Discrete Bat Algorithm for Traveling Salesman Problem

In this paper, we present discrete bat algorithm (DBA) for solving the Traveling Salesman Problem (TSP). In this improved bat algorithm, on the one hand, the subtraction operator of location and location, the multiplication operator of real number and location, and the addition operator of velocity and location are redefined, on the other hand, the initial population are generated by Nearest Neighbor tour construction heuristic, and 2-opt edge-exchange algorithm is introduced to perform the local search. We test series of numerical instances by using 33 benchmark instances with sizes ranging from 55 to 318 nodes from the TSPLIB, and compare DBA with Chen and Chien's method (2011), Marinakis et al.'s method (2005), and Marinakis et al.'s method (2005). The experimental results show that the percentage deviations of DBA are better than that of the three methods and within satisfaction.