Scholarlink

<p>This study aims to develop an integrated experimental and machine learning framework for accurately predicting the tensile behaviour of high-content multi-walled carbon nanotubes (MWCNTs)-reinforced polylactic acid (PLA) composites fabricated using fused filament fabrication (FFF). The composite filaments were extruded using a single-screw process and 3D-printed under varied raster orientation, infill density, and layer height. Mechanical characterization through ASTM-standard tensile testing revealed a maximum tensile strength of 54.269&#xa0;MPa, highlighting the critical influence of process parameters on interlayer bonding. X-ray diffraction analysis confirmed a crystallinity index increase from 36% (neat PLA) to 47.41% for PLA–MWCNTs composites, while SEM imaging demonstrated uniform MWCNTs dispersion without agglomeration. To predict tensile strength and reduce experimental iteration, five machine learning models—Linear Regression, Random Forest, AdaBoost, Ridge Regression, and K-Nearest Neighbors—were trained and evaluated. Among them, LR and RR achieved the highest accuracy with R<sup>2</sup> = 0.99, outperforming ensemble and instance-based models. The proposed framework can be applied to optimize process parameters, reduce experimental effort, and support the design of high-performance polymer nanocomposites for applications in aerospace, automotive, and biomedical engineering.</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

Integrating experimental characterization and machine learning for predicting the tensile behaviour of PLA–MWCNTs FFF composites

  • 期刊论文

  • 发表日期：

    2026年6月19日

查看全文
  • Tapish Raj,
  • Akash Jain,
  • Ankit Sahai,
  • Rahul Swarup Sharma

摘要

This study aims to develop an integrated experimental and machine learning framework for accurately predicting the tensile behaviour of high-content multi-walled carbon nanotubes (MWCNTs)-reinforced polylactic acid (PLA) composites fabricated using fused filament fabrication (FFF). The composite filaments were extruded using a single-screw process and 3D-printed under varied raster orientation, infill density, and layer height. Mechanical characterization through ASTM-standard tensile testing revealed a maximum tensile strength of 54.269 MPa, highlighting the critical influence of process parameters on interlayer bonding. X-ray diffraction analysis confirmed a crystallinity index increase from 36% (neat PLA) to 47.41% for PLA–MWCNTs composites, while SEM imaging demonstrated uniform MWCNTs dispersion without agglomeration. To predict tensile strength and reduce experimental iteration, five machine learning models—Linear Regression, Random Forest, AdaBoost, Ridge Regression, and K-Nearest Neighbors—were trained and evaluated. Among them, LR and RR achieved the highest accuracy with R2 = 0.99, outperforming ensemble and instance-based models. The proposed framework can be applied to optimize process parameters, reduce experimental effort, and support the design of high-performance polymer nanocomposites for applications in aerospace, automotive, and biomedical engineering.