An Adaptive Dominance and Reference Vector Based Evolutionary Algorithm for Many-objective Optimization

Improved dominance method improve the ability to approach the Pareto front, but they show poor ability in balancing convergence and diversity of population. To address this issue, this paper proposes an adaptive dominance and reference vector based evolutionary algorithm for many-objective optimization (ADRVEA). Firstly, the adaptive dominance (AD) is proposed to design an niche mechanism. Then the reference vector is adopted to divide the objective space for the purpose of the enhanced diversity. Finally, the fitness function is constructed to complete the elite selection. The experimental results demonstrated that the proposed ADRVEA not only has significant performance, but also effectively balances the convergence and diversity.