Study On Double Phase Ant Colony Optimization And Shuffled Frog Leaping Algorithm

Ant Colony Optimization (ACO) is a group intelligent optimization algorithms inspired by nature ant foraging behavior. ACO algorithm for solving the problem of discrete optimization mechanism to solve a lot of problems in science and engineering fields:for example, a variety of classic combinatorial optimization problem, the problem of digital image processing and so on. Despite the positive feedback ant colony algorithm has a pheromone update strategy, greedy heuristic search mechanisms can ensure finding better solutions faster, but when the search process of ant colony algorithm to the late, all the ants explore solutions exactly the same,there is no solution space for further development. As the ant colony algorithm coupled with a strong, easy and other heuristics and bionic optimization algorithm combining, paper selected shuffled frog leaping algorithm (SFLA) integration with ant colony algorithm, so the algorithm running late jump out of local optimal solution, increasing the accuracy of the algorithm to solve. the main content of this paper is the following:First, we propose double phase ant colony optimization (DPACO). In order to introduce pheromone update phase ant colony algorithm update rules, a two-stage ant pheromone on the path traversed updated, we propose two pheromone update phase partitioning strategy:hard partitioning strategy, soft partitioning strategy, Furthermore, the value of the pheromone on each path is limited to an integer interval; Trough testing on the classic traveling salesman problem, results show that the improved algorithm can jump out of local optima, search capabilities of the ant colony algorithm is significantly enhanced.Considering the ant colony algorithm has strong ability to couple with other algorithm, we study the integration of shuffled frog leaping algorithm and double phase ant colony algorithm. First, the probability selection rules of the double phase ant colony algorithm is added with experience feedback operator, making the path of the ants build solutions that can be considered long-term impact on the entire path of the elements selected; second, the local search process using genetic algorithms crossover operator and mutation operator to improve the shuffled frog leaping algorithm makes leapfrog algorithm the greatest degree of avoiding premature convergence. Finally, the introduction of local search algorithms leapfrog the global update process at each step of the iterative process of ant colony algorithm in the second order. Experimental results show that the proposed algorithm on solving TSP problem the quality has greatly improved.