SEA: Many-objective evolutionary algorithm with selection evolution strategy

Abstract

Balancing the convergence and diversity of the population is crucial for solving multi-objective problems. As the number of objectives increases, the inherent conflict between maintaining diversity and ensuring convergence becomes more significant. To address this challenge, we propose a novel evolutionary algorithm that selectively emphasizes either convergence or diversity, guided by the convergence and diversity indicators of the current population and a predefined priority criterion. During the iterative process, the algorithm strategically aims to either approach the true Pareto front to improve convergence or foster a more uniform distribution within the current Pareto layer to enhance diversity. Continuous monitoring of these indicators enables the algorithm to effectively manage and fine-tune the convergence and diversity of the population. We meticulously evaluated the performance of the proposed algorithm by comparing it with eight state-of-the-art evolution algorithms on 31 benchmark problems. The experimental results unequivocally demonstrated the outstanding performance of the proposed algorithm in solving multi-objective problems. Furthermore, the algorithm can be seamlessly incorporated into other evolution algorithms to strike a delicate balance between diversity and convergence, thereby empowering them to tackle challenging many-objective optimization tasks with enhanced efficiency and accuracy.