Ant Colony Optimization For The Traveling Salesman Problem Based On Ants With Memory

The scientists who studied the behavior of insects with sociality found that insects'cooperation can solve complicated problems although this kind of cooperation is nearly self-organized in one community and seems very simply in some situation. The study of the behavior of sociality insects supplied the computer scientists powerful solutions to distributed control and optimization.Ant Colony Optimization is the representative example that makes use of group intelligence to solve combination optimization problems. Ant Colony Optimization is a kind of heuristic hunting arithmetic which is applied to combination optimization problems. It possesses positive feedback, distributed calculation and heuristic hunting characters. Ant Colony Optimization is successively applied to TSP issues, resource re-distributed and this kind of classical optimization problems. In this way better effects turned out and many scholars were attracted. The field of the ACO has also been spread into dynamic environment, chaos computer and multiple objects etc., new technology based on ACO also are ongoing in recently years.It only has been more than ten years since Ant Colony Optimization was raised and the whole mathematics system has not been formed. At the same time, it's capability need to be improved. By starting with the NP problem—TSP,this paper gives a specific introduction to the development backgrounds, content, ways to be carried out and capability of Ant Colony Optimization; after deeply studying , we will put forward our special improving methods to it. The major contributions of this dissertation on Ant Colony Optimization are presented as follows:1. Brought in the ants with memorizing capability and realizes it by program that both on ACS and MMAS.2. Respectively simulated the ameliorated ACS and MMAS algorithms on symmetric static TSP of which has ten different sizes. we analyzed many data results and found these advantages: memorized ants could increase the speed of finding the solution while solving the little sized TSP, but the speed and results are worse than the origin ACS and MMAS in TSP whose size are larger.3. Optimized the memorized ants. In Ant System, they use probability-select strategy in local search for better performance, let the ant become high and low probability memorized model, then actualize the new model in code and combine with ACS and MMAS.4. Respectively simulated the ameliorated ACS and MMAS algorithms on ten different symmetric static TSP models. we analyzed many data results and found out that high probability memorized ants got better performances in finding optima solutions than the origin ACS and MMAS in most TSP models.