Symmetrical Generalized Pareto Dominance and Adjusted Reference Vector Cooperative Evolutionary Algorithm for Many-Objective Optimization DOI Open Access
Shuwei Zhu, Lingzao Zeng, Meiji Cui

et al.

Symmetry, Journal Year: 2024, Volume and Issue: 16(11), P. 1484 - 1484

Published: Nov. 6, 2024

In Pareto-based many-objective evolutionary algorithms, performance usually degrades drastically as the number of objectives increases due to poor discriminability Pareto optimality. Although some relaxed domination relations have been proposed relieve loss selection pressure, it is hard maintain good population diversity, especially in late phase evolution. To solve this problem, we propose a symmetrical Generalized Dominance and Adjusted Reference Vectors Cooperative (GPDARVC) algorithm deal with optimization problems. The symmetric version generalized dominance (GPD), an efficient framework, provides sufficient pressure without degrading no matter objectives. Then, reference vectors (RVs), initially generated evenly objective space, guide diversity. cooperation GPD RVs environmental part ensures balance convergence Also, further enhance effectiveness RV-guided selection, regenerate more according proportion valid RVs; thereafter, select most for adjustment after association operation. validate GPDARVC, compare seven representative algorithms on commonly used sets This comprehensive analysis results 26 test problems different numbers 6 practical problems, which show that GPDARVC outperforms other cases, indicating its great potential

Language: Английский

Symmetrical Generalized Pareto Dominance and Adjusted Reference Vector Cooperative Evolutionary Algorithm for Many-Objective Optimization DOI Open Access
Shuwei Zhu, Lingzao Zeng, Meiji Cui

et al.

Symmetry, Journal Year: 2024, Volume and Issue: 16(11), P. 1484 - 1484

Published: Nov. 6, 2024

In Pareto-based many-objective evolutionary algorithms, performance usually degrades drastically as the number of objectives increases due to poor discriminability Pareto optimality. Although some relaxed domination relations have been proposed relieve loss selection pressure, it is hard maintain good population diversity, especially in late phase evolution. To solve this problem, we propose a symmetrical Generalized Dominance and Adjusted Reference Vectors Cooperative (GPDARVC) algorithm deal with optimization problems. The symmetric version generalized dominance (GPD), an efficient framework, provides sufficient pressure without degrading no matter objectives. Then, reference vectors (RVs), initially generated evenly objective space, guide diversity. cooperation GPD RVs environmental part ensures balance convergence Also, further enhance effectiveness RV-guided selection, regenerate more according proportion valid RVs; thereafter, select most for adjustment after association operation. validate GPDARVC, compare seven representative algorithms on commonly used sets This comprehensive analysis results 26 test problems different numbers 6 practical problems, which show that GPDARVC outperforms other cases, indicating its great potential

Language: Английский

Citations

0