Ant Colony Algorithm Applied Research

Ant colony algorithm was proposed first by Italy scholar M.Dorigo. It is another heuristic search algorithm applied in combinational optimized problem followed by simulated annealing algorithm, heredity algorithm, taboo search algorithm, ANN algorithm and so on. The experiment indicates the algorithm has good capability of finding the solution. But ant colony algorithm has some disadvantages as dis-convergence, finding the local solution and so on. This paper research the algorithm model and the applying areas. The work includes as follow.Firstly, it sum up the applying areas of ant colony algorithm and its disadvantages, deeply analyzing the disadvantage in order to improve total capability of ant colony algorithm. It improves the ant colony through by the realizing stochastic ant colony for extent zone, and implements the recent information renewal and the search strategy in order to enhance the solution performance. The new algorithm can adjust the proportion between the stochastic ant colony and the intelligent ant colony in order to controls the convergence rate. The experiment indicates that improve can avoid the algorithm converging quickly and enhance the capability of finding the solution.Secondly, to improve capability of auto classification and the speed of algorithm, the ant colony algorithm based on classification was proposed. The stochastic ants construct classification table, establish intellectual ants and ascertain the center of classification. The intellectual ant classify the image. Meanwhile the all ants begin to extract the edge of image. Comparing the basic ant colony algorithm, the improve can advance the capability of finding the solution and the precision of auto classification.