Scholarlink

<p>Investigating ductility and fracture load is crucial for enhancing the performance of joined materials. Different machine learning models, namely XGBoost, Decision Tree, Random Forest, Linear Regression, and support vector machine, have been utilized to investigate the friction stir spot welding (FSSW) process performances of AA2018-H2 and C10200 bi-metallic joints. Depending on the design of the experiment outlines, the experimental process was executed considering three key parameters: rotational speed (1500–2100&#xa0;rpm), plunge depth (0.1–0.3&#xa0;mm), and dwelling time (2–6&#xa0;S). The regression equations created the correlations between FSSW controlling parameters with the fracture load and elongation. The significant results have been observed in the XGBoost model, with the coefficient of determination (<i>R</i><sup>2</sup>) for fracture load at 0.9762, elongation at 0.2965, and mean squared error for fracture load at 7750.23 (N<sup>2</sup>), while elongation is at 0.0227 (sq.%) following the optimization. The FE-SEM fractographs simultaneously revealed dimple structures, which confirmed the occurrence of ductile fracture in the nugget area of the welded zone, across all specimens. The maximum fracture load and the corresponding elongation have been obtained as 4125&#xa0;N and 1.2%, respectively, in the set of 1500&#xa0;rpm, 0.2&#xa0;mm, 6&#xa0;s. This attempt highlights the importance of a methodological approach in addressing the direct control of the FSSW industrial challenges. </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

Machine Learning-Based Data Prediction in Friction Stir Spot Welding for AA2018-H2 and C10200 on Fracture Load and Ductility

  • 期刊论文

  • 发表日期：

    2025年8月4日

查看全文
  • Suprim Sardar,
  • Sutanu Misra,
  • Partha Pratim Dey

摘要

Investigating ductility and fracture load is crucial for enhancing the performance of joined materials. Different machine learning models, namely XGBoost, Decision Tree, Random Forest, Linear Regression, and support vector machine, have been utilized to investigate the friction stir spot welding (FSSW) process performances of AA2018-H2 and C10200 bi-metallic joints. Depending on the design of the experiment outlines, the experimental process was executed considering three key parameters: rotational speed (1500–2100 rpm), plunge depth (0.1–0.3 mm), and dwelling time (2–6 S). The regression equations created the correlations between FSSW controlling parameters with the fracture load and elongation. The significant results have been observed in the XGBoost model, with the coefficient of determination (R2) for fracture load at 0.9762, elongation at 0.2965, and mean squared error for fracture load at 7750.23 (N2), while elongation is at 0.0227 (sq.%) following the optimization. The FE-SEM fractographs simultaneously revealed dimple structures, which confirmed the occurrence of ductile fracture in the nugget area of the welded zone, across all specimens. The maximum fracture load and the corresponding elongation have been obtained as 4125 N and 1.2%, respectively, in the set of 1500 rpm, 0.2 mm, 6 s. This attempt highlights the importance of a methodological approach in addressing the direct control of the FSSW industrial challenges.