Angle-Based Dual-Association Evolutionary Algorithm for Many-Objective Optimization

As the number of objectives increases, the comprehensive processing performance of multi-objective optimization problems significantly declines. To address this challenge, this paper proposes an Angle-based dual-association Evolutionary Algorithm for Many-Objective Optimization (MOEA-AD). The algorithm enhances the exploration capability of unknown regions by associating empty subspaces with the solutions of the highest fitness through an angle-based bi-association strategy. Additionally, a novel quality assessment scheme is designed to evaluate the convergence and diversity of solutions, introducing dynamic penalty coefficients to balance the relationship between the two. Adaptive hierarchical sorting of solutions is performed based on the global diversity distribution to ensure the selection of optimal solutions. The performance of MOEA-AD is validated on several classic benchmark problems (with up to 20 objectives) and compared with five state-of-the-art multi-objective evolutionary algorithms. Experimental results demonstrate that the algorithm exhibits significant advantages in both convergence and diversity.