Ant Colony Optimization Based On Predatory Search And Its Applications In Traveling Salesman Problem

Traveling Salesman Problem is a classic problem in the research of intelligent algorithm,which is also a benchmark in testing the performance of intelligent algorithm.This paper tries to research on Ant Colony Algorithm (ACO), which is one of the bionicalgorithms . This paper hopes to use Predatory Search (PS) to improve the search process ofACO, and to overcome the defect of the rapid convergence of ACO, and at last, extend thesurvival time of ACO.The work of this paper includes the following aspects:1) Briefly introduce combinatorial optimization problems, which are major applicationsof intelligent algorithms, and briefly introduce Traveling Salesman Problem (TSP).2) The basic principles, mathematical model and implement processes of the ACO areintroduced in detail, and analyze the advantages and disadvantages of the ACO. In addition,this paper introduces the development of ACO, thinks about the future improving idea, andmakes attempt to improve ACO, while at last, drew some interim results.3) The auxiliary algorithm, PS, is introduced. The balance of local and global search inPS is analyzed.4) Based on the characteristics of ACO and PS, this paper introduces a new algorithm ofPSACO. PSACO use PS to optimize one of local optimized solution of ACO, at the momentof ACO falling into local optimized solution, and then optimize the convergence of ACOaccording to the method of adjusting pheromone matrix using the optimized solution.5) Use TSP as platform to build the framework and process of the PSACO. And make lotsof experiment to discuss parameters and settings of PSACO, draw a set of parameters adapt todifferent size problems. At last, PSACO is applied to several TSP, and experimental resultsshow that PSACO optimize the convergence process, so that ACO jump out of local optimum,and the survival time of ACO extends.