Scholarlink

<p>To address the premature convergence of the sparrow search algorithm (SSA) caused by over-reliance on fitness value for producer selection and insufficient utilization of population information, an enhanced SSA with hybrid producer selection and scale-free network-guided update (HSFSSA) is proposed in this paper. Firstly, a producer selection mechanism integrating fitness value, relative distance, and cosine similarity is introduced to ensure producers are both high-quality and widely distributed in the search space. At second, opposite solutions are generated from the weighted centroid of producers and refined via quasi-affine transformation to fully exploit potential high-quality solutions. Finally, the population is mapped to scale-free network nodes, allowing scroungers to update positions using neighbor information and thus improving information exchange efficiency. Experimental results combined with Friedman and Wilcoxon rank-sum tests of the proposed HSFSSA with comparisons against the vanilla SSA, the four advanced SSA variants, and the five state-of-the-art evolutionary algorithms on the public CEC2017 and CEC2022 benchmark suites, illustrate that HSFSSA outperforms all co-evaluated algorithms. HSFSSA was also applied to the unmanned aerial vehicle (UAV) path planning problem, and can also achieve superior performance.</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

Enhanced sparrow optimization algorithm with hybrid producer selection and scale-free network-guided update

  • 期刊论文

  • 发表日期：

    2025年11月19日

查看全文
  • Dahai Li,
  • Rui Huang,
  • Zhendong Wang

摘要

To address the premature convergence of the sparrow search algorithm (SSA) caused by over-reliance on fitness value for producer selection and insufficient utilization of population information, an enhanced SSA with hybrid producer selection and scale-free network-guided update (HSFSSA) is proposed in this paper. Firstly, a producer selection mechanism integrating fitness value, relative distance, and cosine similarity is introduced to ensure producers are both high-quality and widely distributed in the search space. At second, opposite solutions are generated from the weighted centroid of producers and refined via quasi-affine transformation to fully exploit potential high-quality solutions. Finally, the population is mapped to scale-free network nodes, allowing scroungers to update positions using neighbor information and thus improving information exchange efficiency. Experimental results combined with Friedman and Wilcoxon rank-sum tests of the proposed HSFSSA with comparisons against the vanilla SSA, the four advanced SSA variants, and the five state-of-the-art evolutionary algorithms on the public CEC2017 and CEC2022 benchmark suites, illustrate that HSFSSA outperforms all co-evaluated algorithms. HSFSSA was also applied to the unmanned aerial vehicle (UAV) path planning problem, and can also achieve superior performance.