Abstract:

In order to resolve subset problems, a new graph-based ant system called catfish bat algorithm-ant colony optimization (CBA-ACO) is proposed. Based on the subset problem’s structure graph, routes’ probability transition-equation  is used to search for solutions, equivalent routes’ pheromones strengthening is adopted to update pheromone. Some solutions are maintained in archive dynamically. The chaotic map and catfish effect are adopted to improve the bat algorithm (BA) for the enhanced exploration which is initialized by archive information and used to find better solution in case the global optimal solution is not updated after several runs. After one cycle, the best route of this cycle updating and the enhanced exploration updating are two cases of updating pheromone. As a result the convergence speed and searching capability of the algorithm are improved. The algorithm is described, and its complexity is analyzed. Experiments show that the CBA-ACO algorithm has a better stability and capability for obtaining the optimal solution.