Scholarlink

<p>Dynamic Multi-Objective Evolutionary Algorithm (DMOEA) provides a powerful and effective tool for solving Dynamic Multi-Objective Optimization Problems (DMOPs). At present, the research focus of DMOEA is how to accurately predict the new positions of Pareto optimal solution Set (PS) or Pareto optimal Front (PF) in the new environment. Therefore, a DMOEA algorithm based on Multi-Point Perturbation Centroid prediction and Hybrid Spiral prediction (MPPCHS) is proposed. MPPCHS divides DMOEA into two main stages and adopts different strategies for each stage. In the early stage, a multi-point perturbation centroid prediction strategy is employed to evenly distribute the population into two sub-populations. The first sub-population adopts the perturbation strategy, randomly selects individuals in the PF and adds perturbation vectors to them to explore new solution space regions. The second sub-population uses the multi-point initialization strategy to generate multiple new solutions around the initialization point, thereby increasing the diversity of the population and providing a uniformly distributed population basis for the subsequent stage. In the later stage, a hybrid spiral prediction strategy is used, which dynamically adjusts the weight according to the quality of the solution, balancing exploration and exploitation. Experimental results show that compared with several other advanced algorithms, MPPCHS exhibits higher global exploration ability and stronger robustness when dealing with DMOPs.</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

Dynamic multi-objective evolutionary algorithm based on multi-point perturbation centroid prediction and hybrid spiral prediction strategy

  • 期刊论文

  • 发表日期：

    2026年6月2日

查看全文
  • Ying Gan,
  • Yongjie Ma

摘要

Dynamic Multi-Objective Evolutionary Algorithm (DMOEA) provides a powerful and effective tool for solving Dynamic Multi-Objective Optimization Problems (DMOPs). At present, the research focus of DMOEA is how to accurately predict the new positions of Pareto optimal solution Set (PS) or Pareto optimal Front (PF) in the new environment. Therefore, a DMOEA algorithm based on Multi-Point Perturbation Centroid prediction and Hybrid Spiral prediction (MPPCHS) is proposed. MPPCHS divides DMOEA into two main stages and adopts different strategies for each stage. In the early stage, a multi-point perturbation centroid prediction strategy is employed to evenly distribute the population into two sub-populations. The first sub-population adopts the perturbation strategy, randomly selects individuals in the PF and adds perturbation vectors to them to explore new solution space regions. The second sub-population uses the multi-point initialization strategy to generate multiple new solutions around the initialization point, thereby increasing the diversity of the population and providing a uniformly distributed population basis for the subsequent stage. In the later stage, a hybrid spiral prediction strategy is used, which dynamically adjusts the weight according to the quality of the solution, balancing exploration and exploitation. Experimental results show that compared with several other advanced algorithms, MPPCHS exhibits higher global exploration ability and stronger robustness when dealing with DMOPs.