A genetic algorithm with jumping gene and heuristic operators for traveling salesman problem

Abstract

Aiming at the problems of slow convergence speed, low solution quality, and easily falling into a local optimum in solving traveling salesman problem (TSP) with genetic algorithm (GA), a genetic algorithm with jumping gene and heuristic operators (GA-JGHO) is proposed, which contains five modifications: (1) an improved roulette selection of combined fitness function is proposed to maintain population diversity and strengthen the exploitation ability, which is helpful to overcome the low population diversity with the standard roulette selection; (2) a bidirectional heuristic crossover (BHX) operator is proposed, which aims to increase the possibility of the potential offspring produced by crossover operation; (3) the combination mutation operator is presented to balance the exploration and exploitation ability; (4) a jumping gene operator is designed, which is beneficial to expand the searching space and reduce the possibility of falling into a local optimum; (5) a unique operator is added to avoid the occurrence of nimiety identical individuals in the population. Besides, the local search operator is integrated to enhance exploitation ability. Moreover, a large number of instances from TSPLIB and a real-world path optimization problem of the cruise robot are selected to verify the validity of the modifications and the potential of GA-JGHO. Experimental results and statistical analyses demonstrate that GA-JGHO performs better in quality stability, accuracy, and convergence speed compared with the other six algorithms.