Ant-TSL System Algorithm using New Ant Agents with Intensification and Diversification Strategies
スポンサーリンク
概要
- 論文の詳細を見る
Ant system algorithm (AS) proposed by Dorigo and others is a new approach for stochastic combinatorial optimization. They applied the proposed methodology to the classical Traveling Salesman Problem (TSP), and reported simulation results. The results show that the AS for TSP was as effective as tabu search and better than simulated annealing. However, when applying this AS to randomly generated graphs, there is a tendency for the solutions obtained using the AS to be trapped in bad solution. Therefore, we attempt to escape bad solution by improving the original AS, using the following strategies. First, we designed a new agent by using intensification and diversification strategies, such as the tabu search applies, in order to obtain better solutions. And, we tried to solve the problem by using new agents with the ability of local search. Furthermore, the parallel ant system algorithm by the above-mentioned new agents was implemented to reduce computational time. Finally we discuss the characteristics of proposed AS.
- 社団法人日本経営工学会の論文
- 2009-02-15