A Two-Stage Evolutionary Algorithm with Two-Archive for Many-Objective Optimization

Multi-objective evolutionary algorithms (MOEAs) encounters many difficulties when dealing with many objective optimization problems. In order to better balance convergence and diversity, we propose a two-stage evolutionary algorithm with two-archive named TSTA. Specifically, the entire search process is divided into two stages, the first stage mainly pursues convergence, using subregion dominance in the convergent archive, and crowding distance sorting in the diversity archive, and the second stage mainly pursues the balance of convergence and diversity, in which the convergence archive remains the same, and the diversity archive is changed to use fitness. To evaluate the performance of our algorithms, 18 benchmark problems are used as the test suite. The experimental results show that TSTA achieves competitive performance in comparison with 6 state-of-the-art MOEAs.