Scholarlink

<p>In multiple population evolutionary algorithms, the collaborative search of different populations can effectively improve the convergence speed and population diversity. However, most existing multi-population evolutionary algorithms do not take into account the computational redundancy associated with collaboration between different populations. To address this challenge, this paper proposes a Dual-Population Co-evolutionary Algorithm Based on Adaptive Neuro-Fuzzy Inference System (DPCNF). In DPCNF, we design a dual-population co-evolutionary framework that conducts independent searches according to the optimization characteristics of the two sub-populations. Furthermore, to further improve the execution efficiency of the single population, a method is introduced that dynamically adjusts the population size based on optimization efficiency using the adaptive neuro-fuzzy inference system. Simultaneously, this paper proposes a population merging method based on aggregation density, which increases population diversity by relocating individuals with higher aggregation density in the search space. The convergence of the proposed DPCNF algorithm is theoretically established. Experimental results on the CEC 2017 and CEC 2022 benchmark problems demonstrate the competitiveness of DPCNF, particularly highlighted by its superior convergence speed. Exploration and exploitation analysis verified that DPCNF can efficiently allocate computational resources to reduce computational redundancy. Finally, the path planning application is implemented in a three-dimensional ocean environment modeled from real seamount data, verifying the effectiveness and practical significance of DPCNF.</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

Dual-population co-evolution for path planning using neuro-fuzzy systems

  • 期刊论文

  • 发表日期：

    2026年2月12日

查看全文
  • Zhihua Liu,
  • Jingtao Tang,
  • Honglei Jia,
  • Jingyu Huo,
  • Jiaxing Chen

摘要

In multiple population evolutionary algorithms, the collaborative search of different populations can effectively improve the convergence speed and population diversity. However, most existing multi-population evolutionary algorithms do not take into account the computational redundancy associated with collaboration between different populations. To address this challenge, this paper proposes a Dual-Population Co-evolutionary Algorithm Based on Adaptive Neuro-Fuzzy Inference System (DPCNF). In DPCNF, we design a dual-population co-evolutionary framework that conducts independent searches according to the optimization characteristics of the two sub-populations. Furthermore, to further improve the execution efficiency of the single population, a method is introduced that dynamically adjusts the population size based on optimization efficiency using the adaptive neuro-fuzzy inference system. Simultaneously, this paper proposes a population merging method based on aggregation density, which increases population diversity by relocating individuals with higher aggregation density in the search space. The convergence of the proposed DPCNF algorithm is theoretically established. Experimental results on the CEC 2017 and CEC 2022 benchmark problems demonstrate the competitiveness of DPCNF, particularly highlighted by its superior convergence speed. Exploration and exploitation analysis verified that DPCNF can efficiently allocate computational resources to reduce computational redundancy. Finally, the path planning application is implemented in a three-dimensional ocean environment modeled from real seamount data, verifying the effectiveness and practical significance of DPCNF.