Abstract:

Classical ant colony optimization algorithms build solutions by starting with an empty initial solution, and unconditionally accepting selected components. This has become a natural restriction of its intensification ability. To overcome this shortage, an iterative improvement based ant colony optimization algorithm was  presented for the traveling salesman problem. In the process of constructing the solution, the ant always memorizes a complete solution; and it adopts a candidate city only when such an adoption can improve the solution. Reconstructing of a partial solution was used to keep the diversity of swarm and avoid premature convergence. Simulation results showed that the proposed algorithm can obtain better solutions within less iteration numbers.

Key words: ant colony optimization algorithm, iterative improvement, traveling salesman problem, intensification, diversification