Based on analyzing the development of ACO algorithms, this paper brings up a new ACO algorithm NDLACO by studying MMAS and an adaptive ant colony algorithm based on equilibriem of distribution. After researching local search method and observing the degree that ants crawl in the solution space with analysis of fitness-distance correlation, both the nearest neighbor and 3-opt local searching methods are integrated to NDLACO algorithm. NDLACO also adapts the adaptive idea based on equilibriem of distribution and deal with the contradictory between ants' convergence speed and avoiding precocity and stagnation. The thesis solves the capacitated vehicle routing problems(CVRP) successfully using NDLACO algorithm. Based on analyzing the results obtained by concerned experiments, we find NDLACO can tolarant some parameters's variaties. Wherever, it does not influence the quality of solutions. |