Scholarlink

<p>Multi-objective evolutionary algorithms (MOEAs) have attracted many attentions due to their excellent performance in problems involving multiple conflicting objectives. It is very vital to make the convergence and population diversity of MOEAs reach a better balance in addressing complex problems. For this, a multi-population evolutionary algorithm framework integrating the special method of population selection based on the crowded distance method and the grey wolf optimizer based on refraction (MOMEA) are proposed, and it is applied to the dung beetle algorithm (MOMEA/DBO). In MOMEA/DBO, the refraction learning is utilized to generate new refraction individuals from new individual evolved by elite individuals. The grey wolf optimizer is employed to generate the grey wolf individuals which are added to the population to improve the convergence. Meanwhile, a special selection method is introduced to select better individuals in each evolution, which removes less useful individuals and replaces them with better solutions to improve population diversity. In the end, for verifying the superiority of the MOMEA/DBO algorithm framework, the algorithm is applied to some benchmark test suites (DTLZ, UF, CF). According to the results of the experiments, the performance of MOMEA/DBO is superior to others on multi-objective optimization problems.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A multi-population evolutionary algorithm integrating special selection methods and refraction-based grey wolf optimization for multi-objective problems

  • 期刊论文

  • 发表日期：

    2025年9月11日

查看全文
  • Zilong Wang,
  • Peng Shao,
  • Shaoping Zhang

摘要

Multi-objective evolutionary algorithms (MOEAs) have attracted many attentions due to their excellent performance in problems involving multiple conflicting objectives. It is very vital to make the convergence and population diversity of MOEAs reach a better balance in addressing complex problems. For this, a multi-population evolutionary algorithm framework integrating the special method of population selection based on the crowded distance method and the grey wolf optimizer based on refraction (MOMEA) are proposed, and it is applied to the dung beetle algorithm (MOMEA/DBO). In MOMEA/DBO, the refraction learning is utilized to generate new refraction individuals from new individual evolved by elite individuals. The grey wolf optimizer is employed to generate the grey wolf individuals which are added to the population to improve the convergence. Meanwhile, a special selection method is introduced to select better individuals in each evolution, which removes less useful individuals and replaces them with better solutions to improve population diversity. In the end, for verifying the superiority of the MOMEA/DBO algorithm framework, the algorithm is applied to some benchmark test suites (DTLZ, UF, CF). According to the results of the experiments, the performance of MOMEA/DBO is superior to others on multi-objective optimization problems.