability strategy is utilized into population initialization. It can save much more computing resources during the iteration procedure of the algorithm. Furthermore, genetic operators, including two kinds of crossover operators and a directional mutation operator, are used for improving the algorithm's convergence accuracy and population diversity. At last, the proposed method is benchmarked on nine benchmark problems in TSPLIB, and the results are compared with four competitors. From the results, it is observed that the proposed approach significantly outperforms others on most of the nine datasets.