The Improvement Of Ant Colony Algorithm

Ant Colony Algorithm is a new developed bionic optimization algorithm, which simulate the ants'finding food process. Because of its simple mechanism, plus-feedback parallelization, strong lustiness and excellent distributed compu- tational mechanism, it has shown its excellent capability and huge developing potential in solving many complicated optimization problems, and it has been successfully applied to production problem such as scheduling problems and routing problems. But it also has some disadvantages, it need longer searching time than others and tend to precocity and stagnation.Basing on the basic ant colony alogrithm and the internatonal current status, Polymorphic Ant Colony Algorithm is the emphasis in this paper. The disadvan- tage in choosing mechanism and pheromone updating mechanism of Polymor- phic Ant Colony Algorithm is presented by experiment. We also analyse the reason of disadvantage detailedly by experiment. Aiming at the Polymorphic Ant Colony Algorithm's disadvantage in the initialization of pheromone and the choice of transition probability, the new algorithms that an polymorphic ant colony algorithm with weight and polymorphic ant colony algorithm with Ant-Q are presented. The weight values are added in the initialization of pheromone and the choice of transition probability in the polymorphic ant colony algorithm with weight. The Ant-Q choice mechanism is adopted in the polymorphic ant colony algorithm with Ant-Q. The two new algorithms both choose the global updating mechanism of basic ant colony algorithm to update the pheromone. They avoid effectively the disavantage of some cities searched repeatedly and some cities can not be searched in Polymorphic Ant Colony Algorithm. With the symmetrical design, the parameters of the two improved algotihms are set rationally. Lastly, the experiment on TSP problem, shows the validity of theparameters and the algorithm.