A Reference Vector Guided Evolutionary Algorithm with Diversity and Convergence Enhancement Strategies for Many-Objective Optimization

Maintaining the balance between convergence and diversity is a key issue in evolutionary multi-objective optimization and a challenge in many-objective scenarios. Reference-vector-guided selection is an exemplary method for decomposition-based many-objective evolutionary algorithms (MaOEAs). Aiming at solving or alleviating the defects reference vector guided selection confronts, this paper proposes a reference vector guided evolutionary algorithm with diversity and convergence enhancement strategies (RVEA-DCES) for many-objective optimization. RVEA-DCES introduces two new strategies namely adaptive sparse region filling and convergence-only selection for diversity and convergence enhancement. The former is to improve diversity by adaptively adding solutions into sparse regions while the latter is to prevent the elimination of solutions with prominent convergence performance. Experimental results on WFG test suite up to 15 objectives indicate that RVEA-DCES is highly competitive in comparison with five state-of-the-art MaOEAs.