Scholarlink

<p>Welding process optimization is critical for reducing structural deformation and residual stress. However, when the welding procedure is fixed, optimizing the welding sequence becomes crucial for further improving performance, especially for large frame structures where the nonlinear relationships among the welding sequence, structural deformation, and residual stress pose significant challenges. Existing methods, which rely heavily on high-cost finite element simulations and time-intensive manual adjustments during the design and manufacturing process analysis stage, are inefficient and fail to meet production requirements. By integrating the two, the efficiency can be increased, and the deformation and residual stress can be effectively reduced. To address these challenges, a welding sequence optimization framework that combines artificial neural networks with an improved whale optimization algorithm is proposed in this paper. On the basis of thermoelastic‒plastic finite element simulations, input datasets were generated with welding sequences as inputs and structural deformation and residual stress as outputs. The framework predicts optimal welding sequences and corresponding deformation and stress, providing a practical solution for complex structural components. In this study, we propose a chaotic improved whale optimization algorithm with backpropagation (CIWOA-BP) model. Compared with the traditional backpropagation (BP) model, the CIWOA-BP model achieves higher prediction accuracy and greater stability. The experimental results indicate that CIWOA-BP can effectively optimize the welding sequence of complex structural components, improving the welding quality. Specifically, the structural deformation is reduced by 50%, and the residual stress is decreased by 80%, increasing assembly accuracy.</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

Welding sequence optimization based on BP and chaotic whale optimization

  • 期刊论文

  • 发表日期：

    2025年12月5日

查看全文
  • Han Wang,
  • Hui Zhang,
  • Wei Lin,
  • Acun Pan,
  • Chengshun Zhu,
  • Liyan Zhang

摘要

Welding process optimization is critical for reducing structural deformation and residual stress. However, when the welding procedure is fixed, optimizing the welding sequence becomes crucial for further improving performance, especially for large frame structures where the nonlinear relationships among the welding sequence, structural deformation, and residual stress pose significant challenges. Existing methods, which rely heavily on high-cost finite element simulations and time-intensive manual adjustments during the design and manufacturing process analysis stage, are inefficient and fail to meet production requirements. By integrating the two, the efficiency can be increased, and the deformation and residual stress can be effectively reduced. To address these challenges, a welding sequence optimization framework that combines artificial neural networks with an improved whale optimization algorithm is proposed in this paper. On the basis of thermoelastic‒plastic finite element simulations, input datasets were generated with welding sequences as inputs and structural deformation and residual stress as outputs. The framework predicts optimal welding sequences and corresponding deformation and stress, providing a practical solution for complex structural components. In this study, we propose a chaotic improved whale optimization algorithm with backpropagation (CIWOA-BP) model. Compared with the traditional backpropagation (BP) model, the CIWOA-BP model achieves higher prediction accuracy and greater stability. The experimental results indicate that CIWOA-BP can effectively optimize the welding sequence of complex structural components, improving the welding quality. Specifically, the structural deformation is reduced by 50%, and the residual stress is decreased by 80%, increasing assembly accuracy.