A novel evolutionary algorithm for the traveling salesman problem
Chengjun Li, Yong Xia, Xu Si, Wei Zhan · 2011
The traveling salesman problem (TSP) is a famous NP-hard problem. The established evolutionary algorithms (EAs) cannot get satisfactory solutions of large or even medium scale TSP instances. To change this situation, a effective EA based on inver-over operator is introduced. In this algorithm, two novel crossover operators, the segment replacing crossover and the order adjusting crossover, are proposed. Moreover, a mechanism for changing the crossover and the mutation rate of the inver-over operator according to generations and a parameter, the critical value, is employed. The results of the experiment show that this new algorithm outperforms the basic EA based on the inver-over operator in all the 4 instances.