A Many-Objective Evolutionary Algorithm Based on Dual Selection Strategy.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Evolutionary Computation

Background:

• Many-objective optimization problems (MaOPs) in high-dimensional spaces challenge existing algorithms.
• Balancing convergence and diversity is difficult as objectives increase.
• Distinguishing non-dominated solutions and assessing diversity become problematic.

Purpose of the Study:

• To propose a novel many-objective evolutionary algorithm, MaOEA/DS, to address convergence-diversity balance.
• To enhance the ability to distinguish superior solutions and improve population diversity.
• To reduce selection pressure in many-objective optimization.

Main Methods:

• Developed a new distance function for effective distance metric calculation.
• Introduced a point crowding-degree (PC) strategy for superior solution identification.
• Implemented a dual selection strategy focusing first on convergence, then on diversity.

Main Results:

• MaOEA/DS demonstrated superior overall performance compared to state-of-the-art algorithms.
• Experimental results validate the effectiveness of the proposed dual selection strategy.
• The PC strategy effectively enhances the distinction of superior solutions.

Conclusions:

• The proposed MaOEA/DS effectively balances convergence and diversity in many-objective optimization.
• The novel distance function and PC strategy contribute to improved performance.
• MaOEA/DS offers a promising approach for tackling high-dimensional optimization challenges.