Scholarlink

<p>Conventional experimental techniques for evaluating friction and wear properties of aluminum-based nanocomposites are both time-intensive and costly, restricting their widespread adoption in aerospace applications where precise tribological evaluation is essential for improving fuel efficiency and reducing weight. This study aims to develop a machine learning framework for accurately predicting the Coefficient of Friction (CoF) and wear rate of Al7075/B<sub>4</sub>C metal matrix composites. A series of tribological experiments were conducted using a pin-on-disc setup with Al7075 reinforced with varying B<sub>4</sub>C weight percentages (0%, 4%, 8%, and 12%), generating a dataset of 10,800 records. A novel Wear and Friction Ensemble Machine Learning (WFEML) framework was developed, employing sequentially stacked phases that progressively refine prediction accuracy. Applied load, reinforcement percentage, time, sliding velocity, and sliding distance were used as input variables, while the outputs were CoF and wear rate. The 12% B<sub>4</sub>C-reinforced composite demonstrated optimal wear resistance under severe conditions, exhibiting a consistently reduced CoF across all compositions and achieving a 77% reduction in wear rate under severe operating conditions. The WFEML framework achieved <i>R</i><sup><i>2</i></sup> values of 0.98 for CoF and 0.99 for wear rate, confirming high predictive reliability. A tailored multi-phase ensemble machine learning framework is proposed, integrating experimental tribological to support data-driven materials design in aluminum-based composites.</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

Tribo-Informatics Method for Predictive Modeling of Friction and Wear Mechanisms in Al7075/B4C Metal Matrix Composites

  • 期刊论文

  • 发表日期：

    2026年3月30日

查看全文
  • Ranjeet Kumbhar,
  • Appaso M. Gadade,
  • Rajmeet Singh,
  • Mudit Choudhary

摘要

Conventional experimental techniques for evaluating friction and wear properties of aluminum-based nanocomposites are both time-intensive and costly, restricting their widespread adoption in aerospace applications where precise tribological evaluation is essential for improving fuel efficiency and reducing weight. This study aims to develop a machine learning framework for accurately predicting the Coefficient of Friction (CoF) and wear rate of Al7075/B4C metal matrix composites. A series of tribological experiments were conducted using a pin-on-disc setup with Al7075 reinforced with varying B4C weight percentages (0%, 4%, 8%, and 12%), generating a dataset of 10,800 records. A novel Wear and Friction Ensemble Machine Learning (WFEML) framework was developed, employing sequentially stacked phases that progressively refine prediction accuracy. Applied load, reinforcement percentage, time, sliding velocity, and sliding distance were used as input variables, while the outputs were CoF and wear rate. The 12% B4C-reinforced composite demonstrated optimal wear resistance under severe conditions, exhibiting a consistently reduced CoF across all compositions and achieving a 77% reduction in wear rate under severe operating conditions. The WFEML framework achieved R2 values of 0.98 for CoF and 0.99 for wear rate, confirming high predictive reliability. A tailored multi-phase ensemble machine learning framework is proposed, integrating experimental tribological to support data-driven materials design in aluminum-based composites.