Scholarlink

<p>To effectively address the multi-objective optimization challenges in unmanned aerial vehicle (UAV) path planning under multiple constraints—including path length and collision risk—and to overcome the insufficient adaptability of the traditional Dung Beetle Optimizer (DBO) in this domain, a Mix-Strategy Improved Dung Beetle Optimizer (MIDBO) is proposed. Specifically, the Circle-SPM chaotic map is introduced to optimize the population initialization process, effectively mitigating the premature convergence caused by uneven distribution and a lack of population diversity. A fusion mechanism combining adaptive weights and spiral motion is designed to enhance the global search capability, thereby preventing non-convergence issues stemming from an insufficient search range. Furthermore, a boundary adjustment strategy based on fractional calculus is proposed to provide an effective local-optimum escape mechanism during the foraging phase. Additionally, a piecewise model constraint mechanism is constructed to balance the resource allocation between local exploitation and global exploration from the perspective of optimization stage division. The optimization performance of MIDBO is initially verified using the CEC2017 benchmark functions to ensure a fundamental balance between local and global search. Subsequently, a multi-constraint model is established to further optimize this adaptability, efficiently exploring the global optimal solution that balances collision risk, path length, and trajectory smoothness in a static cylindrical obstacle environment. This synergistic fusion mechanism effectively resolves the issues of global optimization imbalance and local decision-making myopia. Experimental results demonstrate that in the CEC2017 benchmark tests, based on the Friedman aggregate ranking and Wilcoxon rank-sum test statistics, the MIDBO algorithm exhibits highly competitive solving accuracy and stability. Although it may not achieve significant superiority on a few specific multimodal functions, its overall statistical ranking secures the first place. In 3D UAV flight missions Ⅰ-Ⅳ within a static obstacle environment, MIDBO demonstrates significant advantages over 11 state-of-the-art algorithms, including PSO, ACO, GWO, GA, WOA, GOA, DBO, AQDBO, EDBO, LODBO, and LTDBO. Compared to the comprehensive average of these comparative algorithms, MIDBO reduces the mean fitness by 4.62%, 5.03%, 10.32%, and 12.91% across the four tasks, respectively. Compared with the aforementioned 11 algorithms, MIDBO achieves substantial enhancements in both cross-modal optimization capability and safe trajectory generation performance in complex static environments.</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

UAV path planning optimization algorithm based on an improved dung beetle optimization algorithm with a mixed strategy

  • 期刊论文

  • 发表日期：

    2026年7月29日

查看全文
  • Jian Deng,
  • Honghai Zhang,
  • Zeyu Liu,
  • Zihan Yu

摘要

To effectively address the multi-objective optimization challenges in unmanned aerial vehicle (UAV) path planning under multiple constraints—including path length and collision risk—and to overcome the insufficient adaptability of the traditional Dung Beetle Optimizer (DBO) in this domain, a Mix-Strategy Improved Dung Beetle Optimizer (MIDBO) is proposed. Specifically, the Circle-SPM chaotic map is introduced to optimize the population initialization process, effectively mitigating the premature convergence caused by uneven distribution and a lack of population diversity. A fusion mechanism combining adaptive weights and spiral motion is designed to enhance the global search capability, thereby preventing non-convergence issues stemming from an insufficient search range. Furthermore, a boundary adjustment strategy based on fractional calculus is proposed to provide an effective local-optimum escape mechanism during the foraging phase. Additionally, a piecewise model constraint mechanism is constructed to balance the resource allocation between local exploitation and global exploration from the perspective of optimization stage division. The optimization performance of MIDBO is initially verified using the CEC2017 benchmark functions to ensure a fundamental balance between local and global search. Subsequently, a multi-constraint model is established to further optimize this adaptability, efficiently exploring the global optimal solution that balances collision risk, path length, and trajectory smoothness in a static cylindrical obstacle environment. This synergistic fusion mechanism effectively resolves the issues of global optimization imbalance and local decision-making myopia. Experimental results demonstrate that in the CEC2017 benchmark tests, based on the Friedman aggregate ranking and Wilcoxon rank-sum test statistics, the MIDBO algorithm exhibits highly competitive solving accuracy and stability. Although it may not achieve significant superiority on a few specific multimodal functions, its overall statistical ranking secures the first place. In 3D UAV flight missions Ⅰ-Ⅳ within a static obstacle environment, MIDBO demonstrates significant advantages over 11 state-of-the-art algorithms, including PSO, ACO, GWO, GA, WOA, GOA, DBO, AQDBO, EDBO, LODBO, and LTDBO. Compared to the comprehensive average of these comparative algorithms, MIDBO reduces the mean fitness by 4.62%, 5.03%, 10.32%, and 12.91% across the four tasks, respectively. Compared with the aforementioned 11 algorithms, MIDBO achieves substantial enhancements in both cross-modal optimization capability and safe trajectory generation performance in complex static environments.