A Staged Multi-Objective Co-Evolutionary Algorithm

NSGAII has rapid convergence rate, but it still has many problems, such as unsatisfactory distribution. In this paper, we propose a staged multi-objective co-evolutionary algorithm based on NSGAII. In order to evaluate the diversity of belief space, we introduce a diversity index; the algorithm extracts knowledge from belief space and utilizes the knowledge to guide the evolution of population; It proposes a concept called dominant ability of population, according to the ability superior populations and inferior populations are evaluated, and the inferior population is annexed by other populations to avoid the waste of computing source; at the same time the algorithm adopts the method of neighborhood mutation to enhance local search in sparse area, which speeds up the convergence rate of the algorithm. Simulation results indicate that the algorithm improves significantly in terms of the performance of convergence and distribution.