Scholarlink

<p>This study investigates drilling forces in multi-layered AA6061 aluminum sheets fabricated via accumulative roll bonding (ARB) under diverse machining conditions, combining experimental characterization with machine learning prediction. Drilling tests were performed on 3–12 layer configurations using three point angles (118°, 130°, 140°), three feed rates (0.08, 0.12, 0.16&#xa0;mm/rev), and three spindle speeds (500, 1000, 1500&#xa0;rpm). Analysis of variance (ANOVA) revealed feed rate (<i>p</i> &lt; 0.0001) as the most significant factor, followed by tip angle, number of layers, and spindle speed, with an <i>R</i><sup>2</sup> of 0.9890 for the fitted model. Main effects analysis showed a reduction in thrust force by tip angle increase and rise in thrust force as feed rate increased. The random forest (RF) model, trained using a 70/30 split, attained high predictive accuracy (<i>R</i><sup>2</sup> &gt; 0.98 in training and &gt; 0.92 in testing) with mean absolute percentage errors below 8%. Feature importance analysis ranked feed rate highest (≈ 42%), followed by point angle, number of layers, and spindle speed. Numerical optimization indicated minimum thrust force (252.6167 N) at 140°, 0.08&#xa0;mm/rev, 1500&#xa0;rpm, and 12 layers, with a desirability of 1.000. The outcomes demonstrate the effectiveness of integrating statistical design of experiments with RF-based predictive modeling to optimize drilling of ARB-processed aluminum alloys, offering both improved understanding of parameter–force interactions and practical guidelines for low-force machining strategies.</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

Experimental Characterization and Random Forest-Based Prediction of Drilling Forces in Multi-Layered AA6061 Sheets Processed by Accumulative Roll Bonding

  • 期刊论文

  • 发表日期：

    2026年2月5日

查看全文
  • Adel Ziae Azar,
  • Amir Mostafapour,
  • Mohammad Baraheni

摘要

This study investigates drilling forces in multi-layered AA6061 aluminum sheets fabricated via accumulative roll bonding (ARB) under diverse machining conditions, combining experimental characterization with machine learning prediction. Drilling tests were performed on 3–12 layer configurations using three point angles (118°, 130°, 140°), three feed rates (0.08, 0.12, 0.16 mm/rev), and three spindle speeds (500, 1000, 1500 rpm). Analysis of variance (ANOVA) revealed feed rate (p < 0.0001) as the most significant factor, followed by tip angle, number of layers, and spindle speed, with an R2 of 0.9890 for the fitted model. Main effects analysis showed a reduction in thrust force by tip angle increase and rise in thrust force as feed rate increased. The random forest (RF) model, trained using a 70/30 split, attained high predictive accuracy (R2 > 0.98 in training and > 0.92 in testing) with mean absolute percentage errors below 8%. Feature importance analysis ranked feed rate highest (≈ 42%), followed by point angle, number of layers, and spindle speed. Numerical optimization indicated minimum thrust force (252.6167 N) at 140°, 0.08 mm/rev, 1500 rpm, and 12 layers, with a desirability of 1.000. The outcomes demonstrate the effectiveness of integrating statistical design of experiments with RF-based predictive modeling to optimize drilling of ARB-processed aluminum alloys, offering both improved understanding of parameter–force interactions and practical guidelines for low-force machining strategies.