Scholarlink

<p>Multi-stage tube extrusion is a widely used manufacturing process for fabricating precision tubular components. In the current study, St37 tubes were extruded using multi-stage forming dies, consisting of two preforming dies and one final forming die. The die’s final length, friction coefficient, and velocity were selected as input parameters, while thickness distribution, extrusion force, and mean square error (MSE) were considered as outputs. Finite Element Modeling was carried out in ABAQUS/Explicit and integrated with optimization algorithms to improve process performance. To identify the optimal combination of processing parameters, a hybrid machine learning framework combining Non-Dominated Sorting Genetic Algorithm II (NSGA-II) and Artificial Neural Networks (ANN) was employed. The model demonstrated high prediction accuracy with strong agreement between simulated and experimental results. Validation through thickness and force measurements confirmed the model’s robustness. The optimal parameters achieved uniform thickness distribution, reduced extrusion force, and minimized MSE. Confirmatory experiments verified the developed model, showing low errors of approximately 3.19% in force and 4.36% in MSE, thereby confirming the reliability and efficiency of the proposed optimization approach.</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

Integrated data-driven and simulation-based approach to optimize St37 tube extrusion

  • 期刊论文

  • 发表日期：

    2026年1月17日

查看全文
  • Mehran Ghasempour-Mouziraji,
  • Morteza Hosseinzadeh,
  • Majid Mohammadhosseinzadeh,
  • Hossein Hajimiri,
  • Ehsan Marzban shirkharkolaei,
  • Mojtaba Najafizadeh,
  • Ricardo Alves De Sousa

摘要

Multi-stage tube extrusion is a widely used manufacturing process for fabricating precision tubular components. In the current study, St37 tubes were extruded using multi-stage forming dies, consisting of two preforming dies and one final forming die. The die’s final length, friction coefficient, and velocity were selected as input parameters, while thickness distribution, extrusion force, and mean square error (MSE) were considered as outputs. Finite Element Modeling was carried out in ABAQUS/Explicit and integrated with optimization algorithms to improve process performance. To identify the optimal combination of processing parameters, a hybrid machine learning framework combining Non-Dominated Sorting Genetic Algorithm II (NSGA-II) and Artificial Neural Networks (ANN) was employed. The model demonstrated high prediction accuracy with strong agreement between simulated and experimental results. Validation through thickness and force measurements confirmed the model’s robustness. The optimal parameters achieved uniform thickness distribution, reduced extrusion force, and minimized MSE. Confirmatory experiments verified the developed model, showing low errors of approximately 3.19% in force and 4.36% in MSE, thereby confirming the reliability and efficiency of the proposed optimization approach.