Study On Applications And Improvement Of Ant Colony Optimization In Logistics System

The modern society people lived is composed with many complex network systems, such as the computer information network, telephone communication network, and energy, assign network, transport service network.As the research object becoming complicated gradually, some traditional certain optimal algorithm based on exact model had encountered difficulty in resolving specific problems, people are enlightened by biology evolution and bionics, and putting forward a lot of heuristic intelligence optimization algorithm. ACO (Ant Colony Optimization) is one of them; it is very suitable to resolve the compound optimization problems. The algorithm was presented by Dorigo and others in the early 1990, more and more people paid attention to it and gave out many improved algorithm, and many of them were applied to some fields successfully.As the ACO algorithm is simulating the way that ants look for food, it adopted a parallel computation mechanism and has strong robustness and is easy to combine with other methods, but its searching speed is slow with much computation and it is easy to fall into the local best results as other evolutionary algorithm. Because of the deficiency, many scholars had presented a lot of improvements for basic ACO. In this dissertation, improvement, application in logistics system optimization and simulate studying of ACO are researched mainly, the main content studying in this article are as follows:In the distribute center selecting problem of material dispatch in two logistics optimization problems, an improved ACO algorithm is presented to make good use of its parallel and positive feedback mechanism, to bring in the local renew regulation to enforce the positive feedback mechanism as to improve search speed, increasing the pheromone on the local optimal route of a loop in the global renew regulation,. In the last half loops, the strategy of pheromone trail smoothing is carried out,which can enhance the probability that ants select the trail with less pheromone,and can increase algorithm's search ability. Simulation results of the example demonstrated the effectiveness of this improved algorithm to solve the compound optimization problem in practical application problems. Because of the TSP model of the second logistics optimization problem and the deficiency of basic ACO, a grouped ACO algorithm with award & penalty strategy is presented to improve the ACO algorithm,grouping the ants and taking advantage of the cooperation between and inside the grouped ants, adopting the global and local renew regulation, renewing the pheromone with award & penalty strategy and improved MMAS strategy. The data of simulation show that it avoid stopping and falling to the local best results. Also, it gets the balance between the results and searching speed.The problem that is to find the optimal Steiner position(s) to struct the Steiner shortest tree in Logistics system supply chains network optimization, which is hardly resolved by traditional algorithms for these non-deterministic polynomial problems (NP).An improved ACO algorithm with MST algorithm to solve the practical application is studied. The experimental results for examples are proved to be valid.