The Improvement And Application Of The Ant Colony Algorithm

Ant Colony algorithm is a novel intelligent optimization algorithm by imitating the principle of real ants looking for food,Its main characteristics are positive feedback,parallelism and the use of aconstructive heuristic.As a new branch of computational intelligence and swarm intelligence,it has attracted more and more attention.In this dissertation,the developmental background,substance and performance of Ant Colony algorithm are introduced particularly,and delve into the algorithm in itself.Make amended scheme and simulated experiment against the defects of algorithm itself and the difference of problem, which is in need of settlement.The major contributions of this dissertation are presented as follows:Against the phenomenon of slowly convergence rate and suboptimization for solving DCMST problem by AS algorithm,a new improved algorithm UDA-ACS is presented in this dissertation.The algorithm adopts the cardinal conception of ACS algorithm,and company with unequal initial pheromone,dynamic negative feedback mechanism,adquately agument the effect of visibility.It is able to restrain the phenomenon of slowly convergence rate and sub-optimization effectively,and enhance the quality and velocity of search.The result of simulated experiment show that improved algorithm emerge more superiorities.A new improved algorithm is presented for MCP in this dissertation.the contribution of this algorithm compare with Vertex-AC algorithm includes that heuristic information based vertex-degree bearing a part in pheromone conduct collectively the choice of next vertex,local pheromone update bearing a part in global pheromone update accomplish collectively the update of vertex pheromone,and employing pseudorandom proportional rule modulate the balance of seeking for the better solution and imposing the available information.It improves effectively the manner of simplex pheromoneupdate and the mechanism of only pheromone conducting the choice of next vertex,it brings on the efficiency of algorithm and the quality of solution. Contrastive experiment employing MCP benchmark test instances verify its' advantage on performance.