Scholarlink

<p>This paper proposed a two-stage optimization design method for metro underframe structures based on the gray wolf algorithm. The study adopted a joint topology-size optimization framework to achieve the dual goals of structural lightweighting and performance enhancement. First, multi-stiffness topology optimization of the underframe cross-section was carried out based on Optistruct software, by reconstructing the distribution of tendons and plates in order to optimize the force transfer path. Second, the particle swarm optimization (PSO) algorithm and BP neural network were combined to construct a high-precision agent model, and the improved multi-objective grey wolf algorithm (AWMOGWO) was introduced as the optimization master program for performance calibration. Finally, the performance of the algorithm was balanced using normalization and perturbation factors based on the optimality seeking principle of the gray wolf algorithm. The improved gray wolf algorithm significantly improves the global search capability and convergence ability under extreme working conditions through the adaptive weight adjustment mechanism and non-dominated sorting strategy. The simulation results show that the vertical displacement performance of the optimized underframe model was improved by 17∼21 % and about 9 % in the static strength-stress performance index, while the mass was reduced by 2∼2.5 %.</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

Adaptive gray wolf optimization algorithm for the optimization of metro underframe structures

  • 期刊论文

  • 发表日期：

    2026年4月9日

查看全文
  • Delei Du,
  • Jian Song,
  • Yana Li,
  • Xingyan Li,
  • Fangyu Liu,
  • Jianxin Xu

摘要

This paper proposed a two-stage optimization design method for metro underframe structures based on the gray wolf algorithm. The study adopted a joint topology-size optimization framework to achieve the dual goals of structural lightweighting and performance enhancement. First, multi-stiffness topology optimization of the underframe cross-section was carried out based on Optistruct software, by reconstructing the distribution of tendons and plates in order to optimize the force transfer path. Second, the particle swarm optimization (PSO) algorithm and BP neural network were combined to construct a high-precision agent model, and the improved multi-objective grey wolf algorithm (AWMOGWO) was introduced as the optimization master program for performance calibration. Finally, the performance of the algorithm was balanced using normalization and perturbation factors based on the optimality seeking principle of the gray wolf algorithm. The improved gray wolf algorithm significantly improves the global search capability and convergence ability under extreme working conditions through the adaptive weight adjustment mechanism and non-dominated sorting strategy. The simulation results show that the vertical displacement performance of the optimized underframe model was improved by 17∼21 % and about 9 % in the static strength-stress performance index, while the mass was reduced by 2∼2.5 %.