Heterogeneous DPSO Algorithm for DTSP

摘要

A new heterogeneous algorithm of Discrete Particle Swarm Optimization has been proposed in this paper to solve the Dynamic Traveling Salesman problem. The test environment is random and variable in time, what requires a rapid adaptation of the algorithm to changes. An inappropriate selection of the algorithm parameters leads to stagnation and quality deterioration of obtained results. The modification, proposed in this paper, enables to reduce the number of the algorithm parameters, regarding the swarm size, the number of iterations and the size of neighborhood. A higher diversity of particles vs. the homogeneous version positively influences the quality of obtained results, what was demonstrated in various experiments.