Evolving Ant Colony Optimization Using the Genetic Algorithm
| Autor: | Yi-Jung Tu, 杜宜蓉 |
|---|---|
| Rok vydání: | 2007 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 95 Genetic algorithm (GA) and ant colony optimization (ACO) are the two of heuristic algorithms. The ACO is inspired from the biological behavior of real ants. It was developed as a viable approach with high performance for achieving stochastic combinatorial optimizations. Although the ACO is effective in solving the global optimization problems, there are many parameters, both explicit and implicit, affect the performance of the algorithms since the search processes of the two algorithms are nonlinear and complex. Therefore, the ACO with well-selected parameter settings may result in good performance. This thesis proposes an evolving ant colony optimization, which employs a GA to find the best set of parameters employed in the ACO, and apply to the large traveling salesman problem (TSP) for the purpose of obtaining the optimal searching tour. In this thesis, there are ten designed parameters include the number of ant m, three weighting factors q0, ?and β, local and global evaporation coefficients ρlocal and ρglobal, parameter of pheromone Q, number of table list l, number of iteration h, and repeated run number i. The benchmarking cases of traveling salesman problem (TSP) from 14 to 225 nodes are computed to generalize a set of optimal parameters through the evolving ACO for applying to large TSPs with over 300 nodes. The results demonstrated that the presented evolving ACO significantly enhances the solution accuracy and speed up the algorithmic convergence. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|